{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dawn Of\n\nSCREENPLAY EXCERPT:\nRoger:\tHIS NAME'S PETER. HE'S ALRIGHT.\n\n\tThe three are already moving toward the helicopter.\n\n\tRoger:\tLET'S HUSTLE.\n\n102\tPeter has stowed the supplies in the rear of the cockpit, and he\n\thas noticed the fuel hose lying on the dock. He tries the nozzle\n\tin the receptacle on the chopper and holds it in until the tank\n\tfills.\n\n103\tThe other \"Policemen\" are still moving cartons of supplies from\n\ttheir van down the dock.\n\n\tRoger:\t(to the other Policemen)\n\t\t\tYOU GUYS BETTER MOVE IT. THERE'S A RADIO\n\t\t\tREPORT ABOUT THE DOCK BEIN' KNOCKED OUT.\n\n104\tThey reach the cockpit. Fran climbs in and crouches on the floor\n\tin the rear of the bubble.\n\n\tFran:\tYOU SURE THIS'LL CARRY US ALL.\n\n\tSteve:\tLITTLE HARDER ON THE FUEL, BUT WE'LL BE OK.\n\n105\tAs Peter climbs aboard, one of the other policemen, carrying a\n\tfinal carton, speaks to Roger.\n\n\tOfficer 2: HEY...YOU GOT ANY CIGARETTES.\n\n\tRoger looks at the others one at a time. Fran shakes her head\n\t\"no\".\n\n\tRoger:\tSORRY. (he trots around to the passenger seat)\n\n\tSteve:\tWHERE YA HEADED?\n\n\tOfficer 2: DOWN RIVER...GOT AN IDEA MAYBE WE CAN MAKE IT TO\n\t\t\tTHE ISLANDS.\n\n\tSteve:\tWHAT ISLANDS? (he starts the engine)\n\n\tOfficer 2: ANY ISLANDS...WHAT ABOUT YOU? WHERE YOU HEADED?\n\n\tSteve:\tSTRAIGHT UP.\n\n106\tThe Policeman rushes off with his two cohorts. As they untie one\n\tof the launches from the dock, the J.A.S. helicopter whines\n\tloudly. Then it lifts off the dock with a smooth motion.\n\n\tThe Police launch starts without a problem, and it pulls out\n\tonto the dark river.\n\n107\tThe lights on the helicopter blink as the metal bird swoops low\n\tover the Philad"}, {"role": "assistant", "content": "## SCENE 1 — INT. TO THE OTHER POLICEMEN\nACTION: Roger:\tHIS NAME'S PETER. HE'S ALRIGHT. The three are already moving toward the helicopter.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dawn Of: Roger:\tHIS NAME'S PETER. HE'S ALRIGHT. The three are already moving toward the helicopter., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Roger:\tHIS NAME'S PETER. HE'S ALRIGHT. The three are already moving toward the helicopter. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: 83 83\nSetting: derived from Crank\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. DOWN THE BLOCK - DAY\nACTION: 83 83 A POLICE MOTORCYCLE is tipped over on a lawn, coughing up\nDIALOGUE — Ve: You're embarrassing.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Crank: 83 83 A POLICE MOTORCYCLE is tipped over on a lawn, coughing up, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. 83 83 A POLICE MOTORCYCLE is tipped over on a lawn, coughing up The camera pushes in with small amplitude at slow speed. Ve (S1) says: [English] You're embarrassing. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Hellboy\"\n\nSCREENPLAY EXCERPT:\nINT. ABANDONED SHOWER ROOM (SET) - DAY\n\n Hellboy enters to find Abe, bleeding but alive, leaning\n against the tile, blue blood all around him. Hellboy tries\n his walkie-talkie. Turns on his locator. It sparks. It's\n damaged and broken.\n\n HELLBOY\n (into his walkie-talkie)\n We need an ambulance. Now!! Over!!"}, {"role": "assistant", "content": "## SCENE 1 — INT. ABANDONED SHOWER ROOM (SET) - DAY\nACTION: Hellboy enters to find Abe, bleeding but alive, leaning against the tile, blue blood all around him. Hellboy tries\nDIALOGUE — Hellboy: We need an ambulance. Now!! Over!!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hellboy\": Hellboy enters to find Abe, bleeding but alive, leaning against the tile, blue blood all around him. Hellboy tries, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hellboy enters to find Abe, bleeding but alive, leaning against the tile, blue blood all around him. Hellboy tries The camera pushes in with small amplitude at slow speed. Hellboy (S1) says: [English] We need an ambulance. Now!! Over!! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I'm going to the ladies room.\nSetting: derived from The Sponge\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. THEY LEAVE.\nACTION: I'm going to the ladies room. I'll go with you. (They leave.)\nDIALOGUE — Lena: I'll go with you. (They leave.)\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Sponge: I'm going to the ladies room. I'll go with you. (They leave.), detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm going to the ladies room. I'll go with you. (They leave.) The camera pushes in with small amplitude at slow speed. Lena (S1) says: [English] I'll go with you. (They leave.) The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Screenplay by Sam Raimi and Ivan Raimi\nSetting: derived from The Army Of Darkness\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: Screenplay by Sam Raimi and Ivan Raimi ------------------------------------------------------------------------------\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Army Of Darkness: Screenplay by Sam Raimi and Ivan Raimi ------------------------------------------------------------------------------, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Screenplay by Sam Raimi and Ivan Raimi ------------------------------------------------------------------------------ The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Hellboy\"\n\nSCREENPLAY EXCERPT:\nEXT. NEARBY - BROOM - NIGHT\n\n He lowers a pair of binoculars and passes them to Whitman.\n\n BROOM\n If he's here, this is worse than I\n thought.\n\n WHITMAN\n (to the radio man)\n Air and sea backup. What's closest?\n\n The RADIOMAN cranks a transmitter to life.\n\n RADIO MAN\n Londonderry, sir. Forty minutes away.\n\n BROOM\n We don't have forty minutes."}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEARBY - BROOM - NIGHT\nACTION: He lowers a pair of binoculars and passes them to Whitman. If he's here, this is worse than I\nDIALOGUE — Broom: If he's here, this is worse than I thought.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hellboy\": He lowers a pair of binoculars and passes them to Whitman. If he's here, this is worse than I, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He lowers a pair of binoculars and passes them to Whitman. If he's here, this is worse than I The camera pushes in with small amplitude at slow speed. Broom (S1) says: [English] If he's here, this is worse than I thought. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BlueHeelers illustrating the trope 'SeriesFauxnale'. A strange example in the episode where Tom marries Grace, which would seem like a fitting end to the show, and instead of previews for next week John Wood announces that Blue Heelers will not be seen. What looks to be a final goodbye turns out to be regular programming being pushed aside because it was one year since September 11.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying SeriesFauxnale.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BlueHeelers, SeriesFauxnale, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for SeriesFauxnale. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Dan and Laurie are back in costume.\nSetting: derived from Watchmen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. OWL-SHIP (MOVING) - NIGHT\nACTION: Dan and Laurie are back in costume. Look, Dan, I'm glad you were able to\nDIALOGUE — Laurie: Look, Dan, I'm glad you were able to rediscover your calling in life, but I think you're getting carried away. You're talking about springing a homicidal...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Watchmen: Dan and Laurie are back in costume. Look, Dan, I'm glad you were able to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Dan and Laurie are back in costume. Look, Dan, I'm glad you were able to The camera pushes in with small amplitude at slow speed. Laurie (S1) says: [English] Look, Dan, I'm glad you were able to rediscover your calling in life, but I think you're getting carried away. You're talking about springing a homicidal... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: There is a long silence.\nSetting: derived from Dawn Of\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TO THE BOY\nACTION: There is a long silence. Roger:\tCOME ON, MARTINEZ!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dawn Of: There is a long silence. Roger:\tCOME ON, MARTINEZ!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. There is a long silence. Roger:\tCOME ON, MARTINEZ! The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Oh! Thank you. Thank you very much.\nSetting: derived from The Revenge\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BLOWING NOSE\nACTION: Oh! Thank you. Thank you very much. (blowing nose) Really. I mean\nDIALOGUE — Elaine: Oh! Thank you. Thank you very much. (blowing nose) Really. I mean\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Revenge: Oh! Thank you. Thank you very much. (blowing nose) Really. I mean, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Oh! Thank you. Thank you very much. (blowing nose) Really. I mean The camera pushes in with small amplitude at slow speed. Elaine (S1) says: [English] Oh! Thank you. Thank you very much. (blowing nose) Really. I mean The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Asteroid City\n\nSCREENPLAY EXCERPT:\nEXT. MOTEL CABIN #7. EVENING\n\nBetween the burned ruin and the canvas tent: the five teenagers\nsit together in a circle on the ground (under a pink oleander,\nnear a humming air conditioning unit). Dinah rapidly explains\nthe rules of a game:\n\n DINAH\n After that, the second person says the\n name the first one said and adds another;\n then the third person says both plus a\n new name; then the next person keeps\n going and so on in a circle. It's a\n memory game. Get it?\n\n RICKY\n I think so.\n\n DINAH\n I'll start. Cleopatra.\n\n RICKY\n (pause)\n Cleopatra, Jagadish Chandra Bose. Like\n that?\n\nDinah nods. The five teenagers adjust and re-situate themselves,\nphysically and mentally. Shelly picks up where Ricky left off:\n\n SHELLY\n Cleopatra, Jagadish Chandra Bose, Antonie\n van Leeuwenhoek.\n\n CLIFFORD\n (eagerly)\n Cleopatra, Jagadish Chandra Bose, Antonie\n van Leeuwenhoek -- Paracelsus.\n\nClifford raises an eyebrow. Woodrow points to each player as he\nrecalls his/her name:\n\n WOODROW\n Cleopatra, Jagadish Chandra Bose, Antonie\n van Leeuwenhoek, Paracelsus --\n (pointing to himself)\n Kurt G�del.\n\nA murmur of appreciation. Camera now pans from one player to the\nnext as Dinah lists:\n 37.\n\n\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. MOTEL CABIN #7. EVENING\nACTION: Between the burned ruin and the canvas tent: the five teenagers sit together in a circle on the ground (under a pink oleander,\nDIALOGUE — Dinah: After that, the second person says the name the first one said and adds another; then the third person says both plus a new name; then the next person keeps...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Asteroid City: Between the burned ruin and the canvas tent: the five teenagers sit together in a circle on the ground (under a pink oleander,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Between the burned ruin and the canvas tent: the five teenagers sit together in a circle on the ground (under a pink oleander, The camera pushes in with small amplitude at slow speed. Dinah (S1) says: [English] After that, the second person says the name the first one said and adds another; then the third person says both plus a new name; then the next person keeps... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Neverending Story\n\nSCREENPLAY EXCERPT:\nEXT. VIEW-\n\nthe courtyard is being taken away by the Nothing. Because of Bastians new\n\nfound disbelief the Nothing has grown stronger and is now attacking the last remaining part of Fantasia.\n\n\n\n Childlike Empress:\n\n Bastian, why don't you do what you dream, Bastian ?\n\n\n\n Bastian:\n\n style=\"mso-tab-count: 1\"> style=\"mso-spacerun: yes\"> But I can't ! I have to keep my feet on the ground!\n\n\n\n Childlike Empress:\n\n Call my name ! Bastian, please ! Save us !\n\n\n\n\n\n Bastian:\n\n All right, I'll do it. I'll save you. I will do what I dream!\n\n\n\nHe climbs up to the window and opens it. He leans out into the storm and calls out the name he had chosen for her.\n\n\n\n Bastian:\n\n MOONCHILD!\n\n\n\nNEXT SCENE-\n\nDarkness, pure and black as night. We hear Bastian speak.\n\n\n\n Bastian:\n\n style=\"mso-tab-count: 2\"> style=\"mso-tab-count: 1\"> style=\"mso-spacerun: yes\"> Why is it so dark?\n\n\n\n Empress Moonchild:\n\n In the beginning it is always dark.\n\n\n\n\n\nA small light appears and starts growing until the two childs faces are\n\n illuminated. It is coming from an object in Moonchild's hand. Bastian looks at it.\n\n\n\n Bastian:\n\n style=\"mso-tab-count: 2\"> style=\"mso-spacerun: yes\"> What is th"}, {"role": "assistant", "content": "## SCENE 1 — EXT. VIEW-\nACTION: the courtyard is being taken away by the Nothing. Because of Bastians new found disbelief the Nothing has grown stronger and is now attacking the last remaining part of Fantasia.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Neverending Story: the courtyard is being taken away by the Nothing. Because of Bastians new found disbelief the Nothing has grown stronger and is now attacking the last remaining part of Fantasia., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. the courtyard is being taken away by the Nothing. Because of Bastians new found disbelief the Nothing has grown stronger and is now attacking the last remaining part of Fantasia. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Blade Runner\n\nSCREENPLAY EXCERPT:\nEXT. PLATEAU - DUSK 85\n\n And beyond on a plateau overlooking the grounds, a\n figure stands watching, waiting like a bird of prey."}, {"role": "assistant", "content": "## SCENE 1 — EXT. PLATEAU - DUSK 85\nACTION: And beyond on a plateau overlooking the grounds, a figure stands watching, waiting like a bird of prey.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade Runner: And beyond on a plateau overlooking the grounds, a figure stands watching, waiting like a bird of prey., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. And beyond on a plateau overlooking the grounds, a figure stands watching, waiting like a bird of prey. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Lord Of The Rings: The Return Of The King\n\nSCREENPLAY EXCERPT:\nEXT. DUNHARROW ENCAMPMENT - DAWN\n\nMUCH ACTIVITY ... HORNS SOUND ... the ENTIRE CAMP is on the\nMOVE ... fires are DOUSED, TENTS pulled down, HORSES saddled."}, {"role": "assistant", "content": "## SCENE 1 — EXT. DUNHARROW ENCAMPMENT - DAWN\nACTION: MUCH ACTIVITY ... HORNS SOUND ... the ENTIRE CAMP is on the MOVE ... fires are DOUSED, TENTS pulled down, HORSES saddled.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Return Of The King: MUCH ACTIVITY ... HORNS SOUND ... the ENTIRE CAMP is on the MOVE ... fires are DOUSED, TENTS pulled down, HORSES saddled., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. MUCH ACTIVITY ... HORNS SOUND ... the ENTIRE CAMP is on the MOVE ... fires are DOUSED, TENTS pulled down, HORSES saddled. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: This futuristic bachelor pad is constructed entirely of brightly\nSetting: derived from The Heavy Metal Movie\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HARRY'S APARTMENT\nACTION: This futuristic bachelor pad is constructed entirely of brightly coloured molden plastic. Bowling trophies sit on a shelf\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Heavy Metal Movie: This futuristic bachelor pad is constructed entirely of brightly coloured molden plastic. Bowling trophies sit on a shelf, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. This futuristic bachelor pad is constructed entirely of brightly coloured molden plastic. Bowling trophies sit on a shelf The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Die Hard\"\n\nSCREENPLAY EXCERPT:\nThey're shooting at the lights.\n\n More shots ring out from the building going over the SWAT\n officers' heads and suddenly the huge dome of one of the\n spotlights shatters behind Mitchell and Robinson's head.\n The glow fades. A moment later the next light twenty feet\n away dies.\n\n ROBINSON\n They're going after the lights!\n\n The two SWAT officers cutting the garage ate suddenly look up\n as their cover starts to disappear.\n\n ROBINSON\n Call them back.\n\n MITCHELL\n No, they're almost in.\n\n Suddenly the third and fourth lights are shot out and the SWAT\n men become sitting ducks.\n\n196 IN HOLLY'S OFFICE - HANS 196\n\n He calmly speak into his CB.\n\n HANS\n Don't get impatient. Just wound them.\n\n197 INT. LOBBY 197\n\n Eddie and Uli fire. They hit one of the officers in the leg,\n the second one in the arm.\n\n198 EXT. POLICE BARRICADES - ON MITCHELL AND ROBINSON 198\n\n MITCHELL"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: They're shooting at the lights. More shots ring out from the building going over the SWAT\nDIALOGUE — Robinson: They're going after the lights!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Die Hard\": They're shooting at the lights. More shots ring out from the building going over the SWAT, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They're shooting at the lights. More shots ring out from the building going over the SWAT The camera pushes in with small amplitude at slow speed. Robinson (S1) says: [English] They're going after the lights! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BitchinKitchen illustrating the trope 'MadlibsCatchphrase'. Nadia G has a recurring rant on Bitchin' Kitchen that goes something like this:\nNadia: What more could you ask for, eh? I know: A [adjective] jetpack. I too once longed for a [adjective] jetpack, but then I thought about it. [two problems with said jetpack] and all for what? So you can [mundane use of jetpack]? It ain't worth it man! Think about it.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying MadlibsCatchphrase.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BitchinKitchen, MadlibsCatchphrase, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for MadlibsCatchphrase. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Just a way of putting me on\nSetting: derived from \"Total Recall\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Just a way of putting me on ice. He'd of reversed it\nDIALOGUE — Bennie: Man! This is real complicated. Go over it again, real slow.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Total Recall\": Just a way of putting me on ice. He'd of reversed it, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Just a way of putting me on ice. He'd of reversed it The camera pushes in with small amplitude at slow speed. Bennie (S1) says: [English] Man! This is real complicated. Go over it again, real slow. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: American Psycho\n\nSCREENPLAY EXCERPT:\nINT. SMITH AND WOLLENSKY RESTAURANT - DAY\n\nKIMBALL and Bateman are sitting at a corner table.\n\nKIMBALL\n(Very surprised)\nNo hash browns?\n\nBATEMAN\nNot in the mood, I guess.\n\nKIMBALL\nBut...everyone orders the hash browns here. I\nmean- it's-have you been here before?\n\nBATEMAN\n(Deliberately nonchalant)\nYes, of course. The hash browns are delicious. I'm just...not...\nordering them.\n\nKIMBALL\n(Looking at him like he's nuts)\nSuit yourself, I guess.\n\nPause.\n\nKIMBALL\nSo, the night he disappeared? Any new thoughts on\nwhat you did?\n\nBATEMAN\nI'm not really sure. I had a shower...and some\nsorbet?\n\nKIMBALL\nI think maybe you've got your dates mixed up.\n\nBATEMAN\nBut how? Where do you place Paul that night?\n\nKIMBALL\nAccording to his date book, and this was verified by his\nsecretary, he had dinner with...Marcus Halberstam.\n\nBATEMAN\nAnd?\n\nKIMBALL\nI've questioned him.\n\nBATEMAN\nMarcus?\n\nKIMBALL\nYes. And he denies it. Though at first he couldn't\nbe sure.\n\nBATEMAN\nBut Marcus denied it?\n\nKIMBALL\nYes.\n\nBATEMAN\nWell, does Marcus have an alibi?\n\nKIMBALL\nYes.\n\nA pause.\n\nBATEMAN\nHe does? You're sure?\n\nKIMBALL\n(smiling)\nI checked it out. It's clean.\n\nBATEMAN\nOh.\nKIMBALL Now where were you?\n(He laughs)\n\nBATEMAN\n(Laughing with him)\nWhere was Marcus?\n\nKIMBALL\nHe wasn't with Paul\nOwen.\n\nBATEMAN\nSo who was he with?\n\nKIMBALL\nHe was at Atlantis with Craig McDermott, Frederick Dibble,\nHarry Newman, George Butner and �\n(He pauses, Then looks up)\n- you.\n\nA moment of stunned silence.\n\nBATEMAN\nOh, right. Of course...We had wanted Paul Owen\nto come. But he said he had plans...I guess I had dinner\nwith Victoria...the f"}, {"role": "assistant", "content": "## SCENE 1 — INT. SMITH AND WOLLENSKY RESTAURANT - DAY\nACTION: KIMBALL and Bateman are sitting at a corner table. Not in the mood, I guess.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American Psycho: KIMBALL and Bateman are sitting at a corner table. Not in the mood, I guess., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. KIMBALL and Bateman are sitting at a corner table. Not in the mood, I guess. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nIt's my condition. I never know if I've\n\t\talready eaten, so I always just eat small\n\t\tamounts.\n\n\t\t\t\tTEDDY\n\t\tYou don't have to remember to be hungry.\n\n\t\t\t\tLEONARD\n\t\tIt's weird, but if you don't eat for a\n\t\twhile then your body stops being hungry.\n\t\tYou get sort of shaky but you don't\n\t\trealize you haven't eaten. Have I told\n\t\tyou about Sammy Jankis?\n\n\t\t\t\tTEDDY\n\t\tYeah, yeah. I heard enough about him.\n\t\tTell me about John G. You still think\n\t\the's here, right?\n\n\t\t\t\tLEONARD\n\t\tWho?\n\n\t\t\t\tTEDDY\n\t\tThe guy you're looking for, Johnny G.\n\t\tThat's why you haven't left. Am I right?\n\nLeonard shrugs. Teddy licks his fingers and frowns.\n\n\t\t\t\tTEDDY (cont'd)\n\t\tLeonard, you need to be very careful.\n\n\t\t\t\tLEONARD\n\t\tWhy?\n\n\t\t\t\tTEDDY\n\t\tWell, the other day you made it sound\n\t\tlike you thought somebody might be trying\n\t\tto set you up. Get you to kill the wrong\n\t\tguy.\n\n\t\t\t\tLEONARD\n\t\tYeah, well I go on facts, not\n\t\trecommendations, okay?\n\n\t\t\t\tTEDDY\n\t\tLenny, you can't trust a man's life to\n\t\tyour little notes and pictures.\n\n\t\t\t\tLEONARD\n\t\tWhy?"}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: It's my condition. I never know if I've already eaten, so I always just eat small\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: It's my condition. I never know if I've already eaten, so I always just eat small, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. It's my condition. I never know if I've already eaten, so I always just eat small The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The CAMERA PANS across the kitchen table where there is a\nSetting: derived from Pleasantville\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. KITCHEN. LATER ...\nACTION: The CAMERA PANS across the kitchen table where there is a full quart of milk and some uneaten co6kies. It finally comes\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Pleasantville: The CAMERA PANS across the kitchen table where there is a full quart of milk and some uneaten co6kies. It finally comes, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The CAMERA PANS across the kitchen table where there is a full quart of milk and some uneaten co6kies. It finally comes The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jay And Silent Bob Strike Back\n\nSCREENPLAY EXCERPT:\nEXT. DAM ROAD--DAY\n\n --pavement near the manhole. She sits there, looking down."}, {"role": "assistant", "content": "## SCENE 1 — EXT. DAM ROAD--DAY\nACTION: --pavement near the manhole. She sits there, looking down.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jay And Silent Bob Strike Back: --pavement near the manhole. She sits there, looking down., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. --pavement near the manhole. She sits there, looking down. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Black Screen\n\nSCREENPLAY EXCERPT:\nINT. X-FORCE WAR ROOM\n\n\t\t\t\tBARNETT\n\t\t\tThe both of you should go home. We\n\t\t\twill keep you abreast.\n\n\t\t\t\tSLOAN\n\t\t\tThis is looking like a needle in a\n\t\t\thaystack. It's hopeless.\n\nDelacroix holds Sloan as he leads her out. She is having\ndifficulty breathing with her asthma.\n\n\t\t\t\tDELACROIX\n\t\t\tThank you for your help.\n\n\t\t\t\tBARNETT\n\t\t\tThe X-Force always gets their man."}, {"role": "assistant", "content": "## SCENE 1 — INT. X-FORCE WAR ROOM\nACTION: The both of you should go home. We will keep you abreast.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: The both of you should go home. We will keep you abreast., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The both of you should go home. We will keep you abreast. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in CaptainScarletAndTheMysterons illustrating the trope 'BritishSeries'. Captain Scarlet and the Mysterons\n Gerry Anderson's New Captain Scarlet\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying BritishSeries.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for CaptainScarletAndTheMysterons, BritishSeries, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for BritishSeries. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: All right, that�s it! (He leans over\nSetting: derived from Saw\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. WIDE-EYED\nACTION: All right, that�s it! (He leans over and picks up his saw, sits down and\nDIALOGUE — Adam: (wide-eyed) I remember everything now. (Lawrence glances over at him.) I remember how I got here.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Saw: All right, that�s it! (He leans over and picks up his saw, sits down and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. All right, that�s it! (He leans over and picks up his saw, sits down and The camera pushes in with small amplitude at slow speed. Adam (S1) says: [English] (wide-eyed) I remember everything now. (Lawrence glances over at him.) I remember how I got here. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Final Destination\"\n\nSCREENPLAY EXCERPT:\nEXT. BEACH - NIGHT\n\nCAMERA, FACING the ocean, moves along the shore revealing a lone image sitting on the beach, looking into the Atlantic.\n\nThe waves are small bur rhythmic, an enternal metronome. Numb, Alex Browning sits in the sand searching for an answer somewhere in the darkness of the evening sea.\n\nBehind him, O.S., approaching in the sand, SOFT FOOTSTEPS. He listens, notes them. Yet rather than turn around... he looks up into the sky.\n\nALEX'S POV - THE SKY\n\nThe shoreline lights cast an orange haze on the stars, breaking through, infinitely above.\n\n ALEX (O.S.)\n Are they up there?\n\nWIDER\n\nClear Rivers approaches, barefoot in the sand. He neddn't turn to know who is behind him.\n\n ALEX (CONT'D)\n Somehow... is 180 still in flight?\n Somewhere... are they still safe?\n\nClear hasn't met eyes with Alex, but she sits nearby, looking into the sky.\n CLEAR\n When I was a kid, like, six or\n\t\tseven... I used to worry so much\n\t\tabout my parents dying. Like lying\n\t\tawake at night... just worrying. I\n\t\tloved them so much. I didn't want\n\t\tthem to get hurt. And what would\n\t\thappen to me? What would life be\n\t\tlike?\n (beat)\n Every night... it seemed.\n\nShe looks out at the stars. He doesn't eye her, either...\n\n ALEX\n Most kids do, I guess.\n\n CLEAR\n Most kids never have it happen.\n\nAlex hangs his head...\n\n CLEAR\n When I was ten... my dad went into a\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. BEACH - NIGHT\nACTION: CAMERA, FACING the ocean, moves along the shore revealing a lone image sitting on the beach, looking into the Atlantic. The waves are small bur rhythmic, an enternal metronome. Numb, Alex Browning sits in the sand searching for an answer somewhere in the darkness of the evening.\nDIALOGUE — Alex: Are they up there?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Final Destination\": CAMERA, FACING the ocean, moves along the shore revealing a lone image sitting on the beach, looking into the Atlantic. The waves are small bur rhythmic, an enternal metronome. Numb, Alex Browning sit, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CAMERA, FACING the ocean, moves along the shore revealing a lone image sitting on the beach, looking into the Atlantic. The waves are small bur rhythmic, an enternal metronome. Numb, Alex Browning sits in the sand searching for an answer somewhere in the darkness of the evening. The camera pushes in with small amplitude at slow speed. Alex (S1) says: [English] Are they up there? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Pauline . . . screaming hysterically.\nSetting: derived from Heavenly Creatures\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. WHARF/OCEANLINER - DAYBLACK & WHITE . . .\nACTION: Pauline . . . screaming hysterically. Juliet on the liner . . . tears streaming down her face.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Heavenly Creatures: Pauline . . . screaming hysterically. Juliet on the liner . . . tears streaming down her face., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Pauline . . . screaming hysterically. Juliet on the liner . . . tears streaming down her face. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Revenge\n\nSCREENPLAY EXCERPT:\nELAINE\n I just like to have and few drinks and\n just let the guy do whatever\n\n\n he wants. Would you close your eyes for a second? I wanna tell\n you a\n\n\n secret about my bra."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I just like to have and few drinks and just let the guy do whatever\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Revenge: I just like to have and few drinks and just let the guy do whatever, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I just like to have and few drinks and just let the guy do whatever The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Luke's tiny X-wing rockets away from the green planet of Dagobah and\nSetting: derived from The Empire Strikes Back\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. SPACE - PLANET DAGOBAH\nACTION: Luke's tiny X-wing rockets away from the green planet of Dagobah and\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Empire Strikes Back: Luke's tiny X-wing rockets away from the green planet of Dagobah and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Luke's tiny X-wing rockets away from the green planet of Dagobah and The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Puffy Shirt\n\nSCREENPLAY EXCERPT:\nYour hands.\n\n\n GEORGE\n What about them?\n\n\n WOMAN\n They're quite exquisite!\n\n\n GEORGE\n They are?\n\n\n WOMAN\n (Mesmerized) Extraordinary! Have you\n ever done any hand modeling?\n\n\n GEORGE\n Hand modeling? (Shakes his head 'No')\n\n\n\n WOMAN\n (Fishes a card out of her purse, then\n hands it to George) Here's my card.\n Why don't you, uh, give me a call? (Walks\n off)"}, {"role": "assistant", "content": "## SCENE 1 — INT. MESMERIZED\nACTION: They're quite exquisite! (Mesmerized) Extraordinary! Have you\nDIALOGUE — George: What about them?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Puffy Shirt: They're quite exquisite! (Mesmerized) Extraordinary! Have you, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They're quite exquisite! (Mesmerized) Extraordinary! Have you The camera pushes in with small amplitude at slow speed. George (S1) says: [English] What about them? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Army Of Darkness\n\nSCREENPLAY EXCERPT:\nTHE ARMY OF DARKNESS\nScreenplay by Sam Raimi and Ivan Raimi\n\n------------------------------------------------------------------------------\n1\t GRAINY BLACK AND WHITE - CLOSE-UP - A MAN'S DESPERATE FACE\t 1\n\t This is Ash, mid twenties, square jaw firmly set and a pair of\n\t haunted eyes which dart about quickly in fear. Ash speaks to\n\t the CAMERA with urgency:\n\t\t\t\t\t ASH\n\t\t\t Why would you say that I am\n\t\t\t insane? I wouldn't say that I've\n\t\t\t lost my mind simply because I've\n\t\t\t heard the voices and seen the\n\t\t\t godless things moving in the\n\t\t\t woods. If anything, I think more\n\t\t\t clearly now than ever before. I\n\t\t\t know now that there is such a\n\t\t\t thing as a living Evil. A dark\n\t\t\t and shapeless thing that lives\n\t\t\t not in the spaces we know, but\n\t\t\t between them. In the Dark. In the\n\t\t\t night. And it wants the exact same\n\t\t\t thing as you and I: a chance at\n\t\t\t warm life on this Earth. It\n\t\t\t doesn't care that is already had\n\t\t\t that chance ...once. Now listen\n\t\t\t closely because there isn't much\n\t\t\t time. Listen and believe, because\n\t\t\t it's all true.\n\t\t\t\t\t\t\t\t\t DISSOLVE TO:\n2\t EXT. A LOG CABIN - NIGHT\t\t\t\t\t\t\t\t2\n\t nestled in a dark forest. Through the window, we see the tiny\n\t figure of Ash picking up a book from a desk.\n\t\t\t\t\t ASH (V.O.)\n\t\t\t I first saw the damn thing at that\n\t\t\t blasted cabin. The Necronomicon.\n\t\t\t An ancient Sumarrian text, bound\n\t\t\t in human flesh and inked in human\n\t\t\t blood. It contained bizarre burial\n\t\t\t rites, prophesies...and\n\t\t\t instruction for demon\n\t\t\t resurrection. It was n"}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: Screenplay by Sam Raimi and Ivan Raimi ------------------------------------------------------------------------------\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Army Of Darkness: Screenplay by Sam Raimi and Ivan Raimi ------------------------------------------------------------------------------, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Screenplay by Sam Raimi and Ivan Raimi ------------------------------------------------------------------------------ The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Walter looks over at the Dude and bellows:\nSetting: derived from The Big Lebowski\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Walter looks over at the Dude and bellows: Dude, are you fucking this up?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Big Lebowski: Walter looks over at the Dude and bellows: Dude, are you fucking this up?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Walter looks over at the Dude and bellows: Dude, are you fucking this up? The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: These are the laws of my\nSetting: derived from Mrs. Teasdale\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: These are the laws of my administration: No one's allowed to\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Mrs. Teasdale: These are the laws of my administration: No one's allowed to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. These are the laws of my administration: No one's allowed to The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ON B&B, still staring at the statue.\nSetting: derived from \"Beavis And Butt-Head Do America\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HOTEL/CASINO - DUSK\nACTION: ON B&B, still staring at the statue. A security guard comes and drags B&B away.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Beavis And Butt-Head Do America\": ON B&B, still staring at the statue. A security guard comes and drags B&B away., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ON B&B, still staring at the statue. A security guard comes and drags B&B away. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The four hackers skate into Grand Central Station and head\nSetting: derived from Hackers\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. GRAND CENTRAL STATION.\nACTION: The four hackers skate into Grand Central Station and head to the lowest level. They meet up with Joey.\nDIALOGUE — Cereal: Hey Joey, you made it!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hackers: The four hackers skate into Grand Central Station and head to the lowest level. They meet up with Joey., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The four hackers skate into Grand Central Station and head to the lowest level. They meet up with Joey. The camera pushes in with small amplitude at slow speed. Cereal (S1) says: [English] Hey Joey, you made it! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade Out:\n\nSCREENPLAY EXCERPT:\nInt. Entrance Hall -- camera shooting up from side of stairs -\n- LS -- Dorothy, Lion, Tin Man, and Scarecrow enter at top of\nstairs -- CAMERA PANS with them as they run down the stairs --\nthey rush toward the entrance, but the huge doors swing shut\nin front of them -- MS -- The Four hammer at the door -- then\nreact, turn and look up to right f.g. as the Witch speaks\no.s. -- CAMERA TRUCKS forward --\n\n DOROTHY\n Oh!\n\nWITCH o.s. Going so soon? I wouldn't.... CS -- Witch and\nNikko looking down from the top of stairs --\n\n WITCH\n ...hear of it. Why, my little par-....\n\nMCS -- Scarecrow, Tin Man, Dorothy and the Lion by the door --\nWITCH o.s. ...ty's just beginning! MLS -- Witch and Nikko at\ntop of stairs -- she laughs -- CAMERA PANS down to reveal the\nWinkies rushing into the entrance hall -- MCS -- Scarecrow,\nTin Man, Dorothy and Lion huddled by door -- the Lion speaks -\n-\n\n LION\n Trapped! Trapped like mice -- er --\n rats! MLS - The Winkies advancing -\n\nMCS - The Four huddled by door - Tin Man speaks to Lion, who\ntries to roar -\n\n TIN MAN\n Go ahead - do something! Roar! Roar!\n\nMCS - The Winkie Guards advance with a menacing roar - MCS -\nThe Four huddled against door - the Lion reacts, speaks -\n\n LION\n What good'll it do us?\n\nELS -- Shooting down from the top of the hall, past the Witch\non landing in f.g. to the Winkies advancing to the four\nhuddled against the door -- the Winkies roar -- CS -- Witch\nand Nikko looking down from the to"}, {"role": "assistant", "content": "## SCENE 1 — INT. ENTRANCE HALL -- CAMERA SHOOTING UP FROM SIDE OF STAIRS -\nACTION: - LS -- Dorothy, Lion, Tin Man, and Scarecrow enter at top of stairs -- CAMERA PANS with them as they run down the stairs --\nDIALOGUE — Dorothy: Oh!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade Out:: - LS -- Dorothy, Lion, Tin Man, and Scarecrow enter at top of stairs -- CAMERA PANS with them as they run down the stairs --, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. - LS -- Dorothy, Lion, Tin Man, and Scarecrow enter at top of stairs -- CAMERA PANS with them as they run down the stairs -- The camera pushes in with small amplitude at slow speed. Dorothy (S1) says: [English] Oh! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: American History X\n\nSCREENPLAY EXCERPT:\nINT. BEDROOM - THE THREESOME\n\nDanny looks at the two.\n\n DANNY\nIf I don't finish this thing I'm dead.\n\n CHRIS\nJust tell 'em you'll do it tomorrow.\nYou can't miss the party.\n\n JASON\nFire pie Lizzy called and told us to\ngrab your ass.\n\n DANNY\nShe did?\n\n CHRIS\n (nodding)\nYou gotta hit that shit. It's the\nonly fun thing that's legal anymore.\n\nThe PHONE RINGS and Danny grabs it.\n\n DANNY\nHello?"}, {"role": "assistant", "content": "## SCENE 1 — INT. BEDROOM - THE THREESOME\nACTION: Danny looks at the two. If I don't finish this thing I'm dead.\nDIALOGUE — Danny: If I don't finish this thing I'm dead.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American History X: Danny looks at the two. If I don't finish this thing I'm dead., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Danny looks at the two. If I don't finish this thing I'm dead. The camera pushes in with small amplitude at slow speed. Danny (S1) says: [English] If I don't finish this thing I'm dead. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Saw\n\nSCREENPLAY EXCERPT:\nAny idea who that is?\n\n JIGSAW\n (on tape) Let the game begin.\n\n ADAM\n He knows us.\n\n LAWRENCE\n Wait a minute. (rewinds more)\n\n ADAM\n (false hope) What do you think? Probably\n a joke, right?\n\n LAWRENCE\n Shh, shh! (presses play)\n\n JIGSAW\n (on tape) Let the game begin.\n\n LAWRENCE\n (softly) Listen...\n\n JIGSAW\n (on tape, in a VERY hushed tone after\n a brief silence)...Follow your heart.\n\n\n Lawrence stops the tape.\n\n ADAM\n What the hell does Follow your heart\n mean?\n\n Lawrence is looking around, on the walls and everywhere else\n in the bathroom for a heart. He spots a small one drawn in brown\n on the toilet next to Adam and points.\n\n LAWRENCE\n There! Right next to you, on the toilet.\n\n\n Adam looks, sees the toilet. Slowly, he bends down, resting his\n hands on the seat, looking down into the bowl of liquid shit,\n not really s"}, {"role": "assistant", "content": "## SCENE 1 — INT. ON TAPE\nACTION: Any idea who that is? (on tape) Let the game begin.\nDIALOGUE — Jigsaw: (on tape) Let the game begin.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Saw: Any idea who that is? (on tape) Let the game begin., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Any idea who that is? (on tape) Let the game begin. The camera pushes in with small amplitude at slow speed. Jigsaw (S1) says: [English] (on tape) Let the game begin. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Bride Of Leprechaun\n\nSCREENPLAY EXCERPT:\nWhere the hell are you?\n\n LEPRECHAUN (O.S.)\n You'll have to open the door, it's\n wrought iron. Remember? Didn't you\n read the book?\n\n In pain, Morty reaches up and starts to turn the combination.\n\n98 INT. CLOSET - CONTINUOUS 98\n\n Cody punches at the glass, repeatedly, it's too thick.\n\n CODY\n Stop it!! Morty!\n\n Cody looks around the closet for something to break the\n glass.\n\n99 INT. MANAGER'S OFFICE - FOLLOWING 99\n\n Screaming in pain, Morty dials in the numbers. When finished,\n he opens up the safe door, and falls back, exhausted. The\n Leprechaun steps out. He dusts himself off and beams with\n pride.\n 78.\n\n\n\n LEPRECHAUN\n Your second wish has been granted.\n\n MORTY"}, {"role": "assistant", "content": "## SCENE 1 — INT. O.S.\nACTION: Where the hell are you? You'll have to open the door, it's\nDIALOGUE — Leprechaun: You'll have to open the door, it's wrought iron. Remember? Didn't you read the book?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bride Of Leprechaun: Where the hell are you? You'll have to open the door, it's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Where the hell are you? You'll have to open the door, it's The camera pushes in with small amplitude at slow speed. Leprechaun (S1) says: [English] You'll have to open the door, it's wrought iron. Remember? Didn't you read the book? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Any idea who that is?\nSetting: derived from Saw\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ON TAPE\nACTION: Any idea who that is? (on tape) Let the game begin.\nDIALOGUE — Jigsaw: (on tape) Let the game begin.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Saw: Any idea who that is? (on tape) Let the game begin., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Any idea who that is? (on tape) Let the game begin. The camera pushes in with small amplitude at slow speed. Jigsaw (S1) says: [English] (on tape) Let the game begin. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Django Unchained\n\nSCREENPLAY EXCERPT:\nPlease come inside Fraulein.\n She does, and just'as Lara Lee is to say something, Schultz says,\n \"Thank You very much,\" and closes the door in her face. Miss Lara looks\n to her Mammy, and the two women head off nonplussed.\n\n INT - SCHULTZ'S GUEST ROOM (BIG HOUSE) - LATE AFTERNOON\n With the door closed, Schultz turns to the weak, frightened,\n disoriented girl.\n He gives her a pleasant smile.\n Dr.SCHULTZ\n They call you Hildi, but your real name\n is Broomhilda, isn't it?\n\n BROOMHILDA\n Yes. How do you know that?\n Dr.SCHULTZ\n Stands to reason who ever taught you\n German would also give you a German name.\n Can I pour you a glass of water,\n Broomhilda?\n Hearing her name being spoke properly for the first time in awhile, not\n to mention with a German accent, does have a bit of a calming influence\n on the frightened girl.\n\n\n\n\n\n\n\n\n\n 107\n\n WE CUT TO DJANGO\n on the other side of the adjoining door, listening, waiting for his cue\n to present himself.\n BACK TO BROOMHILDA AND Dr.SCHULTZ\n As Dr.Schultz calmly pours the young lady a glass of water, he begins\n talking to her in GERMAN SUBTITLED IN ENGLISH;\n Dr.SCHULTZ"}, {"role": "assistant", "content": "## SCENE 1 — INT. BIG HOUSE\nACTION: Please come inside Fraulein. She does, and just'as Lara Lee is to say something, Schultz says,\nDIALOGUE — Broomhilda: Yes. How do you know that? Dr.SCHULTZ Stands to reason who ever taught you German would also give you a German name. Can I pour you a glass of water,...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Django Unchained: Please come inside Fraulein. She does, and just'as Lara Lee is to say something, Schultz says,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Please come inside Fraulein. She does, and just'as Lara Lee is to say something, Schultz says, The camera pushes in with small amplitude at slow speed. Broomhilda (S1) says: [English] Yes. How do you know that? Dr.SCHULTZ Stands to reason who ever taught you German would also give you a German name. Can I pour you a glass of water,... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Good News, Bad News\n\nSCREENPLAY EXCERPT:\nJERRY\n (The phone rings. He picks it up and\n says:) If you know what happened in\n the Mets-game, don't say anything, I\n taped it, hello... Yeah, no, I'm sorry,\n you have the wrong number...Yeah, no\n (somebody knocks at the door) Yeah?"}, {"role": "assistant", "content": "## SCENE 1 — INT. SOMEBODY KNOCKS AT THE DOOR\nACTION: (The phone rings. He picks it up and says:) If you know what happened in\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Good News, Bad News: (The phone rings. He picks it up and says:) If you know what happened in, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. (The phone rings. He picks it up and says:) If you know what happened in The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A COP rushes toward him, seemingly from\nSetting: derived from American Psycho\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ALLEY - NIGHT\nACTION: A COP rushes toward him, seemingly from out of nowhere, and tackles him, trying to get the gun away\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American Psycho: A COP rushes toward him, seemingly from out of nowhere, and tackles him, trying to get the gun away, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A COP rushes toward him, seemingly from out of nowhere, and tackles him, trying to get the gun away The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: BRUCE and VICKI enter. He takes her coat, drops it on a\nSetting: derived from Batman\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. WAYNE MANOR - NIGHT\nACTION: BRUCE and VICKI enter. He takes her coat, drops it on a chair by the door. VICKI is giddy, all champagned up.\nDIALOGUE — Vicki: -- but it's not fair. I'm half drunk and you're not even --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman: BRUCE and VICKI enter. He takes her coat, drops it on a chair by the door. VICKI is giddy, all champagned up., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. BRUCE and VICKI enter. He takes her coat, drops it on a chair by the door. VICKI is giddy, all champagned up. The camera pushes in with small amplitude at slow speed. Vicki (S1) says: [English] -- but it's not fair. I'm half drunk and you're not even -- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Glee illustrating the trope 'GayPanic'. In-Universe example: Santana from Glee manages to throw one of these against herself. When she and Brittany, her BFF and occasional sex-buddy, are necking, Brittany brings up wanting to sing Melissa Etheridge's \"Come to My Window\" with Santana, who promptly freaks out and hurriedly explains that she is not making out with her because she is in love with her and wants to make \"ladybabies\" with her;...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying GayPanic.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Glee, GayPanic, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for GayPanic. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in AgentsOfSHIELD illustrating the trope 'BigBadTriumvirate'. The second season of Agents of S.H.I.E.L.D. starts with a Big-Bad Ensemble of Daniel Whitehall and Calvin Zabo, but in \"A Hen in the Wolf House\", they team up to weaponize the Obelisk and destroy S.H.I.E.L.D. Except that Calvin Zabo actually wants revenge on Whitehall for killing his wife, Jiaying years earlier, and the alliance falls apart halfway through the season.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying BigBadTriumvirate.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for AgentsOfSHIELD, BigBadTriumvirate, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for BigBadTriumvirate. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Neverending Story\n\nSCREENPLAY EXCERPT:\nEXT. VIEW-Ivory Tower\n\n\n\nWide angled view of the nearing Ivory Tower. It shimmers in its only light as the sun sets behind it.\n\nclass=MsoBodyText2>\n\nNighthob:\n\n Stay awake. I told you to stay awake.\n\n\n\nWe span the whole length of this Ivory Tower. It is said to be as big as a city. We pass over the base of the tower, then it�s mid section, and finally we near the very top of the structure. It is very intricate here. We begin to zoom in on the Magnolia pavilion. The very tip of the Ivory Tower, the place that the Empress lives.\n\n\n\n\n\nNighthob:\n\n The home of the Empress. She's our only hope.\n\n\n\nFADE OUT:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. VIEW-IVORY TOWER\nACTION: Wide angled view of the nearing Ivory Tower. It shimmers in its only light as the sun sets behind it. Stay awake. I told you to stay awake.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Neverending Story: Wide angled view of the nearing Ivory Tower. It shimmers in its only light as the sun sets behind it. Stay awake. I told you to stay awake., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Wide angled view of the nearing Ivory Tower. It shimmers in its only light as the sun sets behind it. Stay awake. I told you to stay awake. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Int. Publishing House Reception Area - Day\n\nSCREENPLAY EXCERPT:\nINT. JOEL'S APARTMENT - NIGHT\n\nJoel and Clementine sit and eat dinner in front of the TV.\nIt's hard to make out what they're watching.\nThey sit on opposite ends of the couch. They look bored.\nThe scene quickly degenerates. The room fades.\n\n\t\tPATRICK'S VOICE\n\tOkay, but there's more.\n\nJoel listens.\n\n\t\tPATRICK'S VOICE (CONT'D)\n\tAfter we did her, I went to where she\n\tworked and I asked her out.\n\nJoel looks over at the faded Clementine across the couch.\nShe stares straight ahead at the TV.\n\n\t\tSTAN'S VOICE\n\tPatrick... do you know how unethical...\n\n\t\tJOEL\n\tThat must be the guy I saw you with.\n\n\t\tVOICE-OVER\n\tIn the bookstore that night. The skinny\n\tguy.\n\nThere's a click and Joel finds himself in --"}, {"role": "assistant", "content": "## SCENE 1 — INT. JOEL'S APARTMENT - NIGHT\nACTION: Joel and Clementine sit and eat dinner in front of the TV. It's hard to make out what they're watching.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int. Publishing House Reception Area - Day: Joel and Clementine sit and eat dinner in front of the TV. It's hard to make out what they're watching., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Joel and Clementine sit and eat dinner in front of the TV. It's hard to make out what they're watching. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Les Miserables\n\nSCREENPLAY EXCERPT:\n58.\n\n BEGGARS CHORUS (cont'd)\n Look down, look down,\n Upon your fellow man!\n\nThe fine ladies and grand gentlemen in the carriages avert\ntheir eyes, or raise the blinds of their carriage windows\nto shut out the sight of the losers of their world.\n\nGavroche, bounding over their heads, evading the swipes of\nliveried footmen, lands on the running board of one\nparticularly grand carriage and begs/taunts the RICH\nOCCUPANT.\n\n GAVROCHE\n `Ow do you do? My name's\n Gavroche!\n These are my people, here's my\n patch.\n Not much to look at - nothing\n posh!\n Nothing that you'd call up to\n scratch.\n This is my school, my high\n society!\n Here in the slums of St Michel.\n We live on crumbs of humble piety\n Tough on the teeth - but what the\n hell!\n Think you're poor? Think you're\n free?\n Follow me! Follow me!\n\n BEGGARS CHORUS\n Look down and show some mercy if\n you can!\n Look down, look down, upon your\n fellow man!\n\nSTILL RUNNING WITH Gavroche as he hops onto the back of\nanother very grand carriage, the traffic now moving at\nlast, hitching a ride on the back - one or two of his gang\nhop on back of carriage with him, the others run panting\nafter to hear his political lecture.\n\n GAVROCHE\n There was a time we killed the\n King\n We tried to change the world too\n fast.\n Now we hav"}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: Look down, look down, Upon your fellow man!\nDIALOGUE — Beggars Chorus: Look down, look down, Upon your fellow man!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Les Miserables: Look down, look down, Upon your fellow man!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Look down, look down, Upon your fellow man! The camera pushes in with small amplitude at slow speed. Beggars Chorus (S1) says: [English] Look down, look down, Upon your fellow man! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Biff's car is just pulling onto the street when David sprints\nSetting: derived from Pleasantville\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. SODA SHOP.\nACTION: Biff's car is just pulling onto the street when David sprints into the parking lot. He takes off after the car, sprinting\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Pleasantville: Biff's car is just pulling onto the street when David sprints into the parking lot. He takes off after the car, sprinting, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Biff's car is just pulling onto the street when David sprints into the parking lot. He takes off after the car, sprinting The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Bread illustrating the trope 'BadassFamily'. NEVER mess with one of the Boswells, unless you want the wrath of the entire family descending on you. Which you really, really don't.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying BadassFamily.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Bread, BadassFamily, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for BadassFamily. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: EVELYN WILLIAMS, Patrick Bateman's\nSetting: derived from American Psycho\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TAXI - EVENING\nACTION: EVELYN WILLIAMS, Patrick Bateman's fianc�e, is making notes with a gold Cross pen and sipping\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American Psycho: EVELYN WILLIAMS, Patrick Bateman's fianc�e, is making notes with a gold Cross pen and sipping, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. EVELYN WILLIAMS, Patrick Bateman's fianc�e, is making notes with a gold Cross pen and sipping The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: CAMERA is LOW ANGLE as Tim backs up after his attack. We notice\nSetting: derived from \"Natural Born Killers\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. GRACE'S HOME - NIGHT\nACTION: CAMERA is LOW ANGLE as Tim backs up after his attack. We notice the High School Ring he wears on his right fist as he steps into\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Natural Born Killers\": CAMERA is LOW ANGLE as Tim backs up after his attack. We notice the High School Ring he wears on his right fist as he steps into, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CAMERA is LOW ANGLE as Tim backs up after his attack. We notice the High School Ring he wears on his right fist as he steps into The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: 1 Ext. Cityscape Sunset 1\n\nSCREENPLAY EXCERPT:\nYou know, smoking's bad for you.\n\n JILL\n It's my fucking life.\n\n SAM"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You know, smoking's bad for you. It's my fucking life.\nDIALOGUE — Jill: It's my fucking life.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from 1 Ext. Cityscape Sunset 1: You know, smoking's bad for you. It's my fucking life., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You know, smoking's bad for you. It's my fucking life. The camera pushes in with small amplitude at slow speed. Jill (S1) says: [English] It's my fucking life. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Agent Brown hears Smith on his earphone.\nSetting: derived from The Matrix\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HALL (MATRIX) - DAY\nACTION: Agent Brown hears Smith on his earphone.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Matrix: Agent Brown hears Smith on his earphone., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Agent Brown hears Smith on his earphone. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Allie's eyes flare, Louisa stifles a sob. The Cowboys\nSetting: derived from Tombstone\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Allie's eyes flare, Louisa stifles a sob. The Cowboys snicker. Wyatt clenches his teeth, still staring straight\nDIALOGUE — Curly Bill: Take Frank and finish it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Tombstone: Allie's eyes flare, Louisa stifles a sob. The Cowboys snicker. Wyatt clenches his teeth, still staring straight, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Allie's eyes flare, Louisa stifles a sob. The Cowboys snicker. Wyatt clenches his teeth, still staring straight The camera pushes in with small amplitude at slow speed. Curly Bill (S1) says: [English] Take Frank and finish it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Big Lebowski\n\nSCREENPLAY EXCERPT:\nOkay Dude. Have it your way.\n\nThe Dude unclips the phone from the battery pack.\n\n\t\t\t\tWALTER\n\t\tBut they're amateurs.\n\nThe Dude glares at Walter. Into the phone:\n\n\t\t\t\tDUDE\n\t\tDude here.\n\n\t\t\t\tVOICE\n\t\tOkay, vee proceed. But only if there\n\t\tis no funny stuff.\n\n\t\t\t\tDUDE\n\t\tYeah.\n\n\t\t\t\tVOICE\n\t\tSo no funny stuff. Okay?\n\n\t\t\t\tDUDE\n\t\tHey, just tell me where the fuck you\n\t\twant us to go.\n\nA HIGHWAY SIGN: SIMI VALLEY ROAD\n\nIt flashes by in the headlights of the roaring car.\n\n\t\t\t\tDUDE\n\t\tThat was the sign.\n\nWalter wrestles the car onto the two-lane road.\n\n\t\t\t\tWALTER\n\t\tYeah. So as long as we get her back,\n\t\tnobody's in a position to complain.\n\t\tAnd we keep the baksheesh.\n\n\t\t\t\tDUDE\n\t\tTerrific, Walter. But you haven't\n\t\ttold me how we get her back. Where\n\t\tis she?\n\n\t\t\t\tWALTER\n\t\tThat's the simple part, Dude. When\n\t\twe make the handoff, I grab the guy\n\t\tand beat it out of him.\n\nHe looks at the Dude.\n\n\t\t\t\tWALTER\n\t\t...Huh?\n\n\t\t\t\tDUDE\n\t\tYeah. That's a great plan, Walter.\n\t\tThat's fucking ingenious, if I\n\t\tunderstand it correctly. That's a\n\t\tSwiss fucking watch.\n\n\t\t\t\tWALTER\n\t\tThaaat's right, Dude. The beauty of\n\t\tthis is its simplicity. If the plan\n\t\tgets too complex something always\n\t\tgoes wrong. If there's one thing I\n\t\tlearned in Nam--\n\nThe phone chirps.\n\n\t\t\t\tDUDE\n\t\tDude.\n\n\t\t\t\tVOICE\n\t\tYou are approaching a vooden britch.\n\t\tWhen you cross it you srow ze bag\n\t\tfrom ze left vindow of ze moving\n\t\tkar. Do not slow down. Vee vatch\n\t\tyou.\n\nClick. Dial tone.\n\n\t\t\t\tDUDE\n\t\tFUCK.\n\n\t\t\t\tWALTER\n\t\tWhat'd he say? Where's the hand-\n\t\toff?\n\n\t\t\t\tDUDE\n\t\tThere is no fucking hand-off, Walter!\n\t\tAt a woo"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Okay Dude. Have it your way. The Dude unclips the phone from the battery pack.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Big Lebowski: Okay Dude. Have it your way. The Dude unclips the phone from the battery pack., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Okay Dude. Have it your way. The Dude unclips the phone from the battery pack. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I just like to have and few drinks and\nSetting: derived from The Revenge\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I just like to have and few drinks and just let the guy do whatever\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Revenge: I just like to have and few drinks and just let the guy do whatever, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I just like to have and few drinks and just let the guy do whatever The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Revenge\n\nSCREENPLAY EXCERPT:\nLEVITAN\n God bless you.\n\n\n ELAINE\n Oh! Thank you. Thank you very much.\n (blowing nose) Really. I mean\n\n\n that. I am not one of those people who give insincere thank you's.\n No\n\n\n sir. No sir. When I thank someone I really thank them. So, thank...\n\n\n\n yoooou!\n\n\n LEVITAN\n You're welcome.\n\n\n ELAINE\n People don't say \" God bless you \" as\n much as they used to. Have\n\n\n you noticed that?\n\n\n LEVITAN\n No.\n\n\n ELAINE\n So, I'm going to a nudist colony next\n week.\n\n\n LEVITAN\n Nudist colony?\n\n\n ELAINE\n Oh, yeah. Yeah. I love nudist colonies.\n They help me..unwind. Aah!\n\n\n LEVITAN\n (laughing) I'd never been to a nudist\n colony.\n\n\n ELAINE\n Oh, really? Oh, you should go. They're\n great. They're great. Of\n\n\n course, when it's over, it's - it's hard to get used to all this\n\n\n\n clothing, you know. So, a lot of times, I'll just lock the door\n to my\n\n\n office and I'll just sit there naked.\n\n\n LEVITAN\n Seriously?\n\n\n ELAINE\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. BLOWING NOSE\nACTION: Oh! Thank you. Thank you very much. (blowing nose) Really. I mean\nDIALOGUE — Elaine: Oh! Thank you. Thank you very much. (blowing nose) Really. I mean\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Revenge: Oh! Thank you. Thank you very much. (blowing nose) Really. I mean, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Oh! Thank you. Thank you very much. (blowing nose) Really. I mean The camera pushes in with small amplitude at slow speed. Elaine (S1) says: [English] Oh! Thank you. Thank you very much. (blowing nose) Really. I mean The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Roger:\tHIS NAME'S PETER. HE'S ALRIGHT.\nSetting: derived from Dawn Of\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TO THE OTHER POLICEMEN\nACTION: Roger:\tHIS NAME'S PETER. HE'S ALRIGHT. The three are already moving toward the helicopter.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dawn Of: Roger:\tHIS NAME'S PETER. HE'S ALRIGHT. The three are already moving toward the helicopter., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Roger:\tHIS NAME'S PETER. HE'S ALRIGHT. The three are already moving toward the helicopter. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A single light is burning in the kitchen window.\nSetting: derived from Pleasantville\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. PARKER HOUSE. NIGHT.\nACTION: A single light is burning in the kitchen window. Do you want some more cookies?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Pleasantville: A single light is burning in the kitchen window. Do you want some more cookies?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A single light is burning in the kitchen window. Do you want some more cookies? The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: S T R A N G E D A Y S\n\nSCREENPLAY EXCERPT:\nINT. MAIN FLOOR\n\n Tran impatiently gestures to Corto.\n\n TRAN\n Go get her, Joey. Stupid bitch.\n (to Max)\n We'll talk about this later.\n\n CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — INT. MAIN FLOOR\nACTION: Tran impatiently gestures to Corto. Go get her, Joey. Stupid bitch.\nDIALOGUE — Tran: Go get her, Joey. Stupid bitch. We'll talk about this later.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from S T R A N G E D A Y S: Tran impatiently gestures to Corto. Go get her, Joey. Stupid bitch., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Tran impatiently gestures to Corto. Go get her, Joey. Stupid bitch. The camera pushes in with small amplitude at slow speed. Tran (S1) says: [English] Go get her, Joey. Stupid bitch. We'll talk about this later. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Wide angled view of the nearing Ivory Tower. It shimmers in its only light as the sun sets behind it.\nSetting: derived from The Neverending Story\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. VIEW-IVORY TOWER\nACTION: Wide angled view of the nearing Ivory Tower. It shimmers in its only light as the sun sets behind it. Stay awake. I told you to stay awake.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Neverending Story: Wide angled view of the nearing Ivory Tower. It shimmers in its only light as the sun sets behind it. Stay awake. I told you to stay awake., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Wide angled view of the nearing Ivory Tower. It shimmers in its only light as the sun sets behind it. Stay awake. I told you to stay awake. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Hitchhiker'S Guide To The Galaxy\n\nSCREENPLAY EXCERPT:\nINT. ARTHUR DENT'S BEDROOM - MORNING 2\n\n CLOSE ON the clock radio from which the voice emanates.\n A hand shuts it off. ARTHUR DENT is awake. 30's. Kind\n faced. His feet land perfectly in slippers as he stands\n and shuffles out of his very neat but very boring room.\n\n3 IN THE BATHROOM - Arthur gargles in front of the mirror. 3\n He bends over to spit revealing a YELLOW BULLDOZER\n through the window, crawling toward his house. A TEA\n KETTLE WHISTLES O.S.\n\n4 INT. ARTHUR DENT'S KITCHEN - CONTINUOUS 4\n\n Arthur sets the newspaper (with the headline \"DOLPHINS\n VANISH\") on the table, removes the kettle. Doesn't notice\n through the window 3 MORE LARGE YELLOW BULLDOZERS.\n\n He pours a cup of tea, sets the pot down next o his\n mobile phone. He picks it up, stares at a DIGITAL PHOTO\n ON THE SCREEN of himself and a very pretty woman (Tricia)\n at a costume party.\n\n He presses the phone to his head as if it were a cold\n compress. Suddenly, there's a LOUD RUMBLING o.s. The\n tea cup RATTLES. Arthur looks out the window. Now he\n sees the bulldozers.\n\n The tea cup shatters on the floor.\n CUT TO:\n5 EXT. ARTHUR DENT'S HOUSE - MORNING 5\n\n Arthur, dressed in pajamas and robe, lies in the dirt in\n front of a bulldozer, blocking it from destroying his\n house. He yells into his mobile phone.\n\n ARTHUR\n (into phone)\n What do you mean he's gone out to lunch?\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. ARTHUR DENT'S BEDROOM - MORNING 2\nACTION: CLOSE ON the clock radio from which the voice emanates. A hand shuts it off. ARTHUR DENT is awake. 30's. Kind\nDIALOGUE — Arthur: What do you mean he's gone out to lunch? You just said he was still at breakfast!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hitchhiker'S Guide To The Galaxy: CLOSE ON the clock radio from which the voice emanates. A hand shuts it off. ARTHUR DENT is awake. 30's. Kind, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CLOSE ON the clock radio from which the voice emanates. A hand shuts it off. ARTHUR DENT is awake. 30's. Kind The camera pushes in with small amplitude at slow speed. Arthur (S1) says: [English] What do you mean he's gone out to lunch? You just said he was still at breakfast! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Engrenages illustrating the trope 'GanglandDriveBy'. One takes place in Engrenages, when a man is gunned down in front of Pierre's house by a motorcycle pillion passenger, because of Pierre's flirtation with being a gangster-servicing Amoral Attorney. This leads Pierre to a moral crisis.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying GanglandDriveBy.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Engrenages, GanglandDriveBy, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for GanglandDriveBy. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: It's my condition. I never know if I've\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: It's my condition. I never know if I've already eaten, so I always just eat small\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: It's my condition. I never know if I've already eaten, so I always just eat small, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. It's my condition. I never know if I've already eaten, so I always just eat small The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Blade Runner\n\nSCREENPLAY EXCERPT:\nINT. DRESSING ROOM - NIGHT 64\n\n Musty and cramped. A portable shower, a dressing table\n and not much else. Salome takes the snake from around\n her shoulders and lays it on the dressing table. Deck-\n ard watches it undulate into the warmth of the lights.\n\n DECKARD\n It that mother real?\n\n SALOME\n Of course he's not real. You think\n I'd be working here if I could\n afford a real snake?\n\n DECKARD\n It's a good job.\n\n SALOME\n You mean the snake.\n\n Deckard nods. There's not much costume to take off but\n she's doing it.\n\n SALOME\n The best.\n\n DECKARD\n Does it eat?\n\n SALOME\n Come on.\n\n His hand reaches out to touch it. As his fingers make\n contact there's an electric \"snap.\" He jerks his hand\n back from the shock.\n\n SALOME\n Jeezus!\n\n DECKARD\n Sorry.\n\n SALOME\n Hey! Do your job but don't wreck\n mine, huh?\n\n She slides behind the screen and turns on the shower.\n Deckard starts creeping around pacing around the room\n like he's inspecting the walls.\n"}, {"role": "assistant", "content": "## SCENE 1 — INT. DRESSING ROOM - NIGHT 64\nACTION: Musty and cramped. A portable shower, a dressing table and not much else. Salome takes the snake from around\nDIALOGUE — Deckard: It that mother real?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade Runner: Musty and cramped. A portable shower, a dressing table and not much else. Salome takes the snake from around, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Musty and cramped. A portable shower, a dressing table and not much else. Salome takes the snake from around The camera pushes in with small amplitude at slow speed. Deckard (S1) says: [English] It that mother real? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Ext. Manhattan Island - Day\n\nSCREENPLAY EXCERPT:\nEXT. BROADWAY - NIGHT (CONTINUOUS ACTION)\n\nThe Statue of Liberty is walking up Broadway approaching 42nd Street,\nwith \"Higher and Higher\" BOOMING from the SPEAKERS inside. A great cheer\ngoes up, and the crowd goes wild with joy, dancing and singing along with\nthe MUSIC."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BROADWAY - NIGHT (CONTINUOUS ACTION)\nACTION: The Statue of Liberty is walking up Broadway approaching 42nd Street, with \"Higher and Higher\" BOOMING from the SPEAKERS inside. A great cheer\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ext. Manhattan Island - Day: The Statue of Liberty is walking up Broadway approaching 42nd Street, with \"Higher and Higher\" BOOMING from the SPEAKERS inside. A great cheer, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The Statue of Liberty is walking up Broadway approaching 42nd Street, with \"Higher and Higher\" BOOMING from the SPEAKERS inside. A great cheer The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: As Credits Roll\n\nSCREENPLAY EXCERPT:\nINT. LINGERIE SHOP - DAY\n\nMARIAN and RAQUEL, two exquisite young women, are looking\nover the merchandise.\n\n\t\t\t\tRAQUEL\n\t\tDid you see that about the guy who\n\t\tchopped up his wife into a million\n\t\tpieces?\n\n\t\t\t\tMARIAN\n\t\tHow could I miss it? The TV won't quit\n\t\twith that stuff.\n\n\t\t\t\tRAQUEL\n\t\tThey're gonna cook him.\n\n\t\t\t\tMARIAN\n\t\tAndy's from Utah. He says there you have\n\t\ta choice... You can die by hanging or by\n\t\tfiring squad.\n\n\t\t\t\tRAQUEL\n\t\tWhich would you choose?\n\nMarian holds up a black teddy to her body.\n\n\t\t\t\tMARIAN\n\t\tAndy would go for this, don't you\n\t\tthink?... Firing squad, definitely.\n\n\t\t\t\tRAQUEL\n\t\tDo they aim for the head or for the\n\t\theart?\n\n\t\t\t\tMARIAN\n\t\tThe heart, I guess.\n\n\t\t\t\tRAQUEL\n\t\tI wouldn't... The brain would know what's\n\t\tgoing on. Your heart would be ripped\n\t\topen trying to pump blood, blood pouring\n\t\tinto the chest cavity. Savage pain,\n\t\tMarian.\n\nRaquel takes a red teddy and holds it up to her chest.\n\n\t\t\t\tMARIAN\n\t\tOh, that's hot... So you'd rather be\n\t\thung, huh?\n\nThey both giggle at the obvious joke.\n\n\t\t\t\tRAQUEL\n\t\tAbsolutely... Soon as your neck snaps,\n\t\tyou black out. It might take a while for\n\t\tthe body to die, but you wouldn't feel\n\t\tit.\n\nMarian reaches for a pair of panties with a hole in the crotch.\n\n\t\t\t\tMARIAN\n\t\tYou might be right, Raquel.\n\nMarian sticks a finger through the hole in the panties and\nwiggles it. The girls giggle. Raquel sees Andy walking into\nthe store. He is sneaking up behind Marian, and motions with\na finger to his lips to Raquel not to say anything. When\nAndy gets directly behind Marian, he puts his hands over her\neyes.\n\n\t\t\t"}, {"role": "assistant", "content": "## SCENE 1 — INT. LINGERIE SHOP - DAY\nACTION: MARIAN and RAQUEL, two exquisite young women, are looking over the merchandise.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from As Credits Roll: MARIAN and RAQUEL, two exquisite young women, are looking over the merchandise., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. MARIAN and RAQUEL, two exquisite young women, are looking over the merchandise. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Tran impatiently gestures to Corto.\nSetting: derived from S T R A N G E D A Y S\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. MAIN FLOOR\nACTION: Tran impatiently gestures to Corto. Go get her, Joey. Stupid bitch.\nDIALOGUE — Tran: Go get her, Joey. Stupid bitch. We'll talk about this later.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from S T R A N G E D A Y S: Tran impatiently gestures to Corto. Go get her, Joey. Stupid bitch., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Tran impatiently gestures to Corto. Go get her, Joey. Stupid bitch. The camera pushes in with small amplitude at slow speed. Tran (S1) says: [English] Go get her, Joey. Stupid bitch. We'll talk about this later. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Empire Strikes Back\n\nSCREENPLAY EXCERPT:\nEXT. CLOUD CITY - LANDING PLATFORM\n\nLando stops ten feet from Han. The two men eye each other carefully.\nLando shakes his head.\n\n\t\t\t\tLANDO\n\t\tWhy, you slimy, double-crossing,\n\t\tno-good swindler! You've got a\n\t\tlot of guts coming here, after\n\t\twhat you pulled.\n\nHan points to himself innocently, mouthing, \"Me?\"\n\nLando moves threateningly toward Han. Suddenly, he throws his arms\naround his startled, long-lost friend and embraces him.\n\n\t\t\t\tLANDO\n\t\t\t(laughs)\n\t\tHow you doing, you old pirate?\n\t\tSo good to see you! I never thought\n\t\tI'd catch up with you again. Where\n\t\tyou been?\n\nThe two old friends embrace, laughing and chuckling."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CLOUD CITY - LANDING PLATFORM\nACTION: Lando stops ten feet from Han. The two men eye each other carefully. Lando shakes his head.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Empire Strikes Back: Lando stops ten feet from Han. The two men eye each other carefully. Lando shakes his head., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Lando stops ten feet from Han. The two men eye each other carefully. Lando shakes his head. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Gabrielle, face obscured by her hoodie and trench coat,\nSetting: derived from Malignant\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HOSPITAL - NIGHT\nACTION: Gabrielle, face obscured by her hoodie and trench coat, enters through the automatic doors.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Malignant: Gabrielle, face obscured by her hoodie and trench coat, enters through the automatic doors., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Gabrielle, face obscured by her hoodie and trench coat, enters through the automatic doors. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"The Thing\"\n\nSCREENPLAY EXCERPT:\nEXT. COMPOUND\n\n The dog reaches Norris and Bennings, as they awkwardly wade\n through the snow, toward the downed copter.\n\n THE SURVIVOR\n\n of the crash, his eyes crazed with determination, struggles\n to his feet. Heedless of his companion, he double-times his\n way to the men and the dog. He reloads his gun and bellows\n in his Scandinavian tongue.\n\n Norris and Bennings have no idea what he is saying.\n\n The survivor waves his arms as if shooting them off, screaming\n as he does so; his face now caked with blood.\n\n The two men are bewildered. The dog jumps up, licking and\n pawing them, imploring for safety.\n\n Blam!! The visitor fires. The men jump back in disbelief.\n\n NORRIS\n What the fu...\n\n Blam! Blam! The crazed visitor screams and fires as he stalks\n after them. His countenance ablaze, mad. Ice and snow kick\n up about the terrified Americans. A bullet smacks into the\n dog's hip, sending it skidding and howling in pain.\n\n Childs, the black man by the snowmobile, takes cover, diving\n behind his machine.\n\n Bennings is hit. Norris pulls, drags him back toward the\n compound. The dog crawls along beside them.\n\n The intruder is relentless in his assail. He runs, screaming,\n firing, screaming, reloading and "}, {"role": "assistant", "content": "## SCENE 1 — EXT. COMPOUND\nACTION: The dog reaches Norris and Bennings, as they awkwardly wade through the snow, toward the downed copter.\nDIALOGUE — The Survivor: of the crash, his eyes crazed with determination, struggles to his feet. Heedless of his companion, he double-times his way to the men and the dog. He reloads...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"The Thing\": The dog reaches Norris and Bennings, as they awkwardly wade through the snow, toward the downed copter., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The dog reaches Norris and Bennings, as they awkwardly wade through the snow, toward the downed copter. The camera pushes in with small amplitude at slow speed. The Survivor (S1) says: [English] of the crash, his eyes crazed with determination, struggles to his feet. Heedless of his companion, he double-times his way to the men and the dog. He reloads... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Revenge\n\nSCREENPLAY EXCERPT:\nLEVITAN\n You wouldn't believe it if I told you.\n In fact, uh, I'm glad\n\n\n you're here. You know, maybe I've been a little rough on ya,\n huh?\n\n\n GEORGE\n Oh.\n\n\n LEVITAN\n Why should we let petty, personal differences\n get in the way of\n\n\n business? I, uh, I want\n\n\n you to come back. You can use my bathroom anytime you want.\n\n\n\n GEORGE\n You want me to come back? Uh...\n\n\n LEVITAN\n Hey! How about a toast, huh? Everybody,\n a toast!\n\n\n GEORGE\n Rick.\n\n\n LEVITAN\n Everyone, I wanna propose a toast to\n ten great years at Rick Barr\n\n\n Properties.\n\n\n GEORGE\n Uh, Rick..\n\n\n LEVITAN\n And all the people in this room, (clears\n throat) that made that\n\n\n possible..\n\n\n GEORGE\n Rick.\n\n\n LEVITAN\n I'd also like to welcome back into the\n fold our..our little\n\n\n shrimpy friend, George Costanza who, although he didn't really\n have a\n\n\n very good year -- how you blew that McConnell deal, I'll never\n know. But,\n\n\n hey, what the hell, huh? We've"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You wouldn't believe it if I told you. In fact, uh, I'm glad\nDIALOGUE — George: Oh.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Revenge: You wouldn't believe it if I told you. In fact, uh, I'm glad, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You wouldn't believe it if I told you. In fact, uh, I'm glad The camera pushes in with small amplitude at slow speed. George (S1) says: [English] Oh. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Watchmen\n\nSCREENPLAY EXCERPT:\nINT. PYRAMID INDUSTRIES - PROCESSING PLANT - NIGHT\n\n They enter the large, dark industrial section of the\n building. They pass huge PROCESSING VATS, CONVEYOR BELTS.\n\n (CONTINUED)\n\n\n\n\n\n\n\n\n\n 95.\n\n CONTINUED:\n\n DAN\n I tapped into all the databases--tax\n records, business registries, real estate\n records--trying to get a bead on who's\n behind Pyramid but it's a never ending\n paper trail.\n Rorschach investigates ahead when Dan catches some MOVEMENT--\n --he turns his FLASHLIGHT to find a few HOMELESS PEOPLE\n squatting in the warehouse. Upon seeing Dan in.costume, the\n Homeless People run away.\n\n RORSCHACH\n Daniel.\n\n DAN\n They're harmless, Rorschach.\n\n RORSCHACH\n Daniel.\n Rorschach is looking down at something. Dan heads over and\n sees a DEAD BODY in the early stages of decomposition. He\n turns the body over--bullet between the eyes just like Moloch\n . but we recognize the corpse--it's journalist Doug Roth.\n\n DAN\n Hey, I know this man.\n\n RORSCHACH\n Yes. Doug Roth. Journalist who grilled\n Dr. Manhattan on national television.\n\n DAN\n No I didn't watch that interview.\n This was the guy who was doing a story on\n Adrian when I went to see him.\n\n RORSCHACH\n Hmm. Journalist set up Dr. Manhattan, was\n probably working to set up Veidt also.\n Dan stands t"}, {"role": "assistant", "content": "## SCENE 1 — INT. PYRAMID INDUSTRIES - PROCESSING PLANT - NIGHT\nACTION: They enter the large, dark industrial section of the building. They pass huge PROCESSING VATS, CONVEYOR BELTS.\nDIALOGUE — Dan: I tapped into all the databases--tax records, business registries, real estate records--trying to get a bead on who's behind Pyramid but it's a never ending...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Watchmen: They enter the large, dark industrial section of the building. They pass huge PROCESSING VATS, CONVEYOR BELTS., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They enter the large, dark industrial section of the building. They pass huge PROCESSING VATS, CONVEYOR BELTS. The camera pushes in with small amplitude at slow speed. Dan (S1) says: [English] I tapped into all the databases--tax records, business registries, real estate records--trying to get a bead on who's behind Pyramid but it's a never ending... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nEXT. CATHERINE'S HOUSE - DAY\n\n\tTwo cars parked outside. An old, quirky Volvo and a Land\n\tRover."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CATHERINE'S HOUSE - DAY\nACTION: Two cars parked outside. An old, quirky Volvo and a Land\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: Two cars parked outside. An old, quirky Volvo and a Land, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Two cars parked outside. An old, quirky Volvo and a Land The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Black Screen\n\nSCREENPLAY EXCERPT:\nINT. MANTAN'S DRESSING ROOM\n\nMantan sits in front of his mirror.\n\n\t\t\t\tDELACROIX\n\t\t\tJust want to say good luck.\n\n\t\t\t\tSLOAN\n\t\t\tBreak a leg.\n\nShe gives Mantan a kiss on his cheek.\n\n\t\t\t\tDELACROIX\n\t\t\tWait a minute. Hold up.\n\nThey all laugh.\n\n\t\t\t\t\t\t\t\t\t\t 57.\n\n\n\t\t\t\tDELACROIX (CONT'D)\n\t\t\tCan I kiss you too?\n\n\t\t\t\tMANTAN\n\t\t\tNaw. I'll take the zero.\n\n\t\t\t\tDELACROIX\n\t\t\tYou feel good, not nervous?\n\n\t\t\t\tMANTAN\n\t\t\tI feel fine.\n\n\t\t\t\tDELACROIX\n\t\t\tNot nervous? Relaxed?\n\n\t\t\t\tMANTAN\n\t\t\tSloan, will you take your boss out\n\t\t\tof here so I can get ready."}, {"role": "assistant", "content": "## SCENE 1 — INT. MANTAN'S DRESSING ROOM\nACTION: Mantan sits in front of his mirror. Just want to say good luck.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: Mantan sits in front of his mirror. Just want to say good luck., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Mantan sits in front of his mirror. Just want to say good luck. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: And here he comes like Don Juan!\nSetting: derived from Les Miserables\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: And here he comes like Don Juan! It is better than an opera!\nDIALOGUE — Grantaire: And here he comes like Don Juan! It is better than an opera!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Les Miserables: And here he comes like Don Juan! It is better than an opera!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. And here he comes like Don Juan! It is better than an opera! The camera pushes in with small amplitude at slow speed. Grantaire (S1) says: [English] And here he comes like Don Juan! It is better than an opera! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You could stay here forever... if you\nSetting: derived from Coraline\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. OTHER HALLWAY - CONTINUOUS\nACTION: You could stay here forever... if you Sure...we'll sing and play games, and\nDIALOGUE — Other Mother: You could stay here forever... if you want to.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Coraline: You could stay here forever... if you Sure...we'll sing and play games, and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You could stay here forever... if you Sure...we'll sing and play games, and The camera pushes in with small amplitude at slow speed. Other Mother (S1) says: [English] You could stay here forever... if you want to. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Barbie\n\nSCREENPLAY EXCERPT:\nINT. SUPREME COURT. DAY\n\nBarbie Sharon argues a case passionately in front of the\nSupreme Court, all Barbies.\n 6.\n\n\n BARBIE SHARON\n Only Barbies are Barbies, and we\n would argue that corporations have\n no \"free speech\" rights to begin\n with, so any claim on their part to\n be exercising a right is just their\n attempt to turn our democracy into\n a plutocracy!\n\nThe Gallery erupts into rapturous applause. Some Kens are\nthere for support.\n\n BARBIE SHARON\n This makes me emotional! And I'm\n expressing it. I have no difficulty\n holding both logic and feeling at\n the same time. It does not diminish\n my powers, it expands them.\n\nThe Chief Justice Barbie hits her gavel, but she can't help\nbut smile. Barbie Margot is there, always cheering on, always\nthe supporter."}, {"role": "assistant", "content": "## SCENE 1 — INT. SUPREME COURT. DAY\nACTION: Barbie Sharon argues a case passionately in front of the Supreme Court, all Barbies.\nDIALOGUE — Barbie Sharon: Only Barbies are Barbies, and we would argue that corporations have no \"free speech\" rights to begin with, so any claim on their part to be exercising a right...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Barbie: Barbie Sharon argues a case passionately in front of the Supreme Court, all Barbies., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Barbie Sharon argues a case passionately in front of the Supreme Court, all Barbies. The camera pushes in with small amplitude at slow speed. Barbie Sharon (S1) says: [English] Only Barbies are Barbies, and we would argue that corporations have no \"free speech\" rights to begin with, so any claim on their part to be exercising a right... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: He Phone Message\n\nSCREENPLAY EXCERPT:\nGEORGE\n Wait till you hear this (sees Donna).\n Whoa, ah, I'm sorry, I\n\n\n didn't, I had no idea. (Goes to leave)\n\n\n DONNA\n Wait, wait. He's in the bathroom.\n\n\n GEORGE\n I just wanted to talk to him for a minute,\n but I'll come back.\n\n\n DONNA\n You don't have to leave.\n\n\n GEORGE\n You sure?\n\n\n DONNA\n Yes.\n\n\n GEORGE\n OK.\n\n\n DONNA\n I'm Donna.\n\n\n GEORGE\n Donna. Oh, you're the one that likes\n that commercial!\n\n\n DONNA\n He told you about that.\n\n\n GEORGE\n No, he, he didn't actually tell me that,\n uh, we were talking about\n\n\n that commercial, in fact I think I brought it up because I like\n that commercial.\n\n\n No, he, he would never tell me anything like that. He never discusses\n anything.\n\n\n He's, he's like a clam. You're not gonna mention this, to him..\n (Jerry re-\n\n\n enters)\n\n\n DONNA\n (To Jerry) So you go around telling\n your friends I'm not hip\n\n\n because I like that commercial.\n\n\n JERRY\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. SEES DONNA\nACTION: Wait till you hear this (sees Donna). Whoa, ah, I'm sorry, I\nDIALOGUE — Donna: Wait, wait. He's in the bathroom.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from He Phone Message: Wait till you hear this (sees Donna). Whoa, ah, I'm sorry, I, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Wait till you hear this (sees Donna). Whoa, ah, I'm sorry, I The camera pushes in with small amplitude at slow speed. Donna (S1) says: [English] Wait, wait. He's in the bathroom. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in ConanTheAdventurer illustrating the trope 'SlaveLiberation'. Both Conan and Zula were slaves to Stygia and escaped together with a handful of other slaves.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying SlaveLiberation.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for ConanTheAdventurer, SlaveLiberation, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for SlaveLiberation. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in GreysAnatomy illustrating the trope 'TheDanza'. Long before Sandra Oh was cast in Grey's Anatomy, she won a Best Actress Genie Award (the modern day Canadian equivalent of the Oscars) for her role in Last Night... playing a character named just Sandra.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TheDanza.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for GreysAnatomy, TheDanza, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TheDanza. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Bodyguard\n\nSCREENPLAY EXCERPT:\nIt's a living.\n FRANK\n No, it isn't.\n He moves away into the next room. She hesitates a\n moment, then follows.\n NICKI\n You think my life is pretty\n sickly, don't you?\n FRANK\n No.\n NICKI\n You must. I do.\n Frank steps into a tight, dark, little window alcove,\n checks the windows.\n\n 105.\n\n FRANK\n Why don't you change it?\n NICKI\n Is it that easy?\nFrank turns from the window to find Nicki blocking his exit.\nShe moves close to him and kisses him, her hands on his\nshoulders. The kiss ends and they look at each other.\n FRANK\n You're a lovely woman.\n NICKI\n But you don't want me.\n\nFrank's look affirms it. Nicki is embarrassed, angry.\n NICKI\n I'm surprised. Thorough fellow\n like you. Why stop at one sister\n when you can fuck them both?\n FRANK\n I make my mistakes.\n NICKI\n But you didn't say no to the boss.\nShe starts to turn away from him, but his hands are on her\nnow, holding her close, refusing to let her turn away.\nTheir faces are very close. It is tight and near here.\n FRANK"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: He moves away into the next room. She hesitates a moment, then follows.\nDIALOGUE — Frank: No, it isn't. He moves away into the next room. She hesitates a moment, then follows.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Bodyguard: He moves away into the next room. She hesitates a moment, then follows., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He moves away into the next room. She hesitates a moment, then follows. The camera pushes in with small amplitude at slow speed. Frank (S1) says: [English] No, it isn't. He moves away into the next room. She hesitates a moment, then follows. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Total Recall\"\n\nSCREENPLAY EXCERPT:\nYou can have a new identity\n for the duration of the trip.\n Pick one.\n\n Quail's eyes linger on \"Intelligence Agent.\"\n\n QUAIL\n \"Intelligence Agent\"... wouldn't\n that be dangerous? I might\n attack....\n\n McCLANE"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You can have a new identity for the duration of the trip.\nDIALOGUE — Quail: \"Intelligence Agent\"... wouldn't that be dangerous? I might attack....\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Total Recall\": You can have a new identity for the duration of the trip., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You can have a new identity for the duration of the trip. The camera pushes in with small amplitude at slow speed. Quail (S1) says: [English] \"Intelligence Agent\"... wouldn't that be dangerous? I might attack.... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A Pok�mon is a pok�mon trainer? It can't be.\nSetting: derived from Untitled\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: A Pok�mon is a pok�mon trainer? It can't be. (It becomes apparent that MewTwo is telepathically\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: A Pok�mon is a pok�mon trainer? It can't be. (It becomes apparent that MewTwo is telepathically, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A Pok�mon is a pok�mon trainer? It can't be. (It becomes apparent that MewTwo is telepathically The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Jack has his briefcase on the table, looks at PAPERWORK.\nSetting: derived from F I G H T C L U B\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. KITCHEN -- MORNING\nACTION: Jack has his briefcase on the table, looks at PAPERWORK. Tyler wanders in, carries a dirty pot to the sink. Jack\nDIALOGUE — Tyler: No thanks, I quit.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from F I G H T C L U B: Jack has his briefcase on the table, looks at PAPERWORK. Tyler wanders in, carries a dirty pot to the sink. Jack, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack has his briefcase on the table, looks at PAPERWORK. Tyler wanders in, carries a dirty pot to the sink. Jack The camera pushes in with small amplitude at slow speed. Tyler (S1) says: [English] No thanks, I quit. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Bride Of Leprechaun\n\nSCREENPLAY EXCERPT:\nPut it back.\n 68.\n\n\n\n Morty shrugs and puts the money into the file cabinet. The\n safe, which has wheels, is rolled easily towards the door by\n Cody and Morty.\n\n MORTY\n We'll get a double hernia if we try\n and lift this thing in the hearse.\n See if you can find something to\n make a ramp. I'll go get the car.\n\n Morty leaves. Cody stays behind.\n\n73 EXT. PARKING LOT - FOLLOWING 73\n\n Morty comes outside and goes to the hearse. As he opens the\n door to get in, he's thrown up against the car.\n\n WIGGINS (O.S.)\n Security! Keep your hands where I\n can see them.\n\n MORTY\n Ow! You're breakin' my arm!\n\n Morty realizes that it's WIGGINS, a security cop. He forces\n Morty to lean over the hood of the hearse and handcuffs one\n of Morty's hands.\n\n WIGGINS (CONT'D)\n You tripped the silent alarm,\n buddy. I'm taking you down.\n\n MORTY\n Taking me down where?! You're a\n friggin' security cop!\n\n WIGGINS\n That's security officer, pal. You\n the only perp here?\n\n MORTY\n Perp? What are you talking about?\n\n The security cop handcuffs Morty to the door of the hearse.\n\n WIGGINS (CONT'D)\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. O.S.\nACTION: Morty shrugs and puts the money into the file cabinet. The safe, which has wheels, is rolled easily towards the door by\nDIALOGUE — Morty: We'll get a double hernia if we try and lift this thing in the hearse. See if you can find something to make a ramp. I'll go get the car.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bride Of Leprechaun: Morty shrugs and puts the money into the file cabinet. The safe, which has wheels, is rolled easily towards the door by, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Morty shrugs and puts the money into the file cabinet. The safe, which has wheels, is rolled easily towards the door by The camera pushes in with small amplitude at slow speed. Morty (S1) says: [English] We'll get a double hernia if we try and lift this thing in the hearse. See if you can find something to make a ramp. I'll go get the car. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Oh, this isn't your standard boardroom; this is Mooby Corp., home of Mooby,\nSetting: derived from Dogma\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. MOOBY CORP. BOARDROOM - DAY\nACTION: Oh, this isn't your standard boardroom; this is Mooby Corp., home of Mooby, the Golden Calf - which can only be described as a bovine variation on\nDIALOGUE — Whitland: Good morning, shoppers.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dogma: Oh, this isn't your standard boardroom; this is Mooby Corp., home of Mooby, the Golden Calf - which can only be described as a bovine variation on, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Oh, this isn't your standard boardroom; this is Mooby Corp., home of Mooby, the Golden Calf - which can only be described as a bovine variation on The camera pushes in with small amplitude at slow speed. Whitland (S1) says: [English] Good morning, shoppers. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Airplane\n\nSCREENPLAY EXCERPT:\nEXT. TOWER - NIGHT\n\n GROUND CONTROL (v.o.)\n Air Poland, you are cleared for takeoff."}, {"role": "assistant", "content": "## SCENE 1 — EXT. TOWER - NIGHT\nACTION: Air Poland, you are cleared for takeoff.\nDIALOGUE — Ground Control: Air Poland, you are cleared for takeoff.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Airplane: Air Poland, you are cleared for takeoff., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Air Poland, you are cleared for takeoff. The camera pushes in with small amplitude at slow speed. Ground Control (S1) says: [English] Air Poland, you are cleared for takeoff. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Sponge\n\nSCREENPLAY EXCERPT:\nSUSAN\n Give it to me. (She rips it open.) Come\n on. Come on!\n\n\n GEORGE\n It's too late.\n\n\n\n\n Kramer surrounded by Cedric, Bob, and the other walkers.\n\n\n BOB\n So! What's it going to be? Are you going\n to wear the ribbon?\n\n\n KRAMER\n No! Never.\n\n\n BOB\n But I am wearing the ribbon. He is wearing\n the ribbon. We are all wearing the ribbon!\n So why aren't you going to wear the\n ribbon!?\n\n\n KRAMER\n This is America! I don't have to wear\n anything I don't want to wear!\n\n\n CEDRIC\n What are we gonna do with him?\n\n\n BOB\n I guess we are just going to have to\n teach him to wear the ribbon!\n\n\n (Kramer tries to climb up a fire escape, but the mob grabs him\n and pulls him back down. Kramer screams.)\n\n\n\n\n Jerry and George waiting for Kramer at the finish line.\n\n\n JERRY\n It completely turned her off.\n\n\n GEORGE\n Well, I can see that. What do you have\n to do that for? Who cares about your\n pants size?\n\n\n JERRY\n I don't wanna be a 32.\n\n\n GEORGE\n I'd k"}, {"role": "assistant", "content": "## SCENE 1 — INT. SHE RIPS IT OPEN.\nACTION: Give it to me. (She rips it open.) Come Kramer surrounded by Cedric, Bob, and the other walkers.\nDIALOGUE — George: It's too late.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Sponge: Give it to me. (She rips it open.) Come Kramer surrounded by Cedric, Bob, and the other walkers., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Give it to me. (She rips it open.) Come Kramer surrounded by Cedric, Bob, and the other walkers. The camera pushes in with small amplitude at slow speed. George (S1) says: [English] It's too late. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Needy is fully dressed and garishly made up. Toni holds a\nSetting: derived from Jennifer'S Body\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. NEEDY'S HOUSE (LIVING ROOM) - EVENING\nACTION: Needy is fully dressed and garishly made up. Toni holds a digital camera at arm's length and takes a shot of the\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jennifer'S Body: Needy is fully dressed and garishly made up. Toni holds a digital camera at arm's length and takes a shot of the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Needy is fully dressed and garishly made up. Toni holds a digital camera at arm's length and takes a shot of the The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Beemer shoves Lenny into the alley and shuts the door\nSetting: derived from S T R A N G E D A Y S\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ALLEY/RETINAL FETISH - NIGHT\nACTION: Beemer shoves Lenny into the alley and shuts the door I recognize you. You're Wade\nDIALOGUE — Lenny: I recognize you. You're Wade Beemer. Running back for the Rams, am I right?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from S T R A N G E D A Y S: Beemer shoves Lenny into the alley and shuts the door I recognize you. You're Wade, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Beemer shoves Lenny into the alley and shuts the door I recognize you. You're Wade The camera pushes in with small amplitude at slow speed. Lenny (S1) says: [English] I recognize you. You're Wade Beemer. Running back for the Rams, am I right? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: COURT SKETCH - CU of Grace staring intently. CAMERA widens to\nSetting: derived from \"Natural Born Killers\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. COURTROOM - DAY\nACTION: COURT SKETCH - CU of Grace staring intently. CAMERA widens to include Bisbing standing before the bench as Mickey sits with\nDIALOGUE — Bisbing: No further questions, your honor.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Natural Born Killers\": COURT SKETCH - CU of Grace staring intently. CAMERA widens to include Bisbing standing before the bench as Mickey sits with, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. COURT SKETCH - CU of Grace staring intently. CAMERA widens to include Bisbing standing before the bench as Mickey sits with The camera pushes in with small amplitude at slow speed. Bisbing (S1) says: [English] No further questions, your honor. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in HowIMetYourMother illustrating the trope 'HypocriteHasAPoint'. How I Met Your Mother: In \"The Best Burger In New York\" Marshall reveals that he ended up taking a corporate job at Goliath National Bank, even though his dream job is to become an environmental lawyer and he has already experienced the negatives of working in a corporation. However, at that point, he had failed so many job interviews that he became too depressed to put on pants. At the very least...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying HypocriteHasAPoint.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for HowIMetYourMother, HypocriteHasAPoint, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for HypocriteHasAPoint. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The elevator descends. Then a tone CHIMES, signaling the end of the\nSetting: derived from Blade\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. THE BLACK PEARL, ELEVATOR - NIGHT\nACTION: The elevator descends. Then a tone CHIMES, signaling the end of the ride. The doors hiss open --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade: The elevator descends. Then a tone CHIMES, signaling the end of the ride. The doors hiss open --, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The elevator descends. Then a tone CHIMES, signaling the end of the ride. The doors hiss open -- The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Mantan sits in front of his mirror.\nSetting: derived from Black Screen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. MANTAN'S DRESSING ROOM\nACTION: Mantan sits in front of his mirror. Just want to say good luck.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: Mantan sits in front of his mirror. Just want to say good luck., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Mantan sits in front of his mirror. Just want to say good luck. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in ColdCase illustrating the trope 'GoodGirlsAvoidAbortion'. Averted in Cold Case, which deals with abortion several times.\n One first season episode centers on an underground abortion service during the 1960s. Both of the (highly sympathetic) victims worked for the service and two other sympathetic witnesses in the case also had.\n One third season episode involved a high school couple who decide not to get an abortion after seeing photos given to them by a...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying GoodGirlsAvoidAbortion.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for ColdCase, GoodGirlsAvoidAbortion, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for GoodGirlsAvoidAbortion. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The INSPECTOR is in his fifties. The deep creases in\nSetting: derived from Blade Runner\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. INSPECTOR BRYANT'S OFFICE - NIGHT 11\nACTION: The INSPECTOR is in his fifties. The deep creases in his face, the broken capillaries in his nose say\nDIALOGUE — Deckard: They escaped from the colonies two weeks ago. Killed twenty- three people and jumped a shuttle. An aerial patrol found the ship in the desert. No crew.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade Runner: The INSPECTOR is in his fifties. The deep creases in his face, the broken capillaries in his nose say, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The INSPECTOR is in his fifties. The deep creases in his face, the broken capillaries in his nose say The camera pushes in with small amplitude at slow speed. Deckard (S1) says: [English] They escaped from the colonies two weeks ago. Killed twenty- three people and jumped a shuttle. An aerial patrol found the ship in the desert. No crew. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Sponge\n\nSCREENPLAY EXCERPT:\nElaine and Billy in bed the next morning.\n\n\n ELAINE\n Good morning.\n\n\n BILLY\n How'd you sleep?\n\n\n ELAINE\n Great. You?\n\n\n BILLY\n Fine, fine. Everything o.k.?\n\n\n ELAINE\n Yep.\n\n\n BILLY\n No regrets?\n\n\n ELAINE\n Nope. (Billy leans in to kiss her.)\n What are you doing?\n\n\n BILLY\n What do you mean?\n\n\n ELAINE\n Oh...I don't think so.\n\n\n BILLY\n Why not? I thought you said everything\n was fine.\n\n\n ELAINE\n I wish I could help you, but I can't\n afford two of 'em. (Pats Billy on the\n shoulder and gets out of bed.)\n\n\n\n\n THE END"}, {"role": "assistant", "content": "## SCENE 1 — INT. BILLY LEANS IN TO KISS HER.\nACTION: Elaine and Billy in bed the next morning. Fine, fine. Everything o.k.?\nDIALOGUE — Elaine: Good morning.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Sponge: Elaine and Billy in bed the next morning. Fine, fine. Everything o.k.?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Elaine and Billy in bed the next morning. Fine, fine. Everything o.k.? The camera pushes in with small amplitude at slow speed. Elaine (S1) says: [English] Good morning. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: She PULLS Coraline down the hallway to the mirror at the\nSetting: derived from Coraline\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. OTHER HOUSE HALLWAY - CONTINUOUS\nACTION: She PULLS Coraline down the hallway to the mirror at the end. Coraline flails at the Tall Other Mother with her\nDIALOGUE — Coraline: Ow, that hurts!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Coraline: She PULLS Coraline down the hallway to the mirror at the end. Coraline flails at the Tall Other Mother with her, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. She PULLS Coraline down the hallway to the mirror at the end. Coraline flails at the Tall Other Mother with her The camera pushes in with small amplitude at slow speed. Coraline (S1) says: [English] Ow, that hurts! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: All the South Park men are in the rank and file marching down\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. SOUTH PARK AVENUE\nACTION: All the South Park men are in the rank and file marching down I don't know but I've been told-\nDIALOGUE — Officer: I don't know but I've been told-\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: All the South Park men are in the rank and file marching down I don't know but I've been told-, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. All the South Park men are in the rank and file marching down I don't know but I've been told- The camera pushes in with small amplitude at slow speed. Officer (S1) says: [English] I don't know but I've been told- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Seven\n\nSCREENPLAY EXCERPT:\nINT. MILLS' APARTMENT, LIVING ROOM -- NIGHT\n\n Mills sits down at his desk. He starts looking through police\n paperwork. The RADIO in the other room goes OFF in mid song.\n\n\n INSERT -- TITLE CARD\n\n TUESDAY"}, {"role": "assistant", "content": "## SCENE 1 — INT. MILLS' APARTMENT, LIVING ROOM -- NIGHT\nACTION: Mills sits down at his desk. He starts looking through police paperwork. The RADIO in the other room goes OFF in mid song.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Seven: Mills sits down at his desk. He starts looking through police paperwork. The RADIO in the other room goes OFF in mid song., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Mills sits down at his desk. He starts looking through police paperwork. The RADIO in the other room goes OFF in mid song. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I'm sorry. Did I bring what?\nSetting: derived from Joker\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I'm sorry. Did I bring what? (impatient; she doesn't\nDIALOGUE — Social Worker: (impatient; she doesn't have time for this) Arthur, last time I asked you to bring your journal with you. For these appointments. Do you have it?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Joker: I'm sorry. Did I bring what? (impatient; she doesn't, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm sorry. Did I bring what? (impatient; she doesn't The camera pushes in with small amplitude at slow speed. Social Worker (S1) says: [English] (impatient; she doesn't have time for this) Arthur, last time I asked you to bring your journal with you. For these appointments. Do you have it? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nEXT. SOUTH PARK AVENUE\n\n All the South Park men are in the rank and file marching down\n the street.\n\n OFFICER\n (Singing)\n I don't know but I've been told-\n\n SOLDIERS\n I DON'T KNOW BUT I'VE BEEN TOLD!\n\n They pass Kyle's house."}, {"role": "assistant", "content": "## SCENE 1 — EXT. SOUTH PARK AVENUE\nACTION: All the South Park men are in the rank and file marching down I don't know but I've been told-\nDIALOGUE — Officer: I don't know but I've been told-\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: All the South Park men are in the rank and file marching down I don't know but I've been told-, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. All the South Park men are in the rank and file marching down I don't know but I've been told- The camera pushes in with small amplitude at slow speed. Officer (S1) says: [English] I don't know but I've been told- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: What's that? A celebrity cooking show for\nSetting: derived from Untitled\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: What's that? A celebrity cooking show for No. It's my favourite episode of Videodrome.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: What's that? A celebrity cooking show for No. It's my favourite episode of Videodrome., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What's that? A celebrity cooking show for No. It's my favourite episode of Videodrome. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Int: Bedroom. Early-Morning Hours\n\nSCREENPLAY EXCERPT:\nCute cat. What's its name?\n\n\t\t\t\tRANDAL\n\t\tAnnoying Customer.\n\n\tThe CUSTOMER lets it sink in, and then leaves in a huff.\n\tDANTE puts up cigarettes.\n\n\t\t\t\t\t\t\t\t\t88.\n\n\n\t\t\t\tDANTE\n\t\tCan you imagine being halfway\n\t\tdecent to the customers at least\n\t\tsome of the time?\n\n\t\t\t\tRANDAL\n\t\tLet me borrow your car.\n\n\t\t\t\tDANTE"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Cute cat. What's its name? The CUSTOMER lets it sink in, and then leaves in a huff.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int: Bedroom. Early-Morning Hours: Cute cat. What's its name? The CUSTOMER lets it sink in, and then leaves in a huff., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Cute cat. What's its name? The CUSTOMER lets it sink in, and then leaves in a huff. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Int: Bedroom. Early-Morning Hours\n\nSCREENPLAY EXCERPT:\nThanks.\n\n\tThe door is heard opening and closing-a customer leaving.\n\n\t\t\t\tVERONICA\n\t\tHow much money did you leave up\n\t\tthere?\n\n\t\t\t\tDANTE\n\t\tLike three dollars in mixed change\n\t\tand a couple of singles. People\n\t\tonly get the paper of coffee this\n\t\ttime of morning.\n\n\t\t\t\tVERONICA\n\t\tYou're trusting.\n\n\t\t\t\tDANTE\n\t\tWhy do you say that?\n\n\t\t\t\tVERONICA\n\t\tHow do you know they're taking the\n\t\tright amount of change? Or even\n\t\tpaying for what they take?\n\n\t\t\t\tDANTE\n\t\tTheoretically, people see money on\n\t\tthe counter and nobody around, they\n\t\tthink they're being watched.\n\n\t\t\t\tVERONICA\n\t\tHonesty through paranoia. Why do\n\t\tyou smell like shoe polish?\n\n\t\t\t\t\t\t\t\t\t18.\n\n\n\t\t\t\tDANTE\n\t\tI had to use shoe polish to make\n\t\tthat sign. The smell won't come off.\n\n\t\t\t\tVERONICA\n\t\tDo you think anyone can see us down\n\t\there?\n\n\t\t\t\tDANTE\n\t\tWhy? You wanna have sex or something?\n\n\t\t\t\tVERONICA"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The door is heard opening and closing-a customer leaving. How much money did you leave up\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int: Bedroom. Early-Morning Hours: The door is heard opening and closing-a customer leaving. How much money did you leave up, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The door is heard opening and closing-a customer leaving. How much money did you leave up The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Keep you damn filthy bones outta\nSetting: derived from The Army Of Darkness\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Keep you damn filthy bones outta SNAP! He breaks the boney arm in two and runs away from the sea\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Army Of Darkness: Keep you damn filthy bones outta SNAP! He breaks the boney arm in two and runs away from the sea, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Keep you damn filthy bones outta SNAP! He breaks the boney arm in two and runs away from the sea The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Butt-Head sits. Beavis pauses, still standing.\nSetting: derived from \"Beavis And Butt-Head Do America\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TOUR BUS - DAY\nACTION: Butt-Head sits. Beavis pauses, still standing. Hey wait a minute. What's going on? Why\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Beavis And Butt-Head Do America\": Butt-Head sits. Beavis pauses, still standing. Hey wait a minute. What's going on? Why, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Butt-Head sits. Beavis pauses, still standing. Hey wait a minute. What's going on? Why The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ------------------------------------------------------------------------\nSetting: derived from \"Hunter\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: ------------------------------------------------------------------------ 53-A CONTINUED: 53-A\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hunter\": ------------------------------------------------------------------------ 53-A CONTINUED: 53-A, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ------------------------------------------------------------------------ 53-A CONTINUED: 53-A The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You know I'm not on the manifest? All of you\nSetting: derived from Django Unchained\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You know I'm not on the manifest? All of you know I'm not suppose to be on this trip.\nDIALOGUE — Roy: What's your deal? You tell us who they are and we let ya go?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Django Unchained: You know I'm not on the manifest? All of you know I'm not suppose to be on this trip., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You know I'm not on the manifest? All of you know I'm not suppose to be on this trip. The camera pushes in with small amplitude at slow speed. Roy (S1) says: [English] What's your deal? You tell us who they are and we let ya go? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: They sit on the edge of the bed - Sachiko holds Elizabeth's\nSetting: derived from Dark City\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BEDROOM\nACTION: They sit on the edge of the bed - Sachiko holds Elizabeth's I'm sorry... This is embarrassing -\nDIALOGUE — Elizabeth: I'm sorry... This is embarrassing - my husband doesn't know who I am...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dark City: They sit on the edge of the bed - Sachiko holds Elizabeth's I'm sorry... This is embarrassing -, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They sit on the edge of the bed - Sachiko holds Elizabeth's I'm sorry... This is embarrassing - The camera pushes in with small amplitude at slow speed. Elizabeth (S1) says: [English] I'm sorry... This is embarrassing - my husband doesn't know who I am... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Where is this place? Why am I here?\nSetting: derived from Untitled\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Where is this place? Why am I here? To make you take care of me, I took you from the Pok�mon Center.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: Where is this place? Why am I here? To make you take care of me, I took you from the Pok�mon Center., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Where is this place? Why am I here? To make you take care of me, I took you from the Pok�mon Center. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: S T R A N G E D A Y S\n\nSCREENPLAY EXCERPT:\nEXT. ALLEY/RETINAL FETISH - NIGHT\n\n Beemer shoves Lenny into the alley and shuts the door\n behind him.\n\n LENNY\n I recognize you. You're Wade\n Beemer. Running back for the Rams,\n am I right?\n\n Beemer, who was about to go to work on Lenny, pauses.\n\n BEEMER\n Yeah, that's right.\n\n LENNY\n I saw you play, man. You were good.\n Like a fucking freight train I\n remember saying. So what happened,\n injuries or what?\n\n BEEMER\n Bullshit politics.\n\n LENNY\n It's always politics. Like this\n thing we're in here, he's paying you\n to tune me up, right? But I could\n pay you more not to. See what I\n mean? I could write you a check\n right now--\n\n BEEMER\n Come on, let's go, I got to get\n back.\n\n LENNY\n (pulling money out\n of his pocket)\n Okay cash! Logical. Here's\n everything I have on me, what do you\n say? How about a Rolex?\n\n BEEMER\n (barely a glance at\n it)\n I already got a real one. Come on,\n it won't be too bad. It's not\n personal.\n\n LENNY\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. ALLEY/RETINAL FETISH - NIGHT\nACTION: Beemer shoves Lenny into the alley and shuts the door I recognize you. You're Wade\nDIALOGUE — Lenny: I recognize you. You're Wade Beemer. Running back for the Rams, am I right?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from S T R A N G E D A Y S: Beemer shoves Lenny into the alley and shuts the door I recognize you. You're Wade, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Beemer shoves Lenny into the alley and shuts the door I recognize you. You're Wade The camera pushes in with small amplitude at slow speed. Lenny (S1) says: [English] I recognize you. You're Wade Beemer. Running back for the Rams, am I right? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Wait till you hear this (sees Donna).\nSetting: derived from He Phone Message\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SEES DONNA\nACTION: Wait till you hear this (sees Donna). Whoa, ah, I'm sorry, I\nDIALOGUE — Donna: Wait, wait. He's in the bathroom.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from He Phone Message: Wait till you hear this (sees Donna). Whoa, ah, I'm sorry, I, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Wait till you hear this (sees Donna). Whoa, ah, I'm sorry, I The camera pushes in with small amplitude at slow speed. Donna (S1) says: [English] Wait, wait. He's in the bathroom. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A MALE FORENSIC uses a magnifying glass to study a very clear\nSetting: derived from Seven\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LUXURY APARTMENT/CRIME SCENE, LIVING ROOM -- NIGHT\nACTION: A MALE FORENSIC uses a magnifying glass to study a very clear fingerprint in black powder on the wall.\nDIALOGUE — Forensic: Oh, boy...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Seven: A MALE FORENSIC uses a magnifying glass to study a very clear fingerprint in black powder on the wall., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A MALE FORENSIC uses a magnifying glass to study a very clear fingerprint in black powder on the wall. The camera pushes in with small amplitude at slow speed. Forensic (S1) says: [English] Oh, boy... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: In the corner, Hammond lifts his head and Ellie gently helps him to a\nSetting: derived from Jurassic Park\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: In the corner, Hammond lifts his head and Ellie gently helps him to a Do you mean those raptor\nDIALOGUE — Hammond: Do you mean those raptor s could be loose?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jurassic Park: In the corner, Hammond lifts his head and Ellie gently helps him to a Do you mean those raptor, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. In the corner, Hammond lifts his head and Ellie gently helps him to a Do you mean those raptor The camera pushes in with small amplitude at slow speed. Hammond (S1) says: [English] Do you mean those raptor s could be loose? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages":[{"role":"system","content":"You write MiniMax-H3 multi-beat scripts as sequential SCENE blocks (FL2VA)."},{"role":"user","content":"Premise: Two scavengers argue about a locked vault door, then the door opens.\nTone: tense\nApproximate pages: 1"},{"role":"assistant","content":"## SCENE 1 — INT. UNDERGROUND VAULT CORRIDOR - DIM\nACTION: Two scavengers face a sealed vault door; the camera holds as tension rises, ending on a tighter framing of their hands on the wheel lock.\nDIALOGUE — REX: If this seals us in, we're done.\nSHOT: medium two-shot, static shot then slight push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: dim underground corridor, heavy vault door, two scavengers in worn gear, practical work lights, cinematic grit\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium two-shot of scavengers Rex and Mara before a heavy vault door matches Picture 1. The camera holds a static shot, then pushes in with small amplitude at slow speed on their gloved hands on the wheel lock. Rex (S1) says under his breath: [English] If this seals us in, we're done. Mara (S2) keeps silent, jaw set, as the frame tightens to Picture 2.\n\noverall_soundscape: Low ventilation drone, distant metal creaks, glove leather flexing on cold steel.\n\nnon_diegetic_music: A low electronic pulse at a slow tempo that holds without resolving.\nLORA: none\nAUDIO: none\nDURATION: 5\n\n## SCENE 2 — INT. UNDERGROUND VAULT CORRIDOR - DIM\nACTION: The wheel lock turns and the vault door cracks open, light spilling into the corridor as the camera pushes through the gap.\nSHOT: medium shot, push in with large amplitude at moderate speed\nSTORYBOARD_PROMPT: vault door cracking open, hard white light spill, two silhouetted scavengers, dust motes, cinematic\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, continuing from the tight wheel-lock framing of Picture 1. Mara (S2) heaves the wheel; metal groans as the door cracks and hard white light spills into the corridor. The camera pushes in with large amplitude toward the opening until Picture 2's silhouette-and-light composition is reached. Rex (S1) exhales: [English] After you.\n\noverall_soundscape: Wheel lock clanks, door hinges scream, air whooshes as pressure equalizes, light hum rises.\n\nnon_diegetic_music: The electronic pulse cuts to a single sustained high tone as the door opens.\nLORA: none\nAUDIO: heavy door, light bloom sfx\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Futurama\n\nSCREENPLAY EXCERPT:\nFUTURAMA\n\n Episode 101\n\n \"SPACE PILOT 3000\"\n\n By\n\n David X. Cohen & Matt Groening\n\n Transcribed by Dave, The Neutral Planet\n\n\n\n [Over the caption December 31st 1999 a crude spaceship flies\n through space, cruising over and under planets and a man speaks.]\n\n\n\n MAN\n (voice-over) Space. It seems to go on\n and on forever. But then you get to\n the end and the gorilla starts throwing\n barrels at you.\n\n\n [A planet opens up and a huge gorilla starts throwing barrels\n at the spaceship. It dodges a few but one hits it and it explodes.\n The gorilla thumps its chest and \"Game Over\" flashes on the screen.\n The spaceship and gorilla isn't real and the man, called Fry,\n was playing an arcade game called \"Monkey Fracas Jr\". He is in\n his mid-20s, wears a red jacket and has orange hair with two\n distinct forks at the front. There is a little kid standing next\n to him. The game is against the wall in a pizzeria called Panucci's\n Pizza.]\n\n\n FRY\n And that's how you play the game!\n\n\n KID\n You stink, loser!\n\n\n [Mr Panucc"}, {"role": "assistant", "content": "## SCENE 1 — INT. VOICE-OVER\nACTION: David X. Cohen & Matt Groening Transcribed by Dave, The Neutral Planet\nDIALOGUE — Man: (voice-over) Space. It seems to go on and on forever. But then you get to the end and the gorilla starts throwing barrels at you.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Futurama: David X. Cohen & Matt Groening Transcribed by Dave, The Neutral Planet, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. David X. Cohen & Matt Groening Transcribed by Dave, The Neutral Planet The camera pushes in with small amplitude at slow speed. Man (S1) says: [English] (voice-over) Space. It seems to go on and on forever. But then you get to the end and the gorilla starts throwing barrels at you. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: GOLLUM suddenly LIFTS HIS HANDS to his face ... and BITES\nSetting: derived from The Lord Of The Rings: The Return Of The King\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CRACK OF DOOM\nACTION: GOLLUM suddenly LIFTS HIS HANDS to his face ... and BITES ANGLE ON: FRODO MATERIALIZES as he DROPS TO HIS KNEES,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Return Of The King: GOLLUM suddenly LIFTS HIS HANDS to his face ... and BITES ANGLE ON: FRODO MATERIALIZES as he DROPS TO HIS KNEES,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GOLLUM suddenly LIFTS HIS HANDS to his face ... and BITES ANGLE ON: FRODO MATERIALIZES as he DROPS TO HIS KNEES, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Bottle Rocket\n\nSCREENPLAY EXCERPT:\nEXT. CAR. DAY\n\nLate afternoon. They're driving on the interstate. Anthony's\ngot a copy of Runner's World magazine.\n\n ANTHONY\n Here's mine right here. See?\n\nAnthony lifts up his foot so Dignan can compare his shoe\nwith the picture.\n\n DIGNAN\n See if mine are in there.\n\n ANTHONY\n (looks at Dignan's shoes)\n Dignan, those aren't running shoes.\n\n DIGNAN\n Yes, they are.\n\n ANTHONY\n Look at the treads on those.\n\n DIGNAN\n What about them?\n\n ANTHONY\n They obviously weren't designed for\n racing.\n\n DIGNAN\n Well, those treads stink. You'd\n blow a knee out racing on those.\n\n BOB\n Will you guys shut up? God. It's\n like having two little kids in the\n car.\n\n ANTHONY\n OK, Dad.\n\n DIGNAN\n Really. OK, Dad. But seriously,\n Anthony. These are fast shoes.\n\n ANTHONY\n You've never had a pair of fast\n shoes in your life, Dignan.\n (to Bob)\n In fifth grade Dignan used to wear\n cowboy boots for P.E.\n\n DIGNAN\n That's real cool, Anthony. Yeah, I\n wore boots. My parents wouldn't buy\n me any $200 running shoes like\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. CAR. DAY\nACTION: Late afternoon. They're driving on the interstate. Anthony's got a copy of Runner's World magazine.\nDIALOGUE — Anthony: Here's mine right here. See?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bottle Rocket: Late afternoon. They're driving on the interstate. Anthony's got a copy of Runner's World magazine., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Late afternoon. They're driving on the interstate. Anthony's got a copy of Runner's World magazine. The camera pushes in with small amplitude at slow speed. Anthony (S1) says: [English] Here's mine right here. See? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nEXT. MOVIE THEATRE - DAY\n\n The boys walk out of the theatre with glazed eyes and wide\n smiles.\n\n KYLE\n Dude, that movie was fucking sweet!\n\n CARTMAN\n You bet your fucking ass it was!\n\n STAN\n Fuck, dude, I wanna be just like Terrance\n and Phillip!\n\n TICKET GUY\n Hey wait a minute... Where's your\n guardian?\n\n STAN\n Huh?\n\n TICKET GUY\n I knew it! You PAID a homeless guy to get\n you in, didn't you!\n\n The boys think a second.\n\n CARTMAN\n Suck my balls.\n\n KYLE\n Yeah,\n (Singing)\n Shut your fucking face,\n Unclefucka!!\n\n The boys walk away, merrily. The ticket guy is in shock.\n\n TICKET GUY\n Oh oh, I'm in trouble.\n\n BOYS\n (Singing, fading off)\n You're an asslicking, ball sucking\n Unclefucka!!"}, {"role": "assistant", "content": "## SCENE 1 — EXT. MOVIE THEATRE - DAY\nACTION: The boys walk out of the theatre with glazed eyes and wide Dude, that movie was fucking sweet!\nDIALOGUE — Kyle: Dude, that movie was fucking sweet!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: The boys walk out of the theatre with glazed eyes and wide Dude, that movie was fucking sweet!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The boys walk out of the theatre with glazed eyes and wide Dude, that movie was fucking sweet! The camera pushes in with small amplitude at slow speed. Kyle (S1) says: [English] Dude, that movie was fucking sweet! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Basic Instinct\n\nSCREENPLAY EXCERPT:\nINT. THE STAIRWAY\n\n Nick tears desperately up the stairway -- he hits the fourth\n floor door. It explodes open."}, {"role": "assistant", "content": "## SCENE 1 — INT. THE STAIRWAY\nACTION: Nick tears desperately up the stairway -- he hits the fourth floor door. It explodes open.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Basic Instinct: Nick tears desperately up the stairway -- he hits the fourth floor door. It explodes open., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Nick tears desperately up the stairway -- he hits the fourth floor door. It explodes open. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Black Screen\n\nSCREENPLAY EXCERPT:\nEXT. TROLLSHAW FOREST -- NIGHT\n\nSPEEDING POV: through the forest from the back of the White\nHorse.\n\nANGLE ON: FRODO, BOUNCING IN THE SADDLE,\n\nhe lifts his head weakly. SURREAL SLOW MOTION POV: THE HORSES\nHEAD BOBBING...trees sliding by...moonlight flickers through\nthe trees.\n\n CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. TROLLSHAW FOREST -- NIGHT\nACTION: SPEEDING POV: through the forest from the back of the White ANGLE ON: FRODO, BOUNCING IN THE SADDLE,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: SPEEDING POV: through the forest from the back of the White ANGLE ON: FRODO, BOUNCING IN THE SADDLE,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. SPEEDING POV: through the forest from the back of the White ANGLE ON: FRODO, BOUNCING IN THE SADDLE, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Blade\n\nSCREENPLAY EXCERPT:\nINT. THE VAMPIRE ARCHIVES, HALLWAY - NIGHT\n\n\tBlade and Karen enter the narrow, maze-like \"stacks\" of the vampire\n\tarchives, weaving their way through shelves of climate-controlled,\n\tdigitized records.\n\n\t\t\t\tKAREN\n\t\tWhat is this place?\n\n\t\t\t\tBLADE\n\t\tSome kind of archive --\n\n\t\t\t\tKAREN\n\t\tIsn't this all a little high-tech? I\n\t\tthought vampires were more into\n\t\tcobwebs and coffins.\n\n\t\t\t\tBLADE\n\t\tYou've been watching too much TV.\n\t\tThey've got their claws sunk into\n\t\teverything -- finance, real estate,\n\t\tpolitics. Probably own half of\n\t\tDowntown.\n\n\tBlade pulls a CD-ROM from the shelves, its spine labeled in\n\tindecipherable vampire glyphs.\n\n\t\t\t\tBLADE\n\t\tThis must be where they keep their\n\t\trecords --\n\n\tHe pauses, hearing a DISTANT WHISPER. Blade signals quiet, silently\n\tleading Karen through the stacks towards --\n\n\tPEARL'S LAIR\n\n\tA vestibule of sorts, leading to a larger bed chamber constructed of\n\trice paper shoji screens and tatami mats. Candles glow within,\n\tilluminating a MONSTROUSLY OBESE SHADOW beyond the translucent rice\n\tpaper walls.\n\n\tThe shadow speaks in a tremulous, bird-like voice -- the secret\n\ttongue. As Blade draws his sword, the massive shadow freezes --\n\n\t\t\t\tPEARL'S VOICE\n\t\t\t(frightened)\n\t\tLorca, is that you?\n\n\tBlade slides open the nearest shoji screen --"}, {"role": "assistant", "content": "## SCENE 1 — INT. THE VAMPIRE ARCHIVES, HALLWAY - NIGHT\nACTION: Blade and Karen enter the narrow, maze-like \"stacks\" of the vampire archives, weaving their way through shelves of climate-controlled,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade: Blade and Karen enter the narrow, maze-like \"stacks\" of the vampire archives, weaving their way through shelves of climate-controlled,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Blade and Karen enter the narrow, maze-like \"stacks\" of the vampire archives, weaving their way through shelves of climate-controlled, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Barbie Margot's having a send off party with all of the\nSetting: derived from Barbie\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BARBIE DREAMHOUSE/CUL-DE-SAC. MORNING\nACTION: Barbie Margot's having a send off party with all of the Barbies - there is a big banner that reads:\nDIALOGUE — Ken Simu: I guess she's going without you. 26.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Barbie: Barbie Margot's having a send off party with all of the Barbies - there is a big banner that reads:, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Barbie Margot's having a send off party with all of the Barbies - there is a big banner that reads: The camera pushes in with small amplitude at slow speed. Ken Simu (S1) says: [English] I guess she's going without you. 26. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Suicide\n\nSCREENPLAY EXCERPT:\nJERRY\n That's not too bad. It's not like a\n Sunny von Bulow comma. The doctor said\n he should snap out of it anytime.\n\n\n\n GINA\n You know why he did this? Because I\n told him it was over. I did not want\n to see him anymore.\n\n\n JERRY\n Really? It's over?\n\n\n GINA\n I could not stand it another minute.\n Yesterday he turned over a man's hot\n dog stand because he thought the man\n was looking at me. And then after he\n saw you in the hall. Ach, he was crazy\n with jealousy.\n\n\n JERRY\n Oh boy, did he say anything about me?\n\n\n\n GINA\n He does not like you. And all indications\n are he does not like Drake's Coffee\n Cake.\n\n\n JERRY\n He said that?\n\n\n GINA\n He was screaming about it all night.\n How it's too sweet and it falls apart\n when you eat it.\n\n\n JERRY\n I'm sorry if I caused any trouble. I\n was just being friendly.\n\n\n GINA\n I wasn't.\n\n\n JERRY\n You weren't?\n\n\n GINA\n No, I have thought about you many times.\n Have you "}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: That's not too bad. It's not like a Sunny von Bulow comma. The doctor said\nDIALOGUE — Gina: You know why he did this? Because I told him it was over. I did not want to see him anymore.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Suicide: That's not too bad. It's not like a Sunny von Bulow comma. The doctor said, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. That's not too bad. It's not like a Sunny von Bulow comma. The doctor said The camera pushes in with small amplitude at slow speed. Gina (S1) says: [English] You know why he did this? Because I told him it was over. I did not want to see him anymore. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Jacob'S Ladder\"\n\nSCREENPLAY EXCERPT:\nEXT. BAR - DAY\n\nTHE YOUNG MAN from the bar grabs JACOB under the arms and drags him\ndown the sidewalk.\n\n\t\t\t\tYOUNG MAN\n\t\tJust hold on.\n\n\t\t\t\tJACOB\n\t\tWhere am I? Who are you?\n\nThe YOUNG MAN yanks JACOB around the corner just as another explosion\nconsumes the car. The air is filled with flames and flying debris. The\nYOUNG MAN pulls JACOB into the bar.\n\n\t\t\t\tYOUNG MAN\n\t\tJust lie still. You're okay. You're\n\t\tnot hurt.\n\nThe CUSTOMERS are in a state of bedlam. Part of the wall has blown\napart and bricks and glass are everywhere. The cross from around PAUL's\nneck is buried in the debris. Sirens are heard in the distance. A BLACK\nCAR speeds off down the street. JACOB looks for the YOUNG MAN who had\nhelped him. He is gone."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BAR - DAY\nACTION: THE YOUNG MAN from the bar grabs JACOB under the arms and drags him Where am I? Who are you?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Jacob'S Ladder\": THE YOUNG MAN from the bar grabs JACOB under the arms and drags him Where am I? Who are you?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. THE YOUNG MAN from the bar grabs JACOB under the arms and drags him Where am I? Who are you? The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Les Miserables\n\nSCREENPLAY EXCERPT:\nEnough of that, or I'll forget to\n be nice!\n You heard me ask for something,\n And I never ask twice!\n\nThe innkeeper's gang posing as customers who have been\nloitering outside come in, explaining to a fifth man who is a\nnewer recruit a story from Monsieur Thenardier's past. Madame\nThenardier goes outside to open up the rather pathetic frost\nfair stall on the porch of the inn.\n\n CUSTOMER 1\n Mine host Thenardier\n He was there, so they say,\n At the field of Waterloo.\n\n CUSTOMER 2\n Got there, it's true,\n When the fight was all through\n But he knew just what to do.\n\n CUSTOMER 3\n Crawling through the mud,\n So I've heard it said,"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Enough of that, or I'll forget to You heard me ask for something,\nDIALOGUE — Customer 1: Mine host Thenardier He was there, so they say, At the field of Waterloo.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Les Miserables: Enough of that, or I'll forget to You heard me ask for something,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Enough of that, or I'll forget to You heard me ask for something, The camera pushes in with small amplitude at slow speed. Customer 1 (S1) says: [English] Mine host Thenardier He was there, so they say, At the field of Waterloo. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Ladies and Gentlemen, it's been a\nSetting: derived from Austin Powers: International Man Of Mystery\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. DR. EVIL'S PRIVATE QUARTERS\nACTION: Ladies and Gentlemen, it's been a long time, but I'm back. It's all\nDIALOGUE — Dr. Evil: Ladies and Gentlemen, it's been a long time, but I'm back. It's all gone perfectly to plan except for one small flaw. Because of a technical error, my right...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Austin Powers: International Man Of Mystery: Ladies and Gentlemen, it's been a long time, but I'm back. It's all, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Ladies and Gentlemen, it's been a long time, but I'm back. It's all The camera pushes in with small amplitude at slow speed. Dr. Evil (S1) says: [English] Ladies and Gentlemen, it's been a long time, but I'm back. It's all gone perfectly to plan except for one small flaw. Because of a technical error, my right... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: American History X\n\nSCREENPLAY EXCERPT:\nEXT. VENICE BEACH HOUSE - A RAINY NIGHT - FLASHBACK\n\nAs the rain falls, we see the Vinyards sitting at the\ntable. Stacey and Murray Rosenberg, Doris' then\nboyfriend who we met earlier, round out the clan. Stacey\nsits close to Derek.\n\n DANNY (V/O)\nAnd then came October 4th, 1993."}, {"role": "assistant", "content": "## SCENE 1 — EXT. VENICE BEACH HOUSE - A RAINY NIGHT - FLASHBACK\nACTION: As the rain falls, we see the Vinyards sitting at the table. Stacey and Murray Rosenberg, Doris' then\nDIALOGUE — Danny: And then came October 4th, 1993.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American History X: As the rain falls, we see the Vinyards sitting at the table. Stacey and Murray Rosenberg, Doris' then, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. As the rain falls, we see the Vinyards sitting at the table. Stacey and Murray Rosenberg, Doris' then The camera pushes in with small amplitude at slow speed. Danny (S1) says: [English] And then came October 4th, 1993. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: No, I am uptight. I hate it... I don't want\nSetting: derived from T W I N P E A K S\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: No, I am uptight. I hate it... I don't want to be this way, but Laura I don't... I\nDIALOGUE — Laura: You know, even when I think about your face I get happier.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T W I N P E A K S: No, I am uptight. I hate it... I don't want to be this way, but Laura I don't... I, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. No, I am uptight. I hate it... I don't want to be this way, but Laura I don't... I The camera pushes in with small amplitude at slow speed. Laura (S1) says: [English] You know, even when I think about your face I get happier. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Barbie\n\nSCREENPLAY EXCERPT:\nINT. GLORIA'S CAR.\n\nGloria drives, making crazy turns. She's an incredible\ndriver. Sasha in the passenger seat. Everyone is screaming.\n\n SASHA\n (so embarrassed)\n God I hope nobody from school saw\n us put a life-size Barbie in our\n car. How did this even happen?!\n\nMattel gains on them. Gloria does some crazy amazing driving.\n\n GLORIA\n I don't know!\n\n SASHA\n How are you here? You're like, an\n idea.\n\n BARBIE MARGOT\n A GREAT idea.\n\n GLORIA\n So, I've been a little lonely\n lately and I found the Barbies we\n used play with --\n\n SASHA GLORIA\nI thought we gave those away! And I started playing and\n making drawings like we used\n to do together because I\n thought it would be fun and\n joyful--\n\n BARBIE MARGOT\n (she understands)\n But it wasn't, was it?\n\n GLORIA\n No, because I started feeling sad\n and weird and then the drawings got\n sad and weird...\n (MORE)\n 60.\n GLORIA (CONT'D)\n and maybe because I couldn't be\n like you I ended up making you like\n me?\n\n BARBIE MARGOT\n Did any of these drawings "}, {"role": "assistant", "content": "## SCENE 1 — INT. GLORIA'S CAR.\nACTION: Gloria drives, making crazy turns. She's an incredible driver. Sasha in the passenger seat. Everyone is screaming.\nDIALOGUE — Sasha: God I hope nobody from school saw us put a life-size Barbie in our car. How did this even happen?!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Barbie: Gloria drives, making crazy turns. She's an incredible driver. Sasha in the passenger seat. Everyone is screaming., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Gloria drives, making crazy turns. She's an incredible driver. Sasha in the passenger seat. Everyone is screaming. The camera pushes in with small amplitude at slow speed. Sasha (S1) says: [English] God I hope nobody from school saw us put a life-size Barbie in our car. How did this even happen?! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Delocated illustrating the trope 'TheMafiya'. Played for Laughs in Delocated with Yvgeni Mirminsky, the vodka enthusiast assassin with an ambition for stand-up comedy. Then the next season they bring in his brother Sergei and things get serious.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TheMafiya.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Delocated, TheMafiya, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TheMafiya. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in KaizokuSentaiGokaiger illustrating the trope 'TightsUnderShorts'. Super Sentai:\n Nai from Mahou Sentai Magiranger wears this plus stockings over the tights, while the other half of her Literal Split Personality just wears the typical dress and thigh high combo instead.\n Every single outfit Luka from Kaizoku Sentai Gokaiger wears is a shorts/tights combination. To the point where, if you weren't paying enough attention, you may think that her wardrobe is way smal...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TightsUnderShorts.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for KaizokuSentaiGokaiger, TightsUnderShorts, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TightsUnderShorts. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BlueBloods illustrating the trope 'InstantHumiliationJustAddYoutube'. In one episode of Blue Bloods an anti-NYPD blogger posts a video of a cop beating the crap out of an old man. He had maliciously edited out the part where the old man made a grab for the officer's Glock. Later a music video version turns up, set to \"Another One Bites the Dust\". Frank gets an A.D.A. to subpoena the original and releases it to the media to prove to the world that the officer acted w...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying InstantHumiliationJustAddYoutube.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BlueBloods, InstantHumiliationJustAddYoutube, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for InstantHumiliationJustAddYoutube. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Can we talk about this later?\nSetting: derived from Int: Bedroom. Early-Morning Hours\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Can we talk about this later? The jizz-mopper's job is to clean\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int: Bedroom. Early-Morning Hours: Can we talk about this later? The jizz-mopper's job is to clean, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Can we talk about this later? The jizz-mopper's job is to clean The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DoctorWho illustrating the trope 'TheTapeKnewYouWouldSayThat'. In this The Hero of Three Faces strip, the Third Doctor explains to the agency on the other end of a one-way communication device how to turn it into a two-way device, apparently responding to their confusion even though his point is he can't currently hear them.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TheTapeKnewYouWouldSayThat.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DoctorWho, TheTapeKnewYouWouldSayThat, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TheTapeKnewYouWouldSayThat. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Where the hell are you?\nSetting: derived from Bride Of Leprechaun\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. O.S.\nACTION: Where the hell are you? You'll have to open the door, it's\nDIALOGUE — Leprechaun: You'll have to open the door, it's wrought iron. Remember? Didn't you read the book?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bride Of Leprechaun: Where the hell are you? You'll have to open the door, it's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Where the hell are you? You'll have to open the door, it's The camera pushes in with small amplitude at slow speed. Leprechaun (S1) says: [English] You'll have to open the door, it's wrought iron. Remember? Didn't you read the book? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Watchmen\n\nSCREENPLAY EXCERPT:\nINT. DAN'S NIGHTMARE - NIGHT\n\n CLOSE ON: Dan, naked, on his knees in the center of an\n INFINITE FIELD OF BLACKNESS. No corners, no exits. In the\n DISTANCE, he sees a FEMALE FORM. Desperate, he runs to her.\n It's Laurie, waiting for him IN COSTUME. They kiss. She tears\n at him passionately. His SKIN suddenly TEARS AWAY like\n wrapping paper. Beneath the false skin, Dan wears his NITE-\n OWL COSTUME. Confident, he touches her face. She smiles with\n palpable heat. They kiss.\n In the background, a NUCLEAR BLAST ERUPTS. Still kissing, the\n\n BLAST BLOWS THEM INTO TWO EMBRACING SKELETONS."}, {"role": "assistant", "content": "## SCENE 1 — INT. DAN'S NIGHTMARE - NIGHT\nACTION: CLOSE ON: Dan, naked, on his knees in the center of an INFINITE FIELD OF BLACKNESS. No corners, no exits. In the\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Watchmen: CLOSE ON: Dan, naked, on his knees in the center of an INFINITE FIELD OF BLACKNESS. No corners, no exits. In the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CLOSE ON: Dan, naked, on his knees in the center of an INFINITE FIELD OF BLACKNESS. No corners, no exits. In the The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Barbie\n\nSCREENPLAY EXCERPT:\nEXT. BARBIE DREAMHOUSE/CUL-DE-SAC. MORNING\n\nBarbie Margot's having a send off party with all of the\nBarbies - there is a big banner that reads:\n\n\"Bon Voyage to Reality and Good Luck Restoring The Membrane\nThat Separates Our World From Theirs So You Don't Get\nCellulite!\"\n\nThe Kens all stand to the side, a bit like men at a baby\nshower. There, but not. Ken Ryan Gosling and Ken Simu do a\njealous mingle.\n\n KEN SIMU\n I guess she's going without you.\n 26.\n\n\n KEN RYAN GOSLING\n (clearly lying)\n No, she literally asked me, but I\n preferred to stay here.\n\n KEN SIMU\n (provoking)\n Why? Are you scared?\n\n KEN RYAN GOSLING\n No!\n\n KEN SIMU\n I'll bet you're scared and I bet\n she doesn't even want you to go.\n\n KEN RYAN GOSLING\n You bet both of those things\n incorrectly! I bet the other\n direction!\n\n KEN SIMU\n Which way is that? You don't even\n know.\n\nBarbie gathers around her car with her friends.\n\n BARBIE MARGOT\n I just don't want to leave! I'm\n Trying to find reasons not to\n leave!! I'm going to miss you guys\n so much. I just wish someone could\n come with me... but you can't. I\n should do this alone.\n\nBack to the Kens. The other group of Kens are playing a very\nmysterious game which appears to be m"}, {"role": "assistant", "content": "## SCENE 1 — EXT. BARBIE DREAMHOUSE/CUL-DE-SAC. MORNING\nACTION: Barbie Margot's having a send off party with all of the Barbies - there is a big banner that reads:\nDIALOGUE — Ken Simu: I guess she's going without you. 26.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Barbie: Barbie Margot's having a send off party with all of the Barbies - there is a big banner that reads:, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Barbie Margot's having a send off party with all of the Barbies - there is a big banner that reads: The camera pushes in with small amplitude at slow speed. Ken Simu (S1) says: [English] I guess she's going without you. 26. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Needy and Chip walk to an alley behind the school.\nSetting: derived from Jennifer'S Body\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. KETTLE HIGH CAMPUS - CONTINUOUS\nACTION: Needy and Chip walk to an alley behind the school. Melting snow drips off the eaves.\nDIALOGUE — Needy: Look at this. She thrusts a piece of paper into CHIP's hands.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jennifer'S Body: Needy and Chip walk to an alley behind the school. Melting snow drips off the eaves., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Needy and Chip walk to an alley behind the school. Melting snow drips off the eaves. The camera pushes in with small amplitude at slow speed. Needy (S1) says: [English] Look at this. She thrusts a piece of paper into CHIP's hands. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A monkey dressed like Mussolini addresses a huge crowd of\nSetting: derived from Jay And Silent Bob Strike Back\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BALCONY--DAY\nACTION: A monkey dressed like Mussolini addresses a huge crowd of apes, who wave fists in the air.\nDIALOGUE — Jay: Then all it'll take is one little monkey in a spiffy suit to whip the dumber chimps into a frenzy, until they go all ape-shit and start demanding more bananas,...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jay And Silent Bob Strike Back: A monkey dressed like Mussolini addresses a huge crowd of apes, who wave fists in the air., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A monkey dressed like Mussolini addresses a huge crowd of apes, who wave fists in the air. The camera pushes in with small amplitude at slow speed. Jay (S1) says: [English] Then all it'll take is one little monkey in a spiffy suit to whip the dumber chimps into a frenzy, until they go all ape-shit and start demanding more bananas,... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You wouldn't believe it if I told you.\nSetting: derived from The Revenge\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You wouldn't believe it if I told you. In fact, uh, I'm glad\nDIALOGUE — George: Oh.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Revenge: You wouldn't believe it if I told you. In fact, uh, I'm glad, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You wouldn't believe it if I told you. In fact, uh, I'm glad The camera pushes in with small amplitude at slow speed. George (S1) says: [English] Oh. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I guess I had it coming. Let's call\nSetting: derived from Alien\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: I guess I had it coming. Let's call After a hard stare at Roby, Standard gives him a curt nod and turns\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Alien: I guess I had it coming. Let's call After a hard stare at Roby, Standard gives him a curt nod and turns, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I guess I had it coming. Let's call After a hard stare at Roby, Standard gives him a curt nod and turns The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \" T E R M I N A T O R \"\n\nSCREENPLAY EXCERPT:\nI'm okay, really. Listen, I\n want you to pack some stuff\n and go up to the cabin for a\n few days. Just don't...no,\n don't ask any questions.\n Just do it. I gotta get\n going...gotta go. Bye, bye.\n\n Sarah has been idly leafing through the DIRECTORY. On\n a whim she looks up something.\n She freezes for a moment when she finds the listing.\n Then with a triumphant expression she rips the page out\n of the book.\n CUT TO:\n\n\n192 EXT. SERVICE STATION/PICNIC TABLE - DAY 192\n\n Sarah is sitting at a table under a tree, lettering\n something with a lipstick on a cardboard box-flap.\n\n E.C.U. - SIGN, as the last letters are finished.\n It reads:\n SILICON VALLEY\n\n FULL ON SARAH as she retracts the lipstick and leans\n across to hand it to a girl at the next table.\n\n SARAH\n Thanks a lot.\n\n REESE (V.O.)\n What's that?\n\n Sarah looks up, startled to see him standing beside her.\n\n SARAH\n That's where we're going.\n\n REESE\n Why?\n\n Sarah point to the directory page lying on the table.\n\n MACRO - PAGE\n\n Sarah's finger points to a listing which reads:\n\n CYBER DYNAMICS CORPORATION\n 18144"}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: I'm okay, really. Listen, I want you to pack some stuff\nDIALOGUE — Sarah: Thanks a lot.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \" T E R M I N A T O R \": I'm okay, really. Listen, I want you to pack some stuff, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm okay, really. Listen, I want you to pack some stuff The camera pushes in with small amplitude at slow speed. Sarah (S1) says: [English] Thanks a lot. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You're a stupid, fat arsed,\nSetting: derived from 1 Ext. Cityscape Sunset 1\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You're a stupid, fat arsed, obstructive, fascist moron aren't\nDIALOGUE — Porter: If you say so.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from 1 Ext. Cityscape Sunset 1: You're a stupid, fat arsed, obstructive, fascist moron aren't, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You're a stupid, fat arsed, obstructive, fascist moron aren't The camera pushes in with small amplitude at slow speed. Porter (S1) says: [English] If you say so. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Needy runs into the park and scans the dark horizon. Her\nSetting: derived from Jennifer'S Body\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. MC CULLUM PARK - CONTINUOUS\nACTION: Needy runs into the park and scans the dark horizon. Her bare feet are bloody from running and her face is shiny\nDIALOGUE — Needy: Chip! A terrified SCREAM in the distance. Needy whips around, hikes up her skirt and starts running toward Murphy Pool.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jennifer'S Body: Needy runs into the park and scans the dark horizon. Her bare feet are bloody from running and her face is shiny, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Needy runs into the park and scans the dark horizon. Her bare feet are bloody from running and her face is shiny The camera pushes in with small amplitude at slow speed. Needy (S1) says: [English] Chip! A terrified SCREAM in the distance. Needy whips around, hikes up her skirt and starts running toward Murphy Pool. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I'm not giving you anything 'til I\nSetting: derived from Bride Of Leprechaun\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: I'm not giving you anything 'til I get Bridget. You'll have to kill\nDIALOGUE — Leprechaun: So be it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bride Of Leprechaun: I'm not giving you anything 'til I get Bridget. You'll have to kill, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm not giving you anything 'til I get Bridget. You'll have to kill The camera pushes in with small amplitude at slow speed. Leprechaun (S1) says: [English] So be it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in HogansHeroes illustrating the trope 'CulturalTranslation'. Happens a lot with regional accents: British English, for example, is often rendered as European Spanish in Latin American dubs.\n The Kansai dialect of Japanese was often dubbed in English as a Brooklyn accent, due to similar stereotypes about the people who speak them.\n Though the convention seems to have changed to a Texan accent, which still often works due to different nuances in the stereotyp...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying CulturalTranslation.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for HogansHeroes, CulturalTranslation, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for CulturalTranslation. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Karl, I know you want him, but the\nSetting: derived from \"Die Hard\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. OPPOSITE SIDE\nACTION: Karl, I know you want him, but the police are probably on their way.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Die Hard\": Karl, I know you want him, but the police are probably on their way., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Karl, I know you want him, but the police are probably on their way. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: MUCH ACTIVITY ... HORNS SOUND ... the ENTIRE CAMP is on the\nSetting: derived from The Lord Of The Rings: The Return Of The King\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. DUNHARROW ENCAMPMENT - DAWN\nACTION: MUCH ACTIVITY ... HORNS SOUND ... the ENTIRE CAMP is on the MOVE ... fires are DOUSED, TENTS pulled down, HORSES saddled.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Return Of The King: MUCH ACTIVITY ... HORNS SOUND ... the ENTIRE CAMP is on the MOVE ... fires are DOUSED, TENTS pulled down, HORSES saddled., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. MUCH ACTIVITY ... HORNS SOUND ... the ENTIRE CAMP is on the MOVE ... fires are DOUSED, TENTS pulled down, HORSES saddled. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Two hours later, the SNOWSTORM's grown into a full-fledged\nSetting: derived from Batman 2\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. GOTHAM SQUARE - LATER THAT NIGHT\nACTION: Two hours later, the SNOWSTORM's grown into a full-fledged blizzard. The drifts are two feet deep and the streets are\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman 2: Two hours later, the SNOWSTORM's grown into a full-fledged blizzard. The drifts are two feet deep and the streets are, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Two hours later, the SNOWSTORM's grown into a full-fledged blizzard. The drifts are two feet deep and the streets are The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dawn Of\n\nSCREENPLAY EXCERPT:\nby George A. Romero\n\n\n1\tWe see the face of a young woman. She is asleep. It is very\n\tquiet at first, as credits appear. The woman's face begins to\n\ttwitch, as though she is having a bad dream. She moans slightly\n\tand her expression grows more desperate.\n\n\tA mix of subtle sounds begin to fade in. As they get louder, we\n\tcan discern what sounds like a busy office area. It is actually\n\ta frantic television studio with the hum of panic in a national\n\temergency.\n\n\tThe woman's moans get louder and more desperate as the\n\tbackground sounds reach full volume and the credits stop. The\n\twoman sits up, snapping awake.\n\n2\tShe lurches forwards into the arms of a strong young man. She is\n\tFrancine, twenty three years old and very attractive, although\n\tshe is gritty with dirt. Her hair is hanging, dishevelled and\n\tsweaty. Her jeans and blouse have been worn for several days.\n\n\tShe is sitting on the floor, where she has slept the last\n\tseveral hours, covered by an old overcoat.\n\n\tTony:\tYOU OK?\n\n\tFran stares at the young man. She is shaking. She doesn't speak.\n\n\tTony:\tTHE SHIT'S REALLY HITTING THE FAN.\n\n\tThe girl tries to clear her head as the young man moves on to\n\twhere others sleep on the floor. He wakes them up one at a time.\n\tWe begin to hear voices over the busy hum of the studio. They\n\thave an electronic tinniness, as broadcast over a monitor. Fran\n\tlooks about. She is still shaken from her dream.\n\n3\tWe see the television studio. Reporters buzz about madly.\n\tEverybody looks dishevelled and exhausted. Technicians man\n\tmonitors, and we see people on the little screens, arguing\n\temotiona"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: 1\tWe see the face of a young woman. She is asleep. It is very quiet at first, as credits appear. The woman's face begins to\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dawn Of: 1\tWe see the face of a young woman. She is asleep. It is very quiet at first, as credits appear. The woman's face begins to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. 1\tWe see the face of a young woman. She is asleep. It is very quiet at first, as credits appear. The woman's face begins to The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The giant twin suns of Tatooine slowly disappear behind a\nSetting: derived from Star Wars\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. TATOOINE - LARS HOMESTEAD\nACTION: The giant twin suns of Tatooine slowly disappear behind a distant dune range. Luke stands watching them for a few\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Star Wars: The giant twin suns of Tatooine slowly disappear behind a distant dune range. Luke stands watching them for a few, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The giant twin suns of Tatooine slowly disappear behind a distant dune range. Luke stands watching them for a few The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Star Wars\n\nSCREENPLAY EXCERPT:\nINT. KENOBI'S DWELLING\n\n The small, spartan hovel is cluttered with desert junk but\n still manages to radiate an air of time-worn comfort and\n security. Luke is in one corner repairing Threepio's arm, as\n old Ben sits thinking.\n\n LUKE\n No, my father didn't fight in the\n wars. He was a navigator on a spice\n freighter.\n\n BEN\n That's what your uncle told you. He\n didn't hold with your father's ideals.\n Thought he should have stayed here\n and not gotten involved.\n\n LUKE\n You fought in the Clone Wars?\n\n BEN\n Yes, I was once a Jedi Knight the\n same as your father.\n\n LUKE\n I wish I'd known him.\n\n BEN\n He was the best star-pilot in the\n galaxy, and a cunning warrior. I\n understand you've become quite a\n good pilot yourself. And he was a\n good friend. Which reminds me...\n\n Ben gets up and goes to a chest where he rummages around.\n As Luke finishes repairing Threepio and starts to fit the\n restraini"}, {"role": "assistant", "content": "## SCENE 1 — INT. KENOBI'S DWELLING\nACTION: The small, spartan hovel is cluttered with desert junk but still manages to radiate an air of time-worn comfort and\nDIALOGUE — Luke: No, my father didn't fight in the wars. He was a navigator on a spice freighter.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Star Wars: The small, spartan hovel is cluttered with desert junk but still manages to radiate an air of time-worn comfort and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The small, spartan hovel is cluttered with desert junk but still manages to radiate an air of time-worn comfort and The camera pushes in with small amplitude at slow speed. Luke (S1) says: [English] No, my father didn't fight in the wars. He was a navigator on a spice freighter. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Batman\n\nSCREENPLAY EXCERPT:\nEXT. ACE CHEMICAL - NIGHT\n\n ECKHARDT'S SWAT TEAM goes wide-eyed as a CONVOY OF POLICE\n BLACK-AND-WHITES roars into the Ace parking lot. UNIFORMED\n COPS pile out of their squad cars, relieving the SWAT TEAM.\n ECKHARDT goes livid as COMMISSIONER GORDON approaches.\n\n ECKHARDT\n What are you trying to do, blow the\n collar?\n\n GORDON\n (to SWAT TEAM)\n You men are dismissed. We'll take\n over from here.\n (to UNIFORMED COPS)\n Any man who opens fire on Jack\n Napier... will answer to me.\n\n ECKHARDT tries to slink off. GORDON grabs him roughly.\n\n GORDON (cont.)\n You. Stick around."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ACE CHEMICAL - NIGHT\nACTION: ECKHARDT'S SWAT TEAM goes wide-eyed as a CONVOY OF POLICE BLACK-AND-WHITES roars into the Ace parking lot. UNIFORMED\nDIALOGUE — Eckhardt: What are you trying to do, blow the collar?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman: ECKHARDT'S SWAT TEAM goes wide-eyed as a CONVOY OF POLICE BLACK-AND-WHITES roars into the Ace parking lot. UNIFORMED, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ECKHARDT'S SWAT TEAM goes wide-eyed as a CONVOY OF POLICE BLACK-AND-WHITES roars into the Ace parking lot. UNIFORMED The camera pushes in with small amplitude at slow speed. Eckhardt (S1) says: [English] What are you trying to do, blow the collar? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Untitled\n\nSCREENPLAY EXCERPT:\nAsh\nMewtwo?\n\nUmio\nA Pok�mon is a pok�mon trainer? It can't be.\n\t(It becomes apparent that MewTwo is telepathically\ncontrolling Joy...)\n\nMewtwo\nI've decided my own rules.\n\nMisty\nThat voice!\n\nBrock\nTelepathy!\n\t(Mewtwo raises Umio up in the air and throws him into a\npool"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: A Pok�mon is a pok�mon trainer? It can't be. (It becomes apparent that MewTwo is telepathically\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: A Pok�mon is a pok�mon trainer? It can't be. (It becomes apparent that MewTwo is telepathically, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A Pok�mon is a pok�mon trainer? It can't be. (It becomes apparent that MewTwo is telepathically The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dawn Of\n\nSCREENPLAY EXCERPT:\nSHE'S ALRIGHT. LET HER THROUGH.\n\n\tThe officer reluctantly steps aside.\n\n15\t The young man and Fan move down a crowded hall and into a small\n\tcamera room. The foot traffic is solid. They talk as they walk.\n\n\tFran:\tI DON'T BELIEVE IT.\n\n\tMan:\t ONE OF THOSE LITTLE BADGES CAN OPEN A LOT\n\t\t\tOF DOORS...YOU AVOID A LOT OF HASSLES IF\n\t\t\tYOU GOT A BADGE...ANY KIND OF BADGE...\n\n\tFran:\tIT'S REALLY GOING CRAZY.\n\n16\t They reach a small camera installation. The camera is aimed at a\n\tmachine which rolls out a list of rescue stations. The list is\n\tsuperimposed over the live broadcast as it goes out.\n\n\tCameraman: YOU GOT NEW ONES?\n\n\tFran:\tI GOTTA TYPE 'EM UP. KILL THE OLD ONES.\n\n\tCameraman: GIVENS WANT 'EM...\n\n\tFran:\tKILL 'EM, DICK. TELL GIVENS TO SEE ME!\n\n\tThe man clicks off his camera. Fran moves toward the studio.\n\n17\t On the monitors, we see the rescue stations blink off over shots\n\tof the two men who still argue on the air.\n\n\tTV Man 1: WELL I DON'T BELIEVE IN GHOSTS, DOCTOR.\n\n\tTV Man 2: THESE ARE NOT GHOSTS. NOR ARE THESE HUMANS!\n\t\t\tTHESE ARE DEAD CORPSES. ANY UN-BURIED HUMAN\n\t\t\tCORPSE WITH ITS BRAIN INTACT WILL IN FACT\n\t\t\tRE-ACTIVATE. AND IT'S PRECISELY BECAUSE OF\n\t\t\tINCITEMENT BY IRRESPONSIBLE PUBLIC FIGURES\n\t\t\tLIKE YOURSELF THAT THIS SITUATION IS BEING\n\t\t\tDEALT WITH IRRESPONSIBLY BY THE PUBLIC AT\n\t\t\tLARGE!\n\n18\t Another outraged cry goes up from the stagehands and observers.\n\tDoctor Foster tries to out-scream the cries...\n\n\tTV Man 2: YOU HAVE NOT LISTENED...YOU HAVE NOT LISTENED...\n\t\t\tFOR THE LAST THREE WEEKS...WHAT DOES IT TAKE...\n\t\t\tWHAT DOES IT TAKE TO MAKE PEOPLE SEE?"}, {"role": "assistant", "content": "## SCENE 1 — INT. NOW DISTRAUGHT...ALMOST PLEADING\nACTION: The officer reluctantly steps aside. 15\t The young man and Fan move down a crowded hall and into a small\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dawn Of: The officer reluctantly steps aside. 15\t The young man and Fan move down a crowded hall and into a small, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The officer reluctantly steps aside. 15\t The young man and Fan move down a crowded hall and into a small The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: 196 SARK'S AIRCRAFT CARRIER - THE BRIDGE 196\nSetting: derived from T R O N\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: 196 SARK'S AIRCRAFT CARRIER - THE BRIDGE 196 High above the game we see Sark watching.\nDIALOGUE — Flynn: Here's an easy one...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T R O N: 196 SARK'S AIRCRAFT CARRIER - THE BRIDGE 196 High above the game we see Sark watching., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. 196 SARK'S AIRCRAFT CARRIER - THE BRIDGE 196 High above the game we see Sark watching. The camera pushes in with small amplitude at slow speed. Flynn (S1) says: [English] Here's an easy one... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Stewardesses Elaine and RANDY are greeting boarding\nSetting: derived from Airplane\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PASSENGER CABIN - BOARDING AREA - NIGHT\nACTION: Stewardesses Elaine and RANDY are greeting boarding passengers including the Hammens and SISTER ANGELINA who is\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Airplane: Stewardesses Elaine and RANDY are greeting boarding passengers including the Hammens and SISTER ANGELINA who is, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Stewardesses Elaine and RANDY are greeting boarding passengers including the Hammens and SISTER ANGELINA who is The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Star Wars\n\nSCREENPLAY EXCERPT:\nEXT. TATOOINE - WASTELAND\n\n The speeder stops before what remains of the huge Jawas\n Sandcrawler. Luke and Ben walk among the smoldering rubble\n and scattered bodies.\n\n LUKE\n It looks like Sandpeople did this,\n all right. Look, here are Gaffi\n sticks, Bantha tracks. It's just...\n I never heard of them hitting anything\n this big before.\n\n Ben is crouching in the sand studying the tracks.\n\n BEN\n They didn't. But we are meant to\n think they did. These tracks are\n side by side. Sandpeople always ride\n single file to hide there numbers.\n\n LUKE\n These are the same Jawas that sold\n us Artoo and Threepio.\n\n BEN\n And these blast points, too accurate\n for Sandpeople. Only Imperial\n stormtroopers are so precise.\n\n LUKE\n Why would Imperial troops want to\n slaughter Jawas?\n\n Luke looks back at the speeder where Artoo and Threepio are\n inspecting the dead Jawas, and put two and two together.\n\n LUKE\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. TATOOINE - WASTELAND\nACTION: The speeder stops before what remains of the huge Jawas Sandcrawler. Luke and Ben walk among the smoldering rubble\nDIALOGUE — Luke: It looks like Sandpeople did this, all right. Look, here are Gaffi sticks, Bantha tracks. It's just... I never heard of them hitting anything this big before.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Star Wars: The speeder stops before what remains of the huge Jawas Sandcrawler. Luke and Ben walk among the smoldering rubble, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The speeder stops before what remains of the huge Jawas Sandcrawler. Luke and Ben walk among the smoldering rubble The camera pushes in with small amplitude at slow speed. Luke (S1) says: [English] It looks like Sandpeople did this, all right. Look, here are Gaffi sticks, Bantha tracks. It's just... I never heard of them hitting anything this big before. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: She absentmindedly obeys.\nSetting: derived from Django Unchained\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. .GERMAN\nACTION: She absentmindedly obeys. Now it's myself and our mutual friend's\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Django Unchained: She absentmindedly obeys. Now it's myself and our mutual friend's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. She absentmindedly obeys. Now it's myself and our mutual friend's The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Watchmen\n\nSCREENPLAY EXCERPT:\nINT. PRISON PSYCH ROOM - DAY\n\n The Psychiatrist listens, paralyzed. Horrified.\n\n RORSCHACH\n A tremor of impact shook my arm. Warm\n blood splashed my face. It was Walter\n Kovacs who screamed, who closed his eyes--\n it was Rorschach who opened them again.\n The silence reels out\n RORSCHACH (cont'd)\n You see, Doctor. God didn't kill that\n little girl, fate didn't butcher her and\n destiny didn't feed her to those dogs. If\n God saw what any of us did that night, he\n didn't seem to mind. From then on, I\n knew. God doesn't make the world this\n way. We do.\n The Psychiatrist rises shakily. He crosses to the GUARD.\n\n PSYCHIATRIST\n I can't help him."}, {"role": "assistant", "content": "## SCENE 1 — INT. PRISON PSYCH ROOM - DAY\nACTION: The Psychiatrist listens, paralyzed. Horrified. A tremor of impact shook my arm. Warm\nDIALOGUE — Rorschach: A tremor of impact shook my arm. Warm blood splashed my face. It was Walter Kovacs who screamed, who closed his eyes-- it was Rorschach who opened them again....\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Watchmen: The Psychiatrist listens, paralyzed. Horrified. A tremor of impact shook my arm. Warm, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The Psychiatrist listens, paralyzed. Horrified. A tremor of impact shook my arm. Warm The camera pushes in with small amplitude at slow speed. Rorschach (S1) says: [English] A tremor of impact shook my arm. Warm blood splashed my face. It was Walter Kovacs who screamed, who closed his eyes-- it was Rorschach who opened them again.... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jennifer'S Body\n\nSCREENPLAY EXCERPT:\nINT. NEEDY'S HOUSE (LIVING ROOM) - EVENING\n\n Needy is fully dressed and garishly made up. Toni holds a\n digital camera at arm's length and takes a shot of the\n two of them together.\n The mother/daughter photo looks ghastly. Toni's eyes are\n closed and Needy pouts miserably."}, {"role": "assistant", "content": "## SCENE 1 — INT. NEEDY'S HOUSE (LIVING ROOM) - EVENING\nACTION: Needy is fully dressed and garishly made up. Toni holds a digital camera at arm's length and takes a shot of the\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jennifer'S Body: Needy is fully dressed and garishly made up. Toni holds a digital camera at arm's length and takes a shot of the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Needy is fully dressed and garishly made up. Toni holds a digital camera at arm's length and takes a shot of the The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Ext. Manhattan Island - Day\n\nSCREENPLAY EXCERPT:\nINT. ECTO-1 - DAY (CONTINUOUS ACTION)\n\nStantz tries to start the car, but the engine won't turn over.\n\n\t\t\t\t\t\tWINSTON\n\t\t\tGive it up, Ray. You're living in the past.\n\t\t\tGhostbusters doesn't exist anymore. In a year\n\t\t\tthese kids won't even remember who we are.\n\n\t\t\t\t\t\tSTANTZ\n\t\t\t\t (tries to start the car again)\n\t\t\tUngrateful little Yuppie larvae. After all we\n\t\t\tdid for this city.\n\n\t\t\t\t\t\tWINSTON\n\t\t\tYeah, what did we do, Ray? The last real job\n\t\t\twe had we bubbled up a hundred foot marshmallow\n\t\t\tman and blew the top three floors off an uptown\n\t\t\thighrise.\n\n\t\t\t\t\t\tSTANTZ\n\t\t\tYeah, but what a ride. You can't make a\n\t\t\thamburger without chopping up a cow.\n\nHe turns the key again, the ENGING TURNS OVER, then starts GRINDING and\nCLUNKING disastrously, chewing up vital parts and dropping twisted bits\nof metal onto the pavement. Finally, with a BLAST of black sooty\nexhaust from the tailpipe, Ecto-1 shudders and dies. Frustrated, Stantz\nbangs his head lightly on the steering wheel\n\n\t\t\t\t\t\t\t\t\t\t\t\t CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — INT. ECTO-1 - DAY (CONTINUOUS ACTION)\nACTION: Stantz tries to start the car, but the engine won't turn over. Give it up, Ray. You're living in the past.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ext. Manhattan Island - Day: Stantz tries to start the car, but the engine won't turn over. Give it up, Ray. You're living in the past., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Stantz tries to start the car, but the engine won't turn over. Give it up, Ray. You're living in the past. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The speeder stops before what remains of the huge Jawas\nSetting: derived from Star Wars\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. TATOOINE - WASTELAND\nACTION: The speeder stops before what remains of the huge Jawas Sandcrawler. Luke and Ben walk among the smoldering rubble\nDIALOGUE — Luke: It looks like Sandpeople did this, all right. Look, here are Gaffi sticks, Bantha tracks. It's just... I never heard of them hitting anything this big before.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Star Wars: The speeder stops before what remains of the huge Jawas Sandcrawler. Luke and Ben walk among the smoldering rubble, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The speeder stops before what remains of the huge Jawas Sandcrawler. Luke and Ben walk among the smoldering rubble The camera pushes in with small amplitude at slow speed. Luke (S1) says: [English] It looks like Sandpeople did this, all right. Look, here are Gaffi sticks, Bantha tracks. It's just... I never heard of them hitting anything this big before. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: KIMBALL and Bateman are sitting at a corner table.\nSetting: derived from American Psycho\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SMITH AND WOLLENSKY RESTAURANT - DAY\nACTION: KIMBALL and Bateman are sitting at a corner table. Not in the mood, I guess.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American Psycho: KIMBALL and Bateman are sitting at a corner table. Not in the mood, I guess., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. KIMBALL and Bateman are sitting at a corner table. Not in the mood, I guess. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Airplane\n\nSCREENPLAY EXCERPT:\nINT. PASSENGER CABIN - BOARDING AREA - NIGHT\n\n Stewardesses Elaine and RANDY are greeting boarding\n passengers including the Hammens and SISTER ANGELINA who is\n carrying a guitar."}, {"role": "assistant", "content": "## SCENE 1 — INT. PASSENGER CABIN - BOARDING AREA - NIGHT\nACTION: Stewardesses Elaine and RANDY are greeting boarding passengers including the Hammens and SISTER ANGELINA who is\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Airplane: Stewardesses Elaine and RANDY are greeting boarding passengers including the Hammens and SISTER ANGELINA who is, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Stewardesses Elaine and RANDY are greeting boarding passengers including the Hammens and SISTER ANGELINA who is The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nINT. CAMPBELL CENTER - LABORATORY - DAY\n\n\tIn the darkened procedure room, Miriam finds Catherine\n\tstanding between the two suspension devices, alone.\n\n\t\t\t\tMIRIAM\n\t\tYou okay?\n\n\tCatherine turns, nods, and tries to smile. Has she been\n\tcrying? Miriam steps further into the room.\n\n\t\t\t\tCATHERINE\n\t\tHe hates me.\n\n\t\t\t\tMIRIAM\n\t\tLucien Baines?\n\n\tCatherine nods.\n\n\t\t\t\tMIRIAM (CONT'D)\n\t\tI wouldn't say \"hates.\" Dislikes,\n\t\tdistrusts, maybe.\n\n\tCatherine laughs. Exactly what Miriam wanted.\n\n\t\t\t\tMIRIAM (CONT'D)\n\t\tShe adores you.\n\n\t\t\t\tCATHERINE\n\t\tShe thinks I can bring Edward back.\n\n\t\t\t\tMIRIAM\n\t\tSomeday you can. We all believe that...\n\n\tCatherine halfheartedly nods, but her mind is elsewhere,\n\tfocused on the two apparatuses.\n\n\t\t\t\tCATHERINE\n\t\tI want to try it. Please.\n\n\tMiriam knows what this is about.\n\n\t\t\t\tMIRIAM\n\t\tWe've been over this a dozen times. No.\n\n\t\t\t\tCATHERINE\n\t\tWhy not? Just once. A trial run. What\n\t\tharm could it do?\n\n\t\t\t\tMIRIAM\n\t\tI don't want to find out.\n\n\tEver the doctor, Miriam finds herself surreptitiously\n\texamining Catherine's eyes, flesh tone, a slight trembling in\n\ther fingertips.\n\n\t\t\t\tCATHERINE\n\t\tWhy is it taking so long for us to reach\n\t\thim? Because we've been reactive,\n\t\tthat's why. And we need to be active.\n\n\tMiriam reaches for Catherine's wrist and takes her pulse.\n\n\t\t\t\tCATHERINE (CONT'D)\n\t\t\t(annoyed)\n\t\tMiriam...\n\n\t\t\t\tMIRIAM\n\t\tShush.\n\n\tAs Miriam examines her, Catherine continues:\n\n\t\t\t\tCATHERINE\n\t\tHe creates the obstacles, he decides\n\t\twhich games to play.\n\n\t\t\t\tMIRIAM\n\t\tI've heard this argument before. It's\n\t\tnot happening.\n\n\t\t\t\tCATHERINE\n\t\t\t(frustrated)"}, {"role": "assistant", "content": "## SCENE 1 — INT. CAMPBELL CENTER - LABORATORY - DAY\nACTION: In the darkened procedure room, Miriam finds Catherine standing between the two suspension devices, alone.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: In the darkened procedure room, Miriam finds Catherine standing between the two suspension devices, alone., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. In the darkened procedure room, Miriam finds Catherine standing between the two suspension devices, alone. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Int. Publishing House Reception Area - Day\n\nSCREENPLAY EXCERPT:\nEXT. NEW YORK STREET - DAY\n\nJoel trudges along carrying two big trash bags full of stuff.\nHe's been crying. He looks behind him and finds himself\nlooking out the window of his apartment at the dark van on\nthe snowy street. He turns back to the New York Street and\nspots the address he was looking for: 610 11th Avenue."}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEW YORK STREET - DAY\nACTION: Joel trudges along carrying two big trash bags full of stuff. He's been crying. He looks behind him and finds himself\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int. Publishing House Reception Area - Day: Joel trudges along carrying two big trash bags full of stuff. He's been crying. He looks behind him and finds himself, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Joel trudges along carrying two big trash bags full of stuff. He's been crying. He looks behind him and finds himself The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The PHONE begins to RING as the others crawl in.\nSetting: derived from The Matrix\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. APPLIANCE STORE (MATRIX) - DAY\nACTION: The PHONE begins to RING as the others crawl in. God, I love that sound.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Matrix: The PHONE begins to RING as the others crawl in. God, I love that sound., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The PHONE begins to RING as the others crawl in. God, I love that sound. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: --pavement near the manhole. She sits there, looking down.\nSetting: derived from Jay And Silent Bob Strike Back\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. DAM ROAD--DAY\nACTION: --pavement near the manhole. She sits there, looking down.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jay And Silent Bob Strike Back: --pavement near the manhole. She sits there, looking down., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. --pavement near the manhole. She sits there, looking down. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Could you do something about the labels? This\nSetting: derived from Untitled\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Could you do something about the labels? This is supposed to be a clandestine operation.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: Could you do something about the labels? This is supposed to be a clandestine operation., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Could you do something about the labels? This is supposed to be a clandestine operation. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Lenny and Iris come out the front door.\nSetting: derived from S T R A N G E D A Y S\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CORAL LOUNGE/PARKING LOT - NIGHT\nACTION: Lenny and Iris come out the front door. It's a bad situation... if they get\nDIALOGUE — Iris: It's a bad situation... if they get me I know they're going to -- OH SHIT!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from S T R A N G E D A Y S: Lenny and Iris come out the front door. It's a bad situation... if they get, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Lenny and Iris come out the front door. It's a bad situation... if they get The camera pushes in with small amplitude at slow speed. Iris (S1) says: [English] It's a bad situation... if they get me I know they're going to -- OH SHIT! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Pleasantville\n\nSCREENPLAY EXCERPT:\nINT. KITCHEN. LATER ...\n\n\tThe CAMERA PANS across the kitchen table where there is a\n\tfull quart of milk and some uneaten co6kies. It finally comes\n\tto rest on Jennifer who is staring across the kitchen table\n\twith a concerned look on her face.\n\n\t\t\t\tJENNIFER\n\t\tYou okay?\n\n\tREVERSE ANGLE. BETTY.\n\n\tShe nods--shaken but \"fine.\" Betty stares long and hard at\n\tthe plate of chocolate chip cookies. She's far away.\n\n\t\t\t\tBETTY\n\t\tYes ...\n\t\t\t(softly)\n\t\tIt's just that ...\n\n\t\t\t\tJENNIFER\n\t\t\t(gently)\n\t\tWhat?\n\n\t\t\t\tBETTY\n\t\tWell ...\n\t\t\t(looking up)\n\t\t... Your father would never do anything\n\t\tlike that.\n\n\tSHOT. JENNIFER.\n\n\tShe bites her lip weighing her next sentence ...\n\n\t\t\t\tJENNIFER\n\t\t\t(leaning forward)\n\t\tOh. Hmm ...\n\t\t\t(whisper/\n\t\t\twoman to woman)\n\t\tWell, Mom ... there's ways to \"enjoy\"\n\t\tyourself without Dad.\n\n\tANGLE. BETTY.\n\n\tShe looks across the table, perplexed.\n\n\t\t\t\t\t\t\t\tCUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — INT. KITCHEN. LATER ...\nACTION: The CAMERA PANS across the kitchen table where there is a full quart of milk and some uneaten co6kies. It finally comes\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Pleasantville: The CAMERA PANS across the kitchen table where there is a full quart of milk and some uneaten co6kies. It finally comes, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The CAMERA PANS across the kitchen table where there is a full quart of milk and some uneaten co6kies. It finally comes The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jurassic Park\n\nSCREENPLAY EXCERPT:\nRest stop.\n\nThe kids take off towards the bathroom, running through the rain. Grant\ngets out of his cruiser, strides to Regis. He indicates the fence.\n\n GRANT\n Is that still the T-Rex paddock?\n\n REGIS\n Yes. But she never comes here. I don't\n know why not. Probably too much\n construction.\n\nGENNARO, jacket over his head against the rain, looks down to the ocean.\n\nGENNARO'S POV - THE ANNE B UNLOADS he last cargo crate.\n\nGRANT LOOKS at the concrete moat. Studies its deep curve. He looks up\nat the tell electrical fence with its 10,000 volt warning. He sees\nconventional power lines on the opposite sid\ne of the road.\n\nCAMERA EXAMINES the empty cruisers. Inside, the pre-recorded voice is\nchatting on. It slows eerily and stops. Video SCREENS BLINK OUT.\n\n\nBACK ON CONTROL ROOM -\n\nNedry yawns loudly.\n\n NEDRY\n Yup! Looks like a never-ending weekend\n for me. I'm gonna get a Diet Coke.\n Don't touch my console, ok? Line will\n be clear in five minutes.\n\nNedry leaves. Hammond swings around and growls under his breath.\n\n HAMMOND\n Slob!\n\n ARNOLD\n Well, at least he knows what he's doing.\n\n\nINT UPPER FLOOR, VISITOR'S CENTER - NIGHT\n\nNedry races through the series of security doors. He ignores the\nsecurity x-ray device and just SHOVES each door open with his hand.\n\n\nBACK ON CRUISERS, REST AREA -\n\nTHE QUARTZ LIGHTS GO OUT, leaving the group in shadowy darkness"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The kids take off towards the bathroom, running through the rain. Grant gets out of his cruiser, strides to Regis. He indicates the fence.\nDIALOGUE — Grant: Is that still the T-Rex paddock?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jurassic Park: The kids take off towards the bathroom, running through the rain. Grant gets out of his cruiser, strides to Regis. He indicates the fence., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The kids take off towards the bathroom, running through the rain. Grant gets out of his cruiser, strides to Regis. He indicates the fence. The camera pushes in with small amplitude at slow speed. Grant (S1) says: [English] Is that still the T-Rex paddock? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Blade Runner\n\nSCREENPLAY EXCERPT:\nINT. CORRIDOR - TENTH FLOOR - NIGHT 106\n\n Deckard looks around the corner of the door down the\n hall. Batty's at the other end. Except for jockstrap\n and gym shoes, he's nude.\n\n BATTY\n You wanna play?\n\n Deckard FIRES. Batty's fast. He ducks into a doorway.\n Pops out again.\n\n BATTY\n Not very sporting to fire on an\n unarmed opponent. I thought you\n were supposed to be good. Aren't\n you the man?!\n\n The makeup on Batty's face is somewhere between a Coman-\n che warrior and a transvestite. The immensity of his\n insolence awesome -- the muscles of his body are swol-\n len, trembling from the thrill of it.\n\n BATTY\n This is how we do it up there, lad!\n Come on!\n\n In a blue of lightning-like action, Batty whips down the\n hall, zigzagging off the walls towards Deckard so fast\n that Deckard gets only three SHOTS off before the blur\n crashes through the wall on his left with a laugh.\n\n Deckard stands there a moment -- digesting the impact\n of it, then edges up to the gaping wall. Batty is be-\n hind him.\n\n He knees Deckard in the back and slaps him in the head.\n Deckard goes to his knees, then over on his face.\n Batty kneels next to him.\n\n BATTY\n Not hurt, are you? You bette"}, {"role": "assistant", "content": "## SCENE 1 — INT. CORRIDOR - TENTH FLOOR - NIGHT 106\nACTION: Deckard looks around the corner of the door down the hall. Batty's at the other end. Except for jockstrap\nDIALOGUE — Batty: You wanna play?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade Runner: Deckard looks around the corner of the door down the hall. Batty's at the other end. Except for jockstrap, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Deckard looks around the corner of the door down the hall. Batty's at the other end. Except for jockstrap The camera pushes in with small amplitude at slow speed. Batty (S1) says: [English] You wanna play? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The Statue of Liberty is walking up Broadway approaching 42nd Street,\nSetting: derived from Ext. Manhattan Island - Day\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BROADWAY - NIGHT (CONTINUOUS ACTION)\nACTION: The Statue of Liberty is walking up Broadway approaching 42nd Street, with \"Higher and Higher\" BOOMING from the SPEAKERS inside. A great cheer\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ext. Manhattan Island - Day: The Statue of Liberty is walking up Broadway approaching 42nd Street, with \"Higher and Higher\" BOOMING from the SPEAKERS inside. A great cheer, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The Statue of Liberty is walking up Broadway approaching 42nd Street, with \"Higher and Higher\" BOOMING from the SPEAKERS inside. A great cheer The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: From overhead, the CAMERA CORKSCREWS CLOCKWISE as CHEV\nSetting: derived from Crank\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. REAR EXIT, EVE'S BUILDING - CONTINUOUS\nACTION: From overhead, the CAMERA CORKSCREWS CLOCKWISE as CHEV wrenches the HOOD'S neck COUNTERCLOCKWISE, snapping it.\nDIALOGUE — Eve: You know, I could use a little help here.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Crank: From overhead, the CAMERA CORKSCREWS CLOCKWISE as CHEV wrenches the HOOD'S neck COUNTERCLOCKWISE, snapping it., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. From overhead, the CAMERA CORKSCREWS CLOCKWISE as CHEV wrenches the HOOD'S neck COUNTERCLOCKWISE, snapping it. The camera pushes in with small amplitude at slow speed. Eve (S1) says: [English] You know, I could use a little help here. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Marine Biologist\n\nSCREENPLAY EXCERPT:\nELAINE\n What, what is it?\n\n\n KRAMER\n The sand, it's everywhere!\n\n\n ELAINE\n OK I'll see you later."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The sand, it's everywhere! OK I'll see you later.\nDIALOGUE — Kramer: The sand, it's everywhere!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Marine Biologist: The sand, it's everywhere! OK I'll see you later., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The sand, it's everywhere! OK I'll see you later. The camera pushes in with small amplitude at slow speed. Kramer (S1) says: [English] The sand, it's everywhere! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in CSINY illustrating the trope 'MasochistsMeal'. There's an episode of CSI: NY where the murder is tracked to a festival of bizarre foods such as deep-fired tarantula, with the murder weapon being a live squid. The squid was meant to be eaten alive after its tentacles were bound, but the killer deliberately left them loose so the victim would choke on it.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying MasochistsMeal.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for CSINY, MasochistsMeal, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for MasochistsMeal. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fear & Loathing In Las Vegas\n\nSCREENPLAY EXCERPT:\nINT. MATRIX MEN'S ROOM - NIGHT\n\nStill in slow motion, the YOUNGER DUKE is trying to eat a\nHUGE SPANSULE OF ACID. With difficulty.\n\n 50.\n\n\n DUKE (V/O)\n I decided to eat only half at first.\n Good thinking. But I spilled the\n rest on the sleeve of my red\n Pendleton shirt.\n\nDUKE stares at his sleeve, uncertain what to do. C/U of the\ndoor to the men's room as a MUSICIAN enters speaking in\nslow-motion.\n\n MUSICIAN\n What's the trouble?\n\n DUKE\n (also in slow-motion)\n Well, all this white stuff on my\n sleeve is LSD.\n\nThe MUSICIAN approaches and looks down at DUKE'S arm. A\nlong pause.\n\nCut back to tight shot of door as it opens and a very clean-\ncut, PREPPY, STOCKBROKER TYPE enters. He freezes in horror.\nWe cut to his POV. DUKE is standing in the middle of the\nmen's room with the MUSICIAN hunkered down at his side...\nsucking on his sleeve. A very gross tableau. The\nSTOCKBROKER slowly eases out of the room.\n\n DUKE (V/O)\n With a bit of luck his life was\n ruined -- forever thinking that\n just behind some narrow door in all\n his favorite bars, men in red\n Pendleton shirts are getting\n incredible kicks from things he'll\n never know."}, {"role": "assistant", "content": "## SCENE 1 — INT. MATRIX MEN'S ROOM - NIGHT\nACTION: Still in slow motion, the YOUNGER DUKE is trying to eat a HUGE SPANSULE OF ACID. With difficulty.\nDIALOGUE — Duke: I decided to eat only half at first. Good thinking. But I spilled the rest on the sleeve of my red Pendleton shirt.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fear & Loathing In Las Vegas: Still in slow motion, the YOUNGER DUKE is trying to eat a HUGE SPANSULE OF ACID. With difficulty., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Still in slow motion, the YOUNGER DUKE is trying to eat a HUGE SPANSULE OF ACID. With difficulty. The camera pushes in with small amplitude at slow speed. Duke (S1) says: [English] I decided to eat only half at first. Good thinking. But I spilled the rest on the sleeve of my red Pendleton shirt. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Timmy climbs tentatively. A dinosaur ROARS in the distance.\nSetting: derived from Jurassic Park\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Timmy climbs tentatively. A dinosaur ROARS in the distance. CLOSE ON - Timmy's foot loses its grip.\nDIALOGUE — Grant: A herbivore grazing area. Excellent.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jurassic Park: Timmy climbs tentatively. A dinosaur ROARS in the distance. CLOSE ON - Timmy's foot loses its grip., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Timmy climbs tentatively. A dinosaur ROARS in the distance. CLOSE ON - Timmy's foot loses its grip. The camera pushes in with small amplitude at slow speed. Grant (S1) says: [English] A herbivore grazing area. Excellent. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Final Destination 2\n\nSCREENPLAY EXCERPT:\nEXT. HIGHWAY ON RAMP - DAY\n\nON OLD WOMAN\n\nAs she pushes her supermarket cart up the on ramp. A SQUEAKY WHEEL is\nshrill and disturbing.\n\nDRIVERS of several waiting vehicles avoid staring at her weathered\nface as she passes. Up ahead waiting in line, is Kimberly's red SUV.\n\nUP AHEAD\n\nWaiting in line, is Kimberly's red SUV.\n\n\n SHAINA (O.S.)\n Watch it!\n\nKimberly accidentally KISSES the bumper in front of her while\ndaydreaming. She calls out to the driver in front:\n\n\n KIMBERLY\n Sorry! My fault!\n\nShaina looks over, concerned. In the backseat, the guys DANO and\nFRANKIE laugh hard and some of Frankie's milkshake comes out of his\nnose.\n\n\n SHAINA\n Want me to drive?\n\n KIMBERLY\n No, I'm good.\n\nShaina looks at Kimberly as a HAND suddenly hits Kimberly's window.\nKimberly spins around, startled.\n\nKimberly stares transfixed at the Old Woman, who stares back with\nchilling intensity.\n\nThe kids ad-lib \"Let's go\" and \"Fucking freak.\" The Old Woman's\nPLASTIC BAG BREAKS, sending oranges rolling down the on- ramp. The Old\n\nWoman breaks eye contact with Kimberly as she rushes to retrieve\nthem.\n\nThe signal turns from red to flashing yellow.\n\nKimberly pulls the SU V onto the highway, looking in the side view\nmirror to see the Old Woman look up at her in SLOW-MO.\n\n\n FRANKIE\n Dano, shouldn't we go back and help\n your mother?\n\n DANO\n Blo w me.\n\nCU ON KIMBERLY\n\nAs she "}, {"role": "assistant", "content": "## SCENE 1 — EXT. HIGHWAY ON RAMP - DAY\nACTION: As she pushes her supermarket cart up the on ramp. A SQUEAKY WHEEL is shrill and disturbing.\nDIALOGUE — Shaina: Watch it!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Final Destination 2: As she pushes her supermarket cart up the on ramp. A SQUEAKY WHEEL is shrill and disturbing., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. As she pushes her supermarket cart up the on ramp. A SQUEAKY WHEEL is shrill and disturbing. The camera pushes in with small amplitude at slow speed. Shaina (S1) says: [English] Watch it! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: As she pushes her supermarket cart up the on ramp. A SQUEAKY WHEEL is\nSetting: derived from Final Destination 2\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HIGHWAY ON RAMP - DAY\nACTION: As she pushes her supermarket cart up the on ramp. A SQUEAKY WHEEL is shrill and disturbing.\nDIALOGUE — Shaina: Watch it!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Final Destination 2: As she pushes her supermarket cart up the on ramp. A SQUEAKY WHEEL is shrill and disturbing., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. As she pushes her supermarket cart up the on ramp. A SQUEAKY WHEEL is shrill and disturbing. The camera pushes in with small amplitude at slow speed. Shaina (S1) says: [English] Watch it! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Transcribed by Dave, The Neutral Planet\nSetting: derived from Futurama\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HOARSE\nACTION: Transcribed by Dave, The Neutral Planet [Ships Rec Room. Fry and Bender play a futuristic version of\nDIALOGUE — Fry: Hmm. Horsey to pointy guy six. Check.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Futurama: Transcribed by Dave, The Neutral Planet [Ships Rec Room. Fry and Bender play a futuristic version of, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Transcribed by Dave, The Neutral Planet [Ships Rec Room. Fry and Bender play a futuristic version of The camera pushes in with small amplitude at slow speed. Fry (S1) says: [English] Hmm. Horsey to pointy guy six. Check. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BigTimeRush illustrating the trope 'IronyAsSheIsCast'. In the Big Time Rush episode Big Time Rides, the fictional James has to learn how to ride a bike, while the real-life James already knows how to.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying IronyAsSheIsCast.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BigTimeRush, IronyAsSheIsCast, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for IronyAsSheIsCast. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Flash Gordon\n\nSCREENPLAY EXCERPT:\nINT. PLASMA CAVERN\n (ZARKOV)\n\n HAWKMEN and LIZARD MEN SLAVES engage MING'S PEOPLE.\n\n Aided by VULTAN and TIGER MAN, ZARKOV battles his way up to\n the Mezzanine Control area. He shouts down wildly:\n 82.\n\n ZARKOV\n Attach the electrodes to the shovels!\n Pitch them into the plasma!"}, {"role": "assistant", "content": "## SCENE 1 — INT. PLASMA CAVERN\nACTION: HAWKMEN and LIZARD MEN SLAVES engage MING'S PEOPLE. Aided by VULTAN and TIGER MAN, ZARKOV battles his way up to\nDIALOGUE — Zarkov: Attach the electrodes to the shovels! Pitch them into the plasma!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Flash Gordon: HAWKMEN and LIZARD MEN SLAVES engage MING'S PEOPLE. Aided by VULTAN and TIGER MAN, ZARKOV battles his way up to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. HAWKMEN and LIZARD MEN SLAVES engage MING'S PEOPLE. Aided by VULTAN and TIGER MAN, ZARKOV battles his way up to The camera pushes in with small amplitude at slow speed. Zarkov (S1) says: [English] Attach the electrodes to the shovels! Pitch them into the plasma! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Minority Report\"\n\nSCREENPLAY EXCERPT:\nINT. JUDICIAL CENTER -- DAY\n\n Lisa stands in a long hallway outside a door, labeled:\n COURTROOM 17. She holds a briefcase in one hand. Down the\n hall, other courtrooms are in use, but not this one. She\n enters the dark and vacant Courtroom 17."}, {"role": "assistant", "content": "## SCENE 1 — INT. JUDICIAL CENTER -- DAY\nACTION: Lisa stands in a long hallway outside a door, labeled: COURTROOM 17. She holds a briefcase in one hand. Down the\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Minority Report\": Lisa stands in a long hallway outside a door, labeled: COURTROOM 17. She holds a briefcase in one hand. Down the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Lisa stands in a long hallway outside a door, labeled: COURTROOM 17. She holds a briefcase in one hand. Down the The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Batman 2\n\nSCREENPLAY EXCERPT:\nINT. TOP-FLOOR LANDING - THAT MOMENT\n\n The CATWOMAN's on a landing near the stairwell, just below the\n dormer that leads to the roof. She pries open a pair of\n ELEVATOR DOORS...\n\n ...and a FLOCK of BIRDS pours out of the empty shaft, making\n for the open ACCESS DOOR a half-story above. She reaches\n inside, finds the cord attached to the RAVEN BOX, and reels it\n in.\n\n She rips off the lock and opens the box for a quick look at\n her trophy. It's there, all right -- another RAVEN just like\n the first two. An awful CRASH, from twenty-five stories down,\n RATTLES THE SHAFT...\n\n She makes a disgusted face, grabs the RAVEN, and bolts for the\n roof."}, {"role": "assistant", "content": "## SCENE 1 — INT. TOP-FLOOR LANDING - THAT MOMENT\nACTION: The CATWOMAN's on a landing near the stairwell, just below the dormer that leads to the roof. She pries open a pair of\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman 2: The CATWOMAN's on a landing near the stairwell, just below the dormer that leads to the roof. She pries open a pair of, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The CATWOMAN's on a landing near the stairwell, just below the dormer that leads to the roof. She pries open a pair of The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The man enters the bedroom. We can see a woman lying on the\nSetting: derived from \"Rear Window\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEIGHBORHOOD - DAY - SEMI-LONG SHOT\nACTION: The man enters the bedroom. We can see a woman lying on the far bed. Near her, a small table is covered with medicine\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Rear Window\": The man enters the bedroom. We can see a woman lying on the far bed. Near her, a small table is covered with medicine, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The man enters the bedroom. We can see a woman lying on the far bed. Near her, a small table is covered with medicine The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: We follow the Olds as it cruises around the back of the building,\nSetting: derived from Blade\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BLADE'S OLDS/ABANDONED FACTORY - NIGHT\nACTION: We follow the Olds as it cruises around the back of the building, heading down a concrete loading ramp. At the bottom of the ramp, a\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade: We follow the Olds as it cruises around the back of the building, heading down a concrete loading ramp. At the bottom of the ramp, a, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We follow the Olds as it cruises around the back of the building, heading down a concrete loading ramp. At the bottom of the ramp, a The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Bottle Rocket\n\nSCREENPLAY EXCERPT:\nEXT. HAMBURGER PLACE. NIGHT\n\nAnthony and Bob go out the door. Dignan is walking over to a\nstation wagon. Little Richard is getting in.\n\n DIGNAN\n Little Richard. Wait a second. Hang\n on. I didn't mean to --\n\nLittle Richard opens his door hard into Dignan's legs, then\ngets out fast and takes off his shirt. Dignan tries to hit\nhim but doesn't connect. They grab onto each other and start\nknocking around. Another GUY gets out of the station wagon.\nTwo more GUYS rush out the door of the burger place. Anthony\nand Bob stand back, nervous.\n\n ANTHONY\n Let them fight.\n\n BOB\n Let them fight.\n\nThey all watch. Dignan keeps trying to punch Little Richard,\nbut he's hanging onto him too tight. They keep spinning\naround together, moving down the sidewalk.\n\n DIGNAN\n (calling out)\n Anthony.\n\nAnthony looks at the other guys then moves toward Dignan.\n\n ANTHONY\n OK. Break it up. Break it up.\n\nLittle Richard lets go of Dignan. They separate.\n\n DIGNAN\n I separated my shoulder.\n\nDignan is holding onto his arm. He kind of moans.\n\n ANTHONY\n OK. Hang on.\n\nAnthony grabs hold of Dignan's arm. Everybody's watching.\n\n DIGNAN\n Just pull straight up.\nAnthony pulls up hard on Dignan's arm. Dignan tries not to\nyell, then suddenly he's OK and relieved.\n\n ANTHONY\n Is it back"}, {"role": "assistant", "content": "## SCENE 1 — EXT. HAMBURGER PLACE. NIGHT\nACTION: Anthony and Bob go out the door. Dignan is walking over to a station wagon. Little Richard is getting in.\nDIALOGUE — Dignan: Little Richard. Wait a second. Hang on. I didn't mean to --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bottle Rocket: Anthony and Bob go out the door. Dignan is walking over to a station wagon. Little Richard is getting in., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Anthony and Bob go out the door. Dignan is walking over to a station wagon. Little Richard is getting in. The camera pushes in with small amplitude at slow speed. Dignan (S1) says: [English] Little Richard. Wait a second. Hang on. I didn't mean to -- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Batman\n\nSCREENPLAY EXCERPT:\nEXT. DECK - YACHT - DAY\n\n BRUCE's forty-foot cabin cruiser, aptly christened \"DIE\n FLEIDERMAUS.\" BRUCE and VICKI are on the deck, in chairs,\n soaking up sun, gazing off at the sailboats.\n\n VICKI\n Do you sail?\n\n BRUCE\n Too much work. I'm not really the\n physical type. -- Thank you, Alfred.\n\n ALFRED has just appeared from belowdecks with a tray of\n drinks for BRUCE and VICKI. VICKI watches as BRUCE reaches\n for his glass. His forearm looks like a thin layer of skin\n over braided telephone cables.\n\n VICKI\n You do a very convincing imitation.\n (sipping her drink)\n Mm, this is tasty. What's yours?\n\n He smiles, slides the drink over toward her, gestures for\n her to try a sip.\n\n VICKI (cont.)\n ... Ginger ale?\n\n BRUCE\n Two drinks and I start swinging from\n the rooftops.\n (beat)\n Tell me, Vicki. There's something\n I'm very curious about. What took\n you down to Corto Maltese?\n\n VICKI\n ... I guess I needed a change.\n\n BRUCE\n You were one of the most successful\n magazine photographers in the city.\n Everyone wanted you.\n\n VICKI\n Have you ever been to Corto Maltese?"}, {"role": "assistant", "content": "## SCENE 1 — EXT. DECK - YACHT - DAY\nACTION: BRUCE's forty-foot cabin cruiser, aptly christened \"DIE FLEIDERMAUS.\" BRUCE and VICKI are on the deck, in chairs,\nDIALOGUE — Vicki: Do you sail?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman: BRUCE's forty-foot cabin cruiser, aptly christened \"DIE FLEIDERMAUS.\" BRUCE and VICKI are on the deck, in chairs,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. BRUCE's forty-foot cabin cruiser, aptly christened \"DIE FLEIDERMAUS.\" BRUCE and VICKI are on the deck, in chairs, The camera pushes in with small amplitude at slow speed. Vicki (S1) says: [English] Do you sail? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Malignant\n\nSCREENPLAY EXCERPT:\nINT. HOME - DOWNSTAIRS HALLWAY - NIGHT\nMadison steps into the hallway, looking around. Turns on the\nlight. Sees blood on the floor. Her mouth drops open,\ntrying to process.\n MADISON\n Derek?\nA DARK FIGURE steps out of the living room into the hallway.\nThis is THE KILLER. We do not see the Killer's's face, only\na menacing SILHOUETTE that's almost shapeless, like the\ninverse of a human being. The Killer is holding a curved,\nviciously sharp AMPUTATION KNIFE.\nON MADISON\nFrozen in terror as she sees the FIGURE. She steps back.\nThe Killer steps forward. Madison BOLTS. The Killer gives\nchase, as Madison runs into --"}, {"role": "assistant", "content": "## SCENE 1 — INT. HOME - DOWNSTAIRS HALLWAY - NIGHT\nACTION: Madison steps into the hallway, looking around. Turns on the light. Sees blood on the floor. Her mouth drops open,\nDIALOGUE — Madison: Derek? A DARK FIGURE steps out of the living room into the hallway. This is THE KILLER. We do not see the Killer's's face, only a menacing SILHOUETTE that's...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Malignant: Madison steps into the hallway, looking around. Turns on the light. Sees blood on the floor. Her mouth drops open,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Madison steps into the hallway, looking around. Turns on the light. Sees blood on the floor. Her mouth drops open, The camera pushes in with small amplitude at slow speed. Madison (S1) says: [English] Derek? A DARK FIGURE steps out of the living room into the hallway. This is THE KILLER. We do not see the Killer's's face, only a menacing SILHOUETTE that's... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Please come inside Fraulein.\nSetting: derived from Django Unchained\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BIG HOUSE\nACTION: Please come inside Fraulein. She does, and just'as Lara Lee is to say something, Schultz says,\nDIALOGUE — Broomhilda: Yes. How do you know that? Dr.SCHULTZ Stands to reason who ever taught you German would also give you a German name. Can I pour you a glass of water,...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Django Unchained: Please come inside Fraulein. She does, and just'as Lara Lee is to say something, Schultz says,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Please come inside Fraulein. She does, and just'as Lara Lee is to say something, Schultz says, The camera pushes in with small amplitude at slow speed. Broomhilda (S1) says: [English] Yes. How do you know that? Dr.SCHULTZ Stands to reason who ever taught you German would also give you a German name. Can I pour you a glass of water,... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A free-standing telephone booth next to a fence-post in the\nSetting: derived from Asteroid City\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HIGHWAY. DAY\nACTION: A free-standing telephone booth next to a fence-post in the corner of a vacant plot between the gas station and the motel\nDIALOGUE — Houseboy: Zak residence.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Asteroid City: A free-standing telephone booth next to a fence-post in the corner of a vacant plot between the gas station and the motel, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A free-standing telephone booth next to a fence-post in the corner of a vacant plot between the gas station and the motel The camera pushes in with small amplitude at slow speed. Houseboy (S1) says: [English] Zak residence. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Jeff turns and looks up to Lisa with a grin.\nSetting: derived from \"Rear Window\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. JEFF'S APARTMENT - NIGHT - SEMI-CLOSEUP\nACTION: Jeff turns and looks up to Lisa with a grin. Well, she picked the most prosperous\nDIALOGUE — Jeff: Well, she picked the most prosperous looking one.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Rear Window\": Jeff turns and looks up to Lisa with a grin. Well, she picked the most prosperous, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jeff turns and looks up to Lisa with a grin. Well, she picked the most prosperous The camera pushes in with small amplitude at slow speed. Jeff (S1) says: [English] Well, she picked the most prosperous looking one. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: After-teri, you-a set-ta boom-a,\nSetting: derived from Inglourious Basterds\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: After-teri, you-a set-ta boom-a, five-o moment-o, you-a, fuck-o Pphisst.\nDIALOGUE — Hirschberg: Affirm-ato, affirm-ato.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Inglourious Basterds: After-teri, you-a set-ta boom-a, five-o moment-o, you-a, fuck-o Pphisst., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. After-teri, you-a set-ta boom-a, five-o moment-o, you-a, fuck-o Pphisst. The camera pushes in with small amplitude at slow speed. Hirschberg (S1) says: [English] Affirm-ato, affirm-ato. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Django Unchained\n\nSCREENPLAY EXCERPT:\nPlease drink.\n She absentmindedly obeys.\n Dr.SCHULTZ\n\n (.GERMAN).\n Now it's myself and our mutual friend's\n intention to take you away from here\n forever.\n\n BROOMHILDA"}, {"role": "assistant", "content": "## SCENE 1 — INT. .GERMAN\nACTION: She absentmindedly obeys. Now it's myself and our mutual friend's\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Django Unchained: She absentmindedly obeys. Now it's myself and our mutual friend's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. She absentmindedly obeys. Now it's myself and our mutual friend's The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Alien\n\nSCREENPLAY EXCERPT:\nThe commander normally has first\n publication rights.\n\n MELKONIS\n Maybe we could write it together.\n\n ROBY\n First thing I'm going to do when we\n get back is eat some biological\n food.\n\n MELKONIS\n What's the matter, you don't like\n this stuff?\n\n ROBY\n Tastes like something you'd feed a\n chicken to make it lay more eggs.\n\n STANDARD\n Oh it's okay. I've had better cag\n than this, but I've had worse too,\n if you know what I mean.\n\n FAUST\n I kind of like it.\n\n ROBY\n You like this shit?\n\n FAUST\n It grows on you.\n\n ROBY\n You know what they make this stuff\n out of?\n\n FAUST"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The commander normally has first Maybe we could write it together.\nDIALOGUE — Melkonis: Maybe we could write it together.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Alien: The commander normally has first Maybe we could write it together., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The commander normally has first Maybe we could write it together. The camera pushes in with small amplitude at slow speed. Melkonis (S1) says: [English] Maybe we could write it together. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: David X. Cohen & Matt Groening\nSetting: derived from Futurama\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. VOICE-OVER\nACTION: David X. Cohen & Matt Groening Transcribed by Dave, The Neutral Planet\nDIALOGUE — Man: (voice-over) Space. It seems to go on and on forever. But then you get to the end and the gorilla starts throwing barrels at you.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Futurama: David X. Cohen & Matt Groening Transcribed by Dave, The Neutral Planet, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. David X. Cohen & Matt Groening Transcribed by Dave, The Neutral Planet The camera pushes in with small amplitude at slow speed. Man (S1) says: [English] (voice-over) Space. It seems to go on and on forever. But then you get to the end and the gorilla starts throwing barrels at you. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Crank\n\nSCREENPLAY EXCERPT:\nINT. TV SCREEN\n 5A 5A\n It's RICKY VERONA on the SCREEN, a young, irritatingly slick\n EASTERN EURO ... little to no accent - fast talking,\n sarcastic, a complete dick ...\n\n\n\n (CONTINUED)\n\n\n\n 3.\n CONTINUED:\n\n\n We see him sitting on CHEV'S bed ... CHEV is visible in the\n frame, unconscious behind VERONA. Pale nicotine sunlight\n filters in through the blinds. There are HOODS loitering\n around the room. It was apparently shot only hours before.\n (\n The discernible sound of CHEV'S heartbeat will subliminally\n increase in speed and volume throughout VERONA'S monologue -\n the cell phone continues to ring, somewhere.)\n\n VERONA\n What's shaking, douchebag? Thought I'd\n give you the heads up. You're dead.\n\n On the TV - VERONA points into the CAMERA.\n\n VERONA (CONT'D)\n Yeah, that's right, you little bitch ...\n if you're watching this tape it means\n that I somehow resisted the urge to\n dismember you and shove the pieces down\n the g"}, {"role": "assistant", "content": "## SCENE 1 — INT. TV SCREEN\nACTION: 5A 5A It's RICKY VERONA on the SCREEN, a young, irritatingly slick\nDIALOGUE — Verona: What's shaking, douchebag? Thought I'd give you the heads up. You're dead.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Crank: 5A 5A It's RICKY VERONA on the SCREEN, a young, irritatingly slick, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. 5A 5A It's RICKY VERONA on the SCREEN, a young, irritatingly slick The camera pushes in with small amplitude at slow speed. Verona (S1) says: [English] What's shaking, douchebag? Thought I'd give you the heads up. You're dead. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Int. Publishing House Reception Area - Day\n\nSCREENPLAY EXCERPT:\nINT. ELEVATOR - DAY\n\nJoel gets off on the sixth floor. He searches for a room\nnumber. As he turns the corner, he sees that the hallway is\nfaded, vague and mostly erased. He keeps walking, comes to\nthe door marked Lacuna and opens it. Inside he can see\nvague, erased version of Mary the receptionist.\n\n\t\tMARY\n\t\t(dead monotone)\n\tHi, May I help you?\n\nThese degraded, faded memories allow Joel to detach himself\nand hear what's going on in his bedroom.\n\n\t\tSTAN'S VOICE\n\tSo, Mary's coming over tonight.\n\nJoel looks up."}, {"role": "assistant", "content": "## SCENE 1 — INT. ELEVATOR - DAY\nACTION: Joel gets off on the sixth floor. He searches for a room number. As he turns the corner, he sees that the hallway is\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int. Publishing House Reception Area - Day: Joel gets off on the sixth floor. He searches for a room number. As he turns the corner, he sees that the hallway is, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Joel gets off on the sixth floor. He searches for a room number. As he turns the corner, he sees that the hallway is The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: --------------------------------------------------------------------\nSetting: derived from Untitled\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: -------------------------------------------------------------------- At first we see an eye. Closed and slowly it opens up and looks\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: -------------------------------------------------------------------- At first we see an eye. Closed and slowly it opens up and looks, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. -------------------------------------------------------------------- At first we see an eye. Closed and slowly it opens up and looks The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Hackers\n\nSCREENPLAY EXCERPT:\nINT. GRAND CENTRAL STATION.\n\n The four hackers skate into Grand Central Station and head\n to the lowest level. They meet up with Joey.\n\n CEREAL\n Hey Joey, you made it!"}, {"role": "assistant", "content": "## SCENE 1 — INT. GRAND CENTRAL STATION.\nACTION: The four hackers skate into Grand Central Station and head to the lowest level. They meet up with Joey.\nDIALOGUE — Cereal: Hey Joey, you made it!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hackers: The four hackers skate into Grand Central Station and head to the lowest level. They meet up with Joey., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The four hackers skate into Grand Central Station and head to the lowest level. They meet up with Joey. The camera pushes in with small amplitude at slow speed. Cereal (S1) says: [English] Hey Joey, you made it! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Gloria drives, making crazy turns. She's an incredible\nSetting: derived from Barbie\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. GLORIA'S CAR.\nACTION: Gloria drives, making crazy turns. She's an incredible driver. Sasha in the passenger seat. Everyone is screaming.\nDIALOGUE — Sasha: God I hope nobody from school saw us put a life-size Barbie in our car. How did this even happen?!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Barbie: Gloria drives, making crazy turns. She's an incredible driver. Sasha in the passenger seat. Everyone is screaming., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Gloria drives, making crazy turns. She's an incredible driver. Sasha in the passenger seat. Everyone is screaming. The camera pushes in with small amplitude at slow speed. Sasha (S1) says: [English] God I hope nobody from school saw us put a life-size Barbie in our car. How did this even happen?! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fantastic Mr. Fox\n\nSCREENPLAY EXCERPT:\nINT. BUNCE'S KITCHEN. DAY\n\n Bunce sits on two stacked telephone books on a chair. He guts\n a dead goose, cutting out its liver and mashing it with a\n fork. A plate of doughnuts cools on the table.\n\n BADGER (V.0.)\n He eats only doughnuts with smashed-up\n goose livers injected into them.\n\n CUT TO:\n A tall, skinny man in a long trench-coat. He holds a Luger\n pistol. He stands in front of his farm, which is an apple\n orchard that stretches over thousands of acres. He has a mean\n face. He is Bean.\n\n BADGER (V.0.)\n Franklin Bean is a turkey and apple\n farmer. He keeps his birds in an orchard\n\n (MORE)\n\n\n\n\n\n\n\n\n\n 12.\n BADGER (V.0.) (cont'd)\n where they run around squawking and\n gobbling, surrounded by apples.\n Bean aims his Luger and shoots a humming bird. Crazy turkeys\n run about among the trees."}, {"role": "assistant", "content": "## SCENE 1 — INT. BUNCE'S KITCHEN. DAY\nACTION: Bunce sits on two stacked telephone books on a chair. He guts a dead goose, cutting out its liver and mashing it with a\nDIALOGUE — Badger: He eats only doughnuts with smashed-up goose livers injected into them.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fantastic Mr. Fox: Bunce sits on two stacked telephone books on a chair. He guts a dead goose, cutting out its liver and mashing it with a, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Bunce sits on two stacked telephone books on a chair. He guts a dead goose, cutting out its liver and mashing it with a The camera pushes in with small amplitude at slow speed. Badger (S1) says: [English] He eats only doughnuts with smashed-up goose livers injected into them. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Where'd you get this snotty old\nSetting: derived from A Nightmare On Elm Street\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Where'd you get this snotty old SNORES from ROD. TINA yawns, turns off the light and snuggles\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from A Nightmare On Elm Street: Where'd you get this snotty old SNORES from ROD. TINA yawns, turns off the light and snuggles, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Where'd you get this snotty old SNORES from ROD. TINA yawns, turns off the light and snuggles The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Mrs. Teasdale\n\nSCREENPLAY EXCERPT:\nThese are the laws of my\n administration: No one's allowed to\n smoke or tell a dirty joke -- And\n whistling is forbidden...\n\n ENSEMBLE"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: These are the laws of my administration: No one's allowed to\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Mrs. Teasdale: These are the laws of my administration: No one's allowed to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. These are the laws of my administration: No one's allowed to The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Alien\n\nSCREENPLAY EXCERPT:\nI guess I had it coming. Let's call\n it settled.\n\n After a hard stare at Roby, Standard gives him a curt nod and turns\n his attention to the machinery.\n\n ROBY (CONT'D)"}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: I guess I had it coming. Let's call After a hard stare at Roby, Standard gives him a curt nod and turns\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Alien: I guess I had it coming. Let's call After a hard stare at Roby, Standard gives him a curt nod and turns, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I guess I had it coming. Let's call After a hard stare at Roby, Standard gives him a curt nod and turns The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BuffyTheVampireSlayer illustrating the trope 'EverybodyDoTheEndlessLoop'. On Buffy the Vampire Slayer, Xander does the Snoopy's looped dance from Charlie Brown (as he did every year at Xmas when they were kids) to prove to Willow that he is the real Xander. She stops him after a minute, but given that this is Xander, he probably would have kept looping.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying EverybodyDoTheEndlessLoop.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BuffyTheVampireSlayer, EverybodyDoTheEndlessLoop, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for EverybodyDoTheEndlessLoop. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Must be my imagination.\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Must be my imagination. In CU she turns back toward the sink area when suddenly the\nDIALOGUE — Sandy: Anything else you want?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: Must be my imagination. In CU she turns back toward the sink area when suddenly the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Must be my imagination. In CU she turns back toward the sink area when suddenly the The camera pushes in with small amplitude at slow speed. Sandy (S1) says: [English] Anything else you want? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Sydney's car pulls into the driveway of Madison's house.\nSetting: derived from Malignant\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. MITCHELL HOME - DAY\nACTION: Sydney's car pulls into the driveway of Madison's house.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Malignant: Sydney's car pulls into the driveway of Madison's house., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sydney's car pulls into the driveway of Madison's house. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The sand, it's everywhere!\nSetting: derived from The Marine Biologist\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The sand, it's everywhere! OK I'll see you later.\nDIALOGUE — Kramer: The sand, it's everywhere!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Marine Biologist: The sand, it's everywhere! OK I'll see you later., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The sand, it's everywhere! OK I'll see you later. The camera pushes in with small amplitude at slow speed. Kramer (S1) says: [English] The sand, it's everywhere! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in HitlerTheRiseOfEvil illustrating the trope 'GanglandDriveBy'. Hitler: The Rise of Evil. The Purge montage includes a drive-by shooting of a man getting his milk delivery, by an SS officer using a Bergmann submachine gun.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying GanglandDriveBy.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for HitlerTheRiseOfEvil, GanglandDriveBy, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for GanglandDriveBy. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: GONZO throws up in the toilet bowl.\nSetting: derived from Fear & Loathing In Las Vegas\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HOTEL FLAMINGO SUITE - BATHROOM - NEXT MORNING\nACTION: GONZO throws up in the toilet bowl. In the background, DUKE opens curtains. Daylight blinds him.\nDIALOGUE — Duke: Come on, we're going to be late.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fear & Loathing In Las Vegas: GONZO throws up in the toilet bowl. In the background, DUKE opens curtains. Daylight blinds him., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GONZO throws up in the toilet bowl. In the background, DUKE opens curtains. Daylight blinds him. The camera pushes in with small amplitude at slow speed. Duke (S1) says: [English] Come on, we're going to be late. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Airplane\n\nSCREENPLAY EXCERPT:\nEXT. AIRPLANE - THROUGH COCKPIT WINDOW - NIGHT\n\n Elaine and automatic pilot are relaxed in their seats. Both\n are smoking cigarettes."}, {"role": "assistant", "content": "## SCENE 1 — EXT. AIRPLANE - THROUGH COCKPIT WINDOW - NIGHT\nACTION: Elaine and automatic pilot are relaxed in their seats. Both are smoking cigarettes.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Airplane: Elaine and automatic pilot are relaxed in their seats. Both are smoking cigarettes., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Elaine and automatic pilot are relaxed in their seats. Both are smoking cigarettes. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Blade and Karen enter the narrow, maze-like \"stacks\" of the vampire\nSetting: derived from Blade\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. THE VAMPIRE ARCHIVES, HALLWAY - NIGHT\nACTION: Blade and Karen enter the narrow, maze-like \"stacks\" of the vampire archives, weaving their way through shelves of climate-controlled,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade: Blade and Karen enter the narrow, maze-like \"stacks\" of the vampire archives, weaving their way through shelves of climate-controlled,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Blade and Karen enter the narrow, maze-like \"stacks\" of the vampire archives, weaving their way through shelves of climate-controlled, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Jeff's eyes drop from the ballet dancer's room to the one\nSetting: derived from \"Rear Window\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. APARTMENT - DAY - CLOSEUP\nACTION: Jeff's eyes drop from the ballet dancer's room to the one\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Rear Window\": Jeff's eyes drop from the ballet dancer's room to the one, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jeff's eyes drop from the ballet dancer's room to the one The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: BRUCE's forty-foot cabin cruiser, aptly christened \"DIE\nSetting: derived from Batman\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. DECK - YACHT - DAY\nACTION: BRUCE's forty-foot cabin cruiser, aptly christened \"DIE FLEIDERMAUS.\" BRUCE and VICKI are on the deck, in chairs,\nDIALOGUE — Vicki: Do you sail?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman: BRUCE's forty-foot cabin cruiser, aptly christened \"DIE FLEIDERMAUS.\" BRUCE and VICKI are on the deck, in chairs,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. BRUCE's forty-foot cabin cruiser, aptly christened \"DIE FLEIDERMAUS.\" BRUCE and VICKI are on the deck, in chairs, The camera pushes in with small amplitude at slow speed. Vicki (S1) says: [English] Do you sail? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in CriminalMinds illustrating the trope 'DanBrowned'. Criminal Minds has made some pretty big goofs. In one example, one of the characters claimed the Principle of Double Effect is essentially an argument that the ends justifies the means. Which is completely wrong from start to finish.\n The Spin-Off Criminal Minds: Beyond Borders was advertised as researching the foreign locales to the point of making viewers believe the show was really shot in them...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying DanBrowned.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for CriminalMinds, DanBrowned, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for DanBrowned. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Austin and Vanessa are seated at a table with Frau. WAITERS\nSetting: derived from Austin Powers: International Man Of Mystery\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. DR. EVIL'S PRIVATE QUARTERS\nACTION: Austin and Vanessa are seated at a table with Frau. WAITERS MUSIC: Sexy Matt Helm-type theme\nDIALOGUE — Dr. Evil: I think you'll enjoy the food. I have the best chef in the world. His name is Ezekial. He's made of seventy-five percent plastic.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Austin Powers: International Man Of Mystery: Austin and Vanessa are seated at a table with Frau. WAITERS MUSIC: Sexy Matt Helm-type theme, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Austin and Vanessa are seated at a table with Frau. WAITERS MUSIC: Sexy Matt Helm-type theme The camera pushes in with small amplitude at slow speed. Dr. Evil (S1) says: [English] I think you'll enjoy the food. I have the best chef in the world. His name is Ezekial. He's made of seventy-five percent plastic. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Isabella is pushing again.\nSetting: derived from Final Destination 2\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. DELIVERY ROOM - CONTINUOUS\nACTION: Isabella is pushing again. Almost there. I see the head...\nDIALOGUE — Obstetrician: Almost there. I see the head...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Final Destination 2: Isabella is pushing again. Almost there. I see the head..., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Isabella is pushing again. Almost there. I see the head... The camera pushes in with small amplitude at slow speed. Obstetrician (S1) says: [English] Almost there. I see the head... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Joel trudges along carrying two big trash bags full of stuff.\nSetting: derived from Int. Publishing House Reception Area - Day\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEW YORK STREET - DAY\nACTION: Joel trudges along carrying two big trash bags full of stuff. He's been crying. He looks behind him and finds himself\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int. Publishing House Reception Area - Day: Joel trudges along carrying two big trash bags full of stuff. He's been crying. He looks behind him and finds himself, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Joel trudges along carrying two big trash bags full of stuff. He's been crying. He looks behind him and finds himself The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dune\n\nSCREENPLAY EXCERPT:\nINT. EMPEROR'S SPACE SHIP - GREEN THRONE ROOM - DAY\n\n Suddenly the wall behind Alia and the spinning Baron\n is completely blown away by a BLAST.\n\n The Baron floats and spins out into the storm.\n\n274. EXT. ARRAKEEN - DAY\n\n As the worms enter the Arrakeen basin they meet the\n Sardaukar. The Sardaukar don't have a chance -- the\n worms suck them up by the hundreds. The Fremen on\n the worm's backs fire weapons, killing the few\n Sardaukar the worms leave behind.\n\n275. EXT. LANDING FIELD - ARRAKEEN - DAY\n\n The space ships which were recently airborne are now\n crashing because of the storm.\n--\n272A. NEW SCENE\n\n The Baron floats and spins high in the dust clouds\n of the storm over the Arrakeen Valley.\n\n274A. NEW SCENE\n\n The worm, that Paul, Gurney and Stilgar are riding,\n suddenly rears up. Its mouth opens. The Baron\n appears as a tiny spinning shape. He flies directly\n into the mouth of this collosal worm and is\n devoured.\n\n277. EXT. STEEL TENT - LANDING FIELD - ARRAKEEN - DAY\n\n A thousand Fremen battle Sardaukar outside the\n Emperor's tent. They soon overpower them and rush\n into the giant steel structure.\n\n278. EXT. LANDING FIELD - ARRAKEEN - DAY\n\n The entire airfield is now filled with wild worms,\n Fremen, and dead or dying Sardaukar.\n\n276. EXT. LANDING FIELD - ARRAKEEN - DAY\n\n Alia moves dreamily among the dying Sardaukar.\n\n279. through 280. (Deleted)\n\n281. INT. GREAT HALL - ARRAKEEN PALACE - DAY\n\n Paul enters the Palace hall and the first thing he\n sees is the gi"}, {"role": "assistant", "content": "## SCENE 1 — INT. EMPEROR'S SPACE SHIP - GREEN THRONE ROOM - DAY\nACTION: Suddenly the wall behind Alia and the spinning Baron is completely blown away by a BLAST.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dune: Suddenly the wall behind Alia and the spinning Baron is completely blown away by a BLAST., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Suddenly the wall behind Alia and the spinning Baron is completely blown away by a BLAST. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A DOOR MARKED, \"DR. HENRY PYM,\" flies open. Hank exits,\nSetting: derived from Avengers: Endgame\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CAMP LEHIGH, SHIELD FACILITY, HALLWAY - DAY (1970)\nACTION: A DOOR MARKED, \"DR. HENRY PYM,\" flies open. Hank exits, fuming, brushing past...STEVE.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Avengers: Endgame: A DOOR MARKED, \"DR. HENRY PYM,\" flies open. Hank exits, fuming, brushing past...STEVE., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A DOOR MARKED, \"DR. HENRY PYM,\" flies open. Hank exits, fuming, brushing past...STEVE. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Saw\n\nSCREENPLAY EXCERPT:\nDon�t...Don�t worry, I�ll bring someone\n back, I promise. (He continues crawling\n on, out the door while Adam watches\n him pathetically, still crying and reaching\n out for him.)\n\n Lawrence makes it out the door, and starts dragging himself down\n the hall. Adam is now alone in the room with two corpses. Adam\n looks to the door one more time then turns to Zep�s body. He\n pulls himself on top of him, wincing at his shoulder pain.\n\n\n ADAM\n (thinking) Key...key...\n\n He starts going through Zep�s pockets, patting him down. He finds\n Zep�s wallet, opens it and feels through it; there� s nothing\n of value to him right now. He tosses it down. He continues feeling\n for something, suddenly feels something in Zerp�s coat pocket.\n His eyes widen as he reaches in, wincing from the pain in his\n shoulder. He pulls out a tape recorder. He stares at it, his\n eyes and mouth starting to widen. He glances down at Zep, then\n at the tape recorder. Sitting up a bit more, he presses play.\n It is Jigsaw�s voice.\n\n JIGSAW\n (VO) Hello Mr. Hindle. Or as they called\n you around the hospital: (shot of Zep\n in the hospital with his cleaning cart,\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. THINKING\nACTION: Don�t...Don�t worry, I�ll bring someone back, I promise. (He continues crawling\nDIALOGUE — Adam: (thinking) Key...key...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Saw: Don�t...Don�t worry, I�ll bring someone back, I promise. (He continues crawling, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Don�t...Don�t worry, I�ll bring someone back, I promise. (He continues crawling The camera pushes in with small amplitude at slow speed. Adam (S1) says: [English] (thinking) Key...key... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: JACOB and MICHAEL are walking through the STREETS OF GREENWICH VILLAGE.\nSetting: derived from \"Jacob'S Ladder\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. GREENWICH VILLAGE - DAY\nACTION: JACOB and MICHAEL are walking through the STREETS OF GREENWICH VILLAGE. It is early MORNING and the sidewalks are bustling with PEOPLE. JACOB\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Jacob'S Ladder\": JACOB and MICHAEL are walking through the STREETS OF GREENWICH VILLAGE. It is early MORNING and the sidewalks are bustling with PEOPLE. JACOB, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. JACOB and MICHAEL are walking through the STREETS OF GREENWICH VILLAGE. It is early MORNING and the sidewalks are bustling with PEOPLE. JACOB The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Ext. Manhattan Island - Day\n\nSCREENPLAY EXCERPT:\nEXT. ARMAND'S RESTAURANT - NIGHT (CONTINUOUS ACTION)\n\nThe police car is parked right behind Ecto-2. Spengler stops at the\npolice car and confronts the cops.\n\n\t\t\t\t\t\tSPENGLER\n\t\t\tLook, we're not drunk and we're not crazy.\n\t\t\tThis is a matter of vital importance.\n\nVenkman steps in and looks at the policemen's nametags.\n\n\t\t\t\t\t\tCOP\n\t\t\tWhat are you doing?\n\n\t\t\t\t\t\tVENKMAN\n\t\t\tI just want to get your names right for when\n\t\t\tthe mayor asks us why we didn't let him know\n\t\t\tabout this sooner.\n\nThe Cops look at each other, uncertain about what to do.\n\n\t\t\t\t\t\tCOP\n\t\t\t\t (relenting)\n\t\t\tOkay, Doc. You want to see the mayor, you\n\t\t\tgot it. Follow us.\n\nThey head for their respective vehicles.\n\n\t\t\t\t\t\t\t\t\t\t\t\t CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. ARMAND'S RESTAURANT - NIGHT (CONTINUOUS ACTION)\nACTION: The police car is parked right behind Ecto-2. Spengler stops at the police car and confronts the cops.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ext. Manhattan Island - Day: The police car is parked right behind Ecto-2. Spengler stops at the police car and confronts the cops., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The police car is parked right behind Ecto-2. Spengler stops at the police car and confronts the cops. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Batman 2\n\nSCREENPLAY EXCERPT:\nEXT. GOTHAM SQUARE - LATER THAT NIGHT\n\n Two hours later, the SNOWSTORM's grown into a full-fledged\n blizzard. The drifts are two feet deep and the streets are\n all but empty of cars. A massive SNOW PLOW the size of a\n Panzer tank rumbles past, FILLING THE FRAME...\n\n ...and revealing, as it passes, a group of CAROLERS, all\n bundled up in mufflers and parkas. Unbothered by the weather,\n they walk the street singing, spreading cheer and goodwill to\n the few passersby.\n\n They've just gone into a lovely a capella rendition of \"Silent\n Night\" when an oversized DELIVERY TRUCK, outfitted with snow\n chains, clanks slowly past in the wake of the plow. Its sides\n are decorated with cartoon igloos advertising a popular ice\n cream snack -- POLAR BARS -- oddly inappropriate for this time\n of year.\n\n A streetcorner SANTA, with bell and bucket, WAVES at the truck\n as it rounds a corner. The CAROLERS carol. Then --\n\n -- a VIOLENT EXPLOSION rocks the street -- followed by the\n sound of AUTOMATIC GUNFIRE and a high, wailing SECURITY ALARM."}, {"role": "assistant", "content": "## SCENE 1 — EXT. GOTHAM SQUARE - LATER THAT NIGHT\nACTION: Two hours later, the SNOWSTORM's grown into a full-fledged blizzard. The drifts are two feet deep and the streets are\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman 2: Two hours later, the SNOWSTORM's grown into a full-fledged blizzard. The drifts are two feet deep and the streets are, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Two hours later, the SNOWSTORM's grown into a full-fledged blizzard. The drifts are two feet deep and the streets are The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in KamenRiderGhost illustrating the trope 'MonkeyMoralityPose'. Kamen Rider Ghost has this as the Finishing Move for Sanzo Damashii.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying MonkeyMoralityPose.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for KamenRiderGhost, MonkeyMoralityPose, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for MonkeyMoralityPose. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Pleasantville\n\nSCREENPLAY EXCERPT:\nEXT. SODA SHOP.\n\n\tBiff's car is just pulling onto the street when David sprints\n\tinto the parking lot. He takes off after the car, sprinting\n\tdown the street.\n\n\t\t\t\tDAVID\n\t\tYOU CAN'T DO THIS JENNIFER! HE DOESN'T\n\t\tEXIST! YOU CAN'T DO THIS TO SOMEONE WHO\n\t\tDOESN'T EXIST!\n\n\tThe tail lights recede in the distance as David slows to a\n\tjog and then finally a walk. He pauses, winded, in the\n\tstreet, then looks up toward Heaven with a pissed off look on\n\this face.\n\n\t\t\t\tDAVID (CONT)\n\t\t\t(to God)\n\t\tThanks a lot.\n\n\t\t\t\t\t\t\t\tCUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. SODA SHOP.\nACTION: Biff's car is just pulling onto the street when David sprints into the parking lot. He takes off after the car, sprinting\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Pleasantville: Biff's car is just pulling onto the street when David sprints into the parking lot. He takes off after the car, sprinting, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Biff's car is just pulling onto the street when David sprints into the parking lot. He takes off after the car, sprinting The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in KamenRiderOOO illustrating the trope 'SnakeCharmer'. Kamen Rider OOO's movie-exclusive form, Burakawani Combo, has a molded cobra sculpted into its helmet; by using a pungi flute he can bring the cobra to life and command it to attack his enemies.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying SnakeCharmer.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for KamenRiderOOO, SnakeCharmer, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for SnakeCharmer. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Hunter\"\n\nSCREENPLAY EXCERPT:\n------------------------------------------------------------------------\n 76\n\n\n\n 120 CONTINUED: 120\n\n VOICE"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: ------------------------------------------------------------------------ 120 CONTINUED: 120\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hunter\": ------------------------------------------------------------------------ 120 CONTINUED: 120, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ------------------------------------------------------------------------ 120 CONTINUED: 120 The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The Secretary of Defense still on the floor, spotting the OVERTHRUSTER\nSetting: derived from Buckaroo Banzai\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. THE PIT - NIGHT\t376\nACTION: The Secretary of Defense still on the floor, spotting the OVERTHRUSTER in Penny's purse, taking it just as Buckaroo and New Jersey rush into\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Buckaroo Banzai: The Secretary of Defense still on the floor, spotting the OVERTHRUSTER in Penny's purse, taking it just as Buckaroo and New Jersey rush into, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The Secretary of Defense still on the floor, spotting the OVERTHRUSTER in Penny's purse, taking it just as Buckaroo and New Jersey rush into The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Castle illustrating the trope 'ClosetShuffle'. In the season 5 opener of Castle, the titular novelist has just had passionate sex with Kate Beckett... when his mother arrives. Poor woman gets shoved in the wardrobe while Castle tries to stall his mother... and then realises that she's left her bra by his bed.\n To make it funnier, it's later revealed his mother knew she was there all along!\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying ClosetShuffle.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Castle, ClosetShuffle, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for ClosetShuffle. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Hey, I just thought of a really funny\nSetting: derived from He Phone Message\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Hey, I just thought of a really funny thing for your act. Alright,\nDIALOGUE — Jerry: I don't think so.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from He Phone Message: Hey, I just thought of a really funny thing for your act. Alright,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hey, I just thought of a really funny thing for your act. Alright, The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] I don't think so. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Matrix\n\nSCREENPLAY EXCERPT:\nINT. HALL (MATRIX) - DAY\n\n\tAgent Brown hears Smith on his earphone.\n\n\t\t\t\t\tAGENT BROWTJ\n\t\t\tEighth floor! Move!"}, {"role": "assistant", "content": "## SCENE 1 — INT. HALL (MATRIX) - DAY\nACTION: Agent Brown hears Smith on his earphone.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Matrix: Agent Brown hears Smith on his earphone., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Agent Brown hears Smith on his earphone. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You know, smoking's bad for you.\nSetting: derived from 1 Ext. Cityscape Sunset 1\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You know, smoking's bad for you. It's my fucking life.\nDIALOGUE — Jill: It's my fucking life.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from 1 Ext. Cityscape Sunset 1: You know, smoking's bad for you. It's my fucking life., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You know, smoking's bad for you. It's my fucking life. The camera pushes in with small amplitude at slow speed. Jill (S1) says: [English] It's my fucking life. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Suicide\n\nSCREENPLAY EXCERPT:\nJERRY\n Boy she's sexy isn't she?\n\n\n (Jerry leaves garbage bag by Kramer's apartment and knocks on\n the door. As he and George enter Jerry's apartment Kramer comes\n out and takes the garbage bag.)\n\n\n JERRY\n Do you believe that guy?\n\n\n ELAINE\n What guy?\n\n\n JERRY\n My neighbour:\n\n\n ELAINE\n Oh, that creepy guy?\n\n\n JERRY\n Yeah, did he think I was flirting with\n her?\n\n\n GEORGE\n He didn't seem too pleased.\n\n\n ELAINE\n Maybe I'll get a steak with french fried\n onion wings.\n\n\n GEORGE\n hey, you know what? I just remembered\n something. I had a dream about that\n guy last night. This is amazing.\n\n\n JERRY\n What's so amazing? You've seen him before.\n\n\n\n GEORGE\n I haven't seen him for months.\n\n\n JERRY\n What was the dream?\n\n\n GEORGE\n I was doing standup comedy in Kennebunkport\n Maine. ??? night club. The stage was\n on a cliff and the audience was throwing\n all the comics off.\n\n\n JERRY\n I think I've played "}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Boy she's sexy isn't she? (Jerry leaves garbage bag by Kramer's apartment and knocks on\nDIALOGUE — Jerry: Do you believe that guy?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Suicide: Boy she's sexy isn't she? (Jerry leaves garbage bag by Kramer's apartment and knocks on, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Boy she's sexy isn't she? (Jerry leaves garbage bag by Kramer's apartment and knocks on The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] Do you believe that guy? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BreakingBad illustrating the trope 'IAmTheTrope'. In one of Breaking Bad's most iconic scenes, formerly mild-mannered Walt delivers a shocking Badass Boast to wife Skyler. All together now:\nWalt: I am not in danger, Skyler. I am the danger! A guy opens his door and gets shot and you think that of me? No. I am the one who knocks!\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying IAmTheTrope.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BreakingBad, IAmTheTrope, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for IAmTheTrope. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: He Phone Message\n\nSCREENPLAY EXCERPT:\nGEORGE\n Hey! What are you doing here? I thought\n I was supposed to call you\n\n\n when I got home.\n\n\n GEORGE\n I, I couldn't wait. I was too anxious\n to see you.\n\n\n CAROL\n Oh, that's so sweet.\n\n\n GEORGE\n Oh, this is my friend, Jerry Seinfeld.\n I just bumped into him\n\n\n around the corner. Isn't that a coincidence? The funny thing\n is, I see him all\n\n\n the time.\n\n\n JERRY\n All the time.\n\n\n CAROL\n It's nice to meet you.\n\n\n JERRY\n Hi.\n\n\n CAROL\n So, I'm starving. Where are we gonna\n eat?\n\n\n GEORGE\n You know, we could go uptown, and that\n way we could give Jerry a\n\n\n ride home.\n\n\n CAROL\n OK. Let's go, I'm ready, where'd you\n park?\n\n\n GEORGE\n Don't you wanna go upstairs first?\n\n\n\n CAROL\n No, what for? I'll just give my bag\n to the doorman.\n\n\n JERRY\n You know, I really need to use the bathroom.\n\n\n\n CAROL\n Oh well there's a bathroom in the coffee\n shop just ne"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Hey! What are you doing here? I thought I was supposed to call you\nDIALOGUE — George: I, I couldn't wait. I was too anxious to see you.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from He Phone Message: Hey! What are you doing here? I thought I was supposed to call you, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hey! What are you doing here? I thought I was supposed to call you The camera pushes in with small amplitude at slow speed. George (S1) says: [English] I, I couldn't wait. I was too anxious to see you. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: F I G H T C L U B\n\nSCREENPLAY EXCERPT:\nINT. KITCHEN -- MORNING\n\n Jack has his briefcase on the table, looks at PAPERWORK.\n Tyler wanders in, carries a dirty pot to the sink. Jack\n takes out a cigarette, lights up. He offers the pack...\n\n TYLER\n No thanks, I quit.\n\n JACK\n You quit?\n\n TYLER\n Yeah. Where you headed?\n\n JACK\n Work. Going to work.\n\n Tyler scratches his chin absently.\n\n JACK\n What... ?\n\n TYLER\n Nothing. Do what you like.\n\n Tyler walks out the way he came."}, {"role": "assistant", "content": "## SCENE 1 — INT. KITCHEN -- MORNING\nACTION: Jack has his briefcase on the table, looks at PAPERWORK. Tyler wanders in, carries a dirty pot to the sink. Jack\nDIALOGUE — Tyler: No thanks, I quit.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from F I G H T C L U B: Jack has his briefcase on the table, looks at PAPERWORK. Tyler wanders in, carries a dirty pot to the sink. Jack, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack has his briefcase on the table, looks at PAPERWORK. Tyler wanders in, carries a dirty pot to the sink. Jack The camera pushes in with small amplitude at slow speed. Tyler (S1) says: [English] No thanks, I quit. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Austin Powers: International Man Of Mystery\n\nSCREENPLAY EXCERPT:\nINT. DR. EVIL'S PRIVATE QUARTERS\n\n DR. EVIL\n (face again unseen)\n Ladies and Gentlemen, it's been a\n long time, but I'm back. It's all\n gone perfectly to plan except for\n one small flaw. Because of a\n technical error, my right arm was\n not frozen. I was therefore by\n definition only partially frozen.\n\n ANGLE ON EVIL ASSOCIATE MUSTAFA. He is terrified and sweaty,\n eyes darting left and right.\n\n MUSTAFA\n But my design was perfect! Your\n autonomic functions were shut down,\n and even though your arm wasn't\n frozen, the aging was retarded,\n therefore your right arm is only\n slightly older than the left.\n\n DR. EVIL\n Can't you see I'm only half a man?\n Look at me, I'm a freak!\n\n He holds up his older right arm, which looks normal.\n\n MUSTAFA\n But Dr. Evil, all you need to do\n is&emdash;\n (holding up tennis\n ball)\n --work with this tennis ball. Squeeze\n it for "}, {"role": "assistant", "content": "## SCENE 1 — INT. DR. EVIL'S PRIVATE QUARTERS\nACTION: Ladies and Gentlemen, it's been a long time, but I'm back. It's all\nDIALOGUE — Dr. Evil: Ladies and Gentlemen, it's been a long time, but I'm back. It's all gone perfectly to plan except for one small flaw. Because of a technical error, my right...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Austin Powers: International Man Of Mystery: Ladies and Gentlemen, it's been a long time, but I'm back. It's all, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Ladies and Gentlemen, it's been a long time, but I'm back. It's all The camera pushes in with small amplitude at slow speed. Dr. Evil (S1) says: [English] Ladies and Gentlemen, it's been a long time, but I'm back. It's all gone perfectly to plan except for one small flaw. Because of a technical error, my right... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Total Recall\"\n\nSCREENPLAY EXCERPT:\nJust a way of putting me on\n ice. He'd of reversed it\n once the planet was evacuated.\n\n BENNIE\n Man! This is real complicated.\n Go over it again, real slow.\n\n\n119 INT. VEHICLE - DAY\n\n QUAIL\n (ignoring Bennie's\n aside)\n What he hadn't counted on was\n your effect on me.\n\n MELINA"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Just a way of putting me on ice. He'd of reversed it\nDIALOGUE — Bennie: Man! This is real complicated. Go over it again, real slow.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Total Recall\": Just a way of putting me on ice. He'd of reversed it, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Just a way of putting me on ice. He'd of reversed it The camera pushes in with small amplitude at slow speed. Bennie (S1) says: [English] Man! This is real complicated. Go over it again, real slow. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Empire Strikes Back\n\nSCREENPLAY EXCERPT:\nEXT. BESPIN SYSTEM - MILLENNIUM FALCON - DAWN\n\nThe powerful pirate starship blasts through space as it heads toward\nthe soft pink planet of Bespin."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BESPIN SYSTEM - MILLENNIUM FALCON - DAWN\nACTION: The powerful pirate starship blasts through space as it heads toward the soft pink planet of Bespin.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Empire Strikes Back: The powerful pirate starship blasts through space as it heads toward the soft pink planet of Bespin., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The powerful pirate starship blasts through space as it heads toward the soft pink planet of Bespin. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Lisa stands in a long hallway outside a door, labeled:\nSetting: derived from \"Minority Report\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. JUDICIAL CENTER -- DAY\nACTION: Lisa stands in a long hallway outside a door, labeled: COURTROOM 17. She holds a briefcase in one hand. Down the\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Minority Report\": Lisa stands in a long hallway outside a door, labeled: COURTROOM 17. She holds a briefcase in one hand. Down the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Lisa stands in a long hallway outside a door, labeled: COURTROOM 17. She holds a briefcase in one hand. Down the The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: An old circular clothesline turns like a pinwheel in the wind. In the center of the yard is an oblong, tarped above ground pool. In the furthest corner of the yard, her dog is chained to a tree, BARKING AND FREAKED OUT. An old tire swing rocks in the storm. From the manner in which the house is situated and the length of the snapped power lines, one is capable of reaching the backyard.\nSetting: derived from \"Final Destination\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BACKYARD - CLEAR RIVER'S HOUSE - NIGHT - CLEAR'S POV\nACTION: An old circular clothesline turns like a pinwheel in the wind. In the center of the yard is an oblong, tarped above ground pool. In the furthest corner of the yard, her dog is chained to a tree, BARKING AND FREAKED OUT. An old tire swing rocks in the storm. From the manner in.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Final Destination\": An old circular clothesline turns like a pinwheel in the wind. In the center of the yard is an oblong, tarped above ground pool. In the furthest corner of the yard, her dog is chained to a tree, BARKI, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. An old circular clothesline turns like a pinwheel in the wind. In the center of the yard is an oblong, tarped above ground pool. In the furthest corner of the yard, her dog is chained to a tree, BARKING AND FREAKED OUT. An old tire swing rocks in the storm. From the manner in. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Empire Strikes Back\n\nSCREENPLAY EXCERPT:\nEXT. HOTH - ICE PLAIN - BATTLEFIELD\n\nA speeder banks through and away from the legs of a walker. Two other\nspeeders pass the first speeder from the opposite direction. Other\nRebel craft race just above the icy plain.\n\nA giant walker head swivels and fires, striking a snowspeeder and\nsending it crashing in a ball of flames."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HOTH - ICE PLAIN - BATTLEFIELD\nACTION: A speeder banks through and away from the legs of a walker. Two other speeders pass the first speeder from the opposite direction. Other\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Empire Strikes Back: A speeder banks through and away from the legs of a walker. Two other speeders pass the first speeder from the opposite direction. Other, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A speeder banks through and away from the legs of a walker. Two other speeders pass the first speeder from the opposite direction. Other The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: -- because I don't want anything\nSetting: derived from Ninja Assassin\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: -- because I don't want anything He glances up at her, and it's clear: he has affection\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ninja Assassin: -- because I don't want anything He glances up at her, and it's clear: he has affection, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. -- because I don't want anything He glances up at her, and it's clear: he has affection The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Obviously, Gail whatever his\nSetting: derived from \"Die Hard\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Obviously, Gail whatever his affiliation, it's safe to say\nDIALOGUE — Holly: Where are you going?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Die Hard\": Obviously, Gail whatever his affiliation, it's safe to say, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Obviously, Gail whatever his affiliation, it's safe to say The camera pushes in with small amplitude at slow speed. Holly (S1) says: [English] Where are you going? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: (To Kramer) Yeah...(in the phone) Yeah,\nSetting: derived from Good News, Bad News\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ENTERS\nACTION: (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the\nDIALOGUE — Jerry: (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the big trucks out on the street? Yeah, no problem (hangs up the phone).\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Good News, Bad News: (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the big trucks out on the street? Yeah, no problem (hangs up the phone). The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Rear Window\"\n\nSCREENPLAY EXCERPT:\nINT. JEFF'S APARTMENT - NIGHT - SEMI-CLOSEUP\n\n Jeff turns and looks up to Lisa with a grin.\n\n JEFF\n Well, she picked the most prosperous\n looking one.\n\n LISA\n She's not in love with him -- or any\n of them.\n\n JEFF\n How can you tell that -- from here?\n\n LISA\n You said it resembled my apartment,\n didn't you?\n\n She moves away with a significant look to him. THE CAMERA\n MOVES IN until Jeff is in semi-closeup, alone.\n\n He ponders over her last remark, then changes his look to\n another direction."}, {"role": "assistant", "content": "## SCENE 1 — INT. JEFF'S APARTMENT - NIGHT - SEMI-CLOSEUP\nACTION: Jeff turns and looks up to Lisa with a grin. Well, she picked the most prosperous\nDIALOGUE — Jeff: Well, she picked the most prosperous looking one.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Rear Window\": Jeff turns and looks up to Lisa with a grin. Well, she picked the most prosperous, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jeff turns and looks up to Lisa with a grin. Well, she picked the most prosperous The camera pushes in with small amplitude at slow speed. Jeff (S1) says: [English] Well, she picked the most prosperous looking one. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Buckaroo Banzai\n\nSCREENPLAY EXCERPT:\nINT. SPEED OPERATOR'S PULPIT (YOYODYNE) - NIGHT\t318\n\nMeanwhile in a grimy narrow control booth high above the factory floor,\na terrified Penny Priddy sits BOUND and GAGGED, John O'Connor SPREADING\nTHE CONTENTS FROM A HONEY-BEAR SQUEEZE BOTTLE over her naked arms while\nlong trails of ORDINARY PICNIC ANTS make their way up her legs...John\nBigboot� looking on...\n\nPenny's UNOPENED big clear plastic purse on Doctor Lizardo's desk, the\nOVERTHRUSTER right under his nose and he doesn't even know it as he\nspeaks into the phone handed to him by an ASSISTANT. BEGIN\nINTERCUTTING.\n\nDOCTOR LIZARDO\nWell, well...Dr. Banzai. Perhaps you don't remember me? Ah, I'm\nflattered. We know the same people. In fact, one of them is here with\nme now...your associate, Doctor Penny Priddy.\n\n\t319\nBUCKAROO BANZAI\n\"Doctor\"--?\n\n\t320\nDOCTOR LIZARDO\nMay I pass along my congratulations for your great achievement. In the\nmiserable annals of the Earth, you will be duly enshrined! However...\n(a pause)\n...Miss Priddy claims to be unable to solve my problem. And provide the\ncrucial missing circuit for my Overthruster. Perhaps you can convince\nher to try.\n\nLizardo holding the phone toward Penny, ripping her gag off...\n\nPENNY PRIDDY\nI'm not worth it, Buckaroo! Forget me!\n\nBUCKAROO BANZAI\nPenny--?! Are you all right?\n\nDoctor Lizardo yanking the phone from Penny, remaining silent for a\nsecond to torture Buckaroo's soul...\n\nBUCKAROO BANZAI\nShe knows nothing, Lizardo.\n\nDOCTOR LIZARDO\nThen shall we say \"A Penny for your thoughts?\" You can come in her\nplace. Yoyodyne, Dr. Banzai. Come alone, with your Overthruster!"}, {"role": "assistant", "content": "## SCENE 1 — INT. SPEED OPERATOR'S PULPIT (YOYODYNE) - NIGHT\t318\nACTION: Meanwhile in a grimy narrow control booth high above the factory floor, a terrified Penny Priddy sits BOUND and GAGGED, John O'Connor SPREADING\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Buckaroo Banzai: Meanwhile in a grimy narrow control booth high above the factory floor, a terrified Penny Priddy sits BOUND and GAGGED, John O'Connor SPREADING, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Meanwhile in a grimy narrow control booth high above the factory floor, a terrified Penny Priddy sits BOUND and GAGGED, John O'Connor SPREADING The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Coraline\n\nSCREENPLAY EXCERPT:\nINT. OTHER HOUSE HALLWAY - CONTINUOUS\n\n\n\n She PULLS Coraline down the hallway to the mirror at the\n end. Coraline flails at the Tall Other Mother with her\n fists.\n\n\n\n CORALINE\n Ow, that hurts!\n\n\n Tall Other Mother SHOVES Coraline RIGHT THROUGH THE\n MIRROR as if it was water!"}, {"role": "assistant", "content": "## SCENE 1 — INT. OTHER HOUSE HALLWAY - CONTINUOUS\nACTION: She PULLS Coraline down the hallway to the mirror at the end. Coraline flails at the Tall Other Mother with her\nDIALOGUE — Coraline: Ow, that hurts!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Coraline: She PULLS Coraline down the hallway to the mirror at the end. Coraline flails at the Tall Other Mother with her, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. She PULLS Coraline down the hallway to the mirror at the end. Coraline flails at the Tall Other Mother with her The camera pushes in with small amplitude at slow speed. Coraline (S1) says: [English] Ow, that hurts! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: (Shrugs) I - I don't get it.\nSetting: derived from The Puffy Shirt\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SHRUGS\nACTION: (Shrugs) I - I don't get it. This woman just set me up for a job!\nDIALOGUE — George: Me neither!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Puffy Shirt: (Shrugs) I - I don't get it. This woman just set me up for a job!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. (Shrugs) I - I don't get it. This woman just set me up for a job! The camera pushes in with small amplitude at slow speed. George (S1) says: [English] Me neither! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Bottom illustrating the trope 'ContinuityNod'. There are a couple The uneaten sprouts from the Christmas Episode make a comeback in the Halloween Episode . In \"Apocalypse\", we see Richie struggling to finish War and Peace , but he finally manages to do it in \"Dough\" two seasons later.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying ContinuityNod.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Bottom, ContinuityNod, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for ContinuityNod. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in House illustrating the trope 'BailEqualsFreedom'. Averted in House. House and Wilson met at a medical conference in New Orleans, where Wilson (very depressed and drunk from finding out his then-wife had filed for divorce) started a fight at a bar and caused property damage. House bailed him out and the pair went back to New Jersey. It wasn't until years later when House got Wilson arrested for speeding (long story) did Wilson find out that the st...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying BailEqualsFreedom.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for House, BailEqualsFreedom, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for BailEqualsFreedom. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Batman\n\nSCREENPLAY EXCERPT:\nEXT. SIDE STREET - THAT MOMENT - NIGHT\n\n A deserted access street, sidewalks lined with the husks of\n stripped-down cars. MOM, DAD, and JIMMY take a deep breath\n and march down the darkened street. A VOICE startles them.\n\n VOICE\n Hey, mister. Gimme a dollar?\n\n The VOICE belongs to a DERELICT -- nineteen or twenty,\n acne-scarred -- who sits between two garbage cans, his palm\n uplifted. His ratty t-shirt reads: 'I LOVE GOTHAM CITY.'\n\n MOM, DAD, and JIMMY pause for the merest of seconds, then\n move on -- pretending not to hear.\n\n DERELICT\n Mister. How about it. One dollar?\n (standing up)\n One dollar, man. Are you deaf?\n Are you deaf? -- Do you speak\n English??\n\n By now the TOURISTS are halfway across the street.\n Mercifully, the DERELICT doesn't seem to be following.\n\n They pick up their pace. They don't see the SHADOWY FIGURE\n in the alleyway. They don't see the GUN until a gloved hand\n brings it down, butt-first, across the back of DAD's neck.\n\n DAD crumples. MOM grabs JIMMY and backs up against a brick\n wall, too terrified to scream. The DERELICT races across\n the street to join his confederate, the STREET PUNK, who's\n already searching for DAD's wallet.\n\n MOM's mouth opens in panic. They can see she's about to\n snap -- so the STREET PUNK, still in a crouch, trains his\n gun on JIMMY.\n\n STREET PUNK\n Do the k"}, {"role": "assistant", "content": "## SCENE 1 — EXT. SIDE STREET - THAT MOMENT - NIGHT\nACTION: A deserted access street, sidewalks lined with the husks of stripped-down cars. MOM, DAD, and JIMMY take a deep breath\nDIALOGUE — Voice: Hey, mister. Gimme a dollar?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman: A deserted access street, sidewalks lined with the husks of stripped-down cars. MOM, DAD, and JIMMY take a deep breath, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A deserted access street, sidewalks lined with the husks of stripped-down cars. MOM, DAD, and JIMMY take a deep breath The camera pushes in with small amplitude at slow speed. Voice (S1) says: [English] Hey, mister. Gimme a dollar? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fear & Loathing In Las Vegas\n\nSCREENPLAY EXCERPT:\nINT. HOTEL FLAMINGO SUITE - AFTERNOON\n\nDUKE enters. The door hits something with a thud.\n\nA 16-year-old GIRL with the aura of an angry Pit Bull.\n\nGONZO stands in the bathroom doorway -- stark naked with a\ndrug-addled grin on his face.\n\n DUKE\n You degenerate pig!\n\n GONZO\n It can't be helped. This is Lucy.\n (laughing distractedly)\n You know--like \"Lucy In The Sky\n With Diamonds.\"\n\nLUCY eyes DUKE venomously.\n\n GONZO\n Lucy! Lucy, be cool, goddamnit!\n Remember what happened at the\n airport! No more of that, okay?\n\nLUCY keeps her eyes on DUKE. GONZO idles over and puts his\narm round her shoulder.\n\n GONZO\n Lucy... this is my client. This is\n Mr. Duke, the famous journalist.\n He's paying for this suite, Lucy.\n He's on our side.\n\nDUKE flops onto the sofa.\n\n GONZO\n Mr. Duke is my friend. He loves\n artists.\n\nDUKE notices for the first time that the room is full of\nartwork. Maybe 40 or 50 portraits, some in oil, some in\ncharcoal, all more or less the same size and same face.\n\n GONZO\n Lucy paints portraits of Barbra\n Streisand.\n\n 67.\n\n\n LUCY\n I drew these from TV.\n\n GONZO\n Fantastic. She came all the way\n down here from "}, {"role": "assistant", "content": "## SCENE 1 — INT. HOTEL FLAMINGO SUITE - AFTERNOON\nACTION: DUKE enters. The door hits something with a thud. A 16-year-old GIRL with the aura of an angry Pit Bull.\nDIALOGUE — Duke: You degenerate pig!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fear & Loathing In Las Vegas: DUKE enters. The door hits something with a thud. A 16-year-old GIRL with the aura of an angry Pit Bull., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. DUKE enters. The door hits something with a thud. A 16-year-old GIRL with the aura of an angry Pit Bull. The camera pushes in with small amplitude at slow speed. Duke (S1) says: [English] You degenerate pig! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: - LS -- Dorothy, Lion, Tin Man, and Scarecrow enter at top of\nSetting: derived from Fade Out:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ENTRANCE HALL -- CAMERA SHOOTING UP FROM SIDE OF STAIRS -\nACTION: - LS -- Dorothy, Lion, Tin Man, and Scarecrow enter at top of stairs -- CAMERA PANS with them as they run down the stairs --\nDIALOGUE — Dorothy: Oh!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade Out:: - LS -- Dorothy, Lion, Tin Man, and Scarecrow enter at top of stairs -- CAMERA PANS with them as they run down the stairs --, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. - LS -- Dorothy, Lion, Tin Man, and Scarecrow enter at top of stairs -- CAMERA PANS with them as they run down the stairs -- The camera pushes in with small amplitude at slow speed. Dorothy (S1) says: [English] Oh! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Mrs. Fox licks the stump of Fox's tail and mends it with\nSetting: derived from Fantastic Mr. Fox\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. FOX'S TREE. NIGHT\nACTION: Mrs. Fox licks the stump of Fox's tail and mends it with gauze and medical tape. She looks furious. Kylie and the Fox\nDIALOGUE — Ash: It'll grow back, won't it?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fantastic Mr. Fox: Mrs. Fox licks the stump of Fox's tail and mends it with gauze and medical tape. She looks furious. Kylie and the Fox, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Mrs. Fox licks the stump of Fox's tail and mends it with gauze and medical tape. She looks furious. Kylie and the Fox The camera pushes in with small amplitude at slow speed. Ash (S1) says: [English] It'll grow back, won't it? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: S T R A N G E D A Y S\n\nSCREENPLAY EXCERPT:\nEXT. STREET - NIGHT\n\n Now both Spreg and Engelman are stalking Mace through the\n crowd. Spreg catches a glimpse of her running and he\n charges after her, like a bull, with his gun drawn. He\n hammers through anybody that gets in his way.\n\n Mace kicks off her high heels and goes flat out, an Indian\n runner.\n\n Sprinting through the crowd, she sees an apparition ahead\n of her... a guy wearing Death's Head make-up and a black\n shroud, carrying a huge cardboard scythe in one hand and a\n doll (New Year's Baby) in the other. Death watches her\n pass.\n\n CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. STREET - NIGHT\nACTION: Now both Spreg and Engelman are stalking Mace through the crowd. Spreg catches a glimpse of her running and he\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from S T R A N G E D A Y S: Now both Spreg and Engelman are stalking Mace through the crowd. Spreg catches a glimpse of her running and he, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Now both Spreg and Engelman are stalking Mace through the crowd. Spreg catches a glimpse of her running and he The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Elaine and Billy in bed the next morning.\nSetting: derived from The Sponge\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BILLY LEANS IN TO KISS HER.\nACTION: Elaine and Billy in bed the next morning. Fine, fine. Everything o.k.?\nDIALOGUE — Elaine: Good morning.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Sponge: Elaine and Billy in bed the next morning. Fine, fine. Everything o.k.?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Elaine and Billy in bed the next morning. Fine, fine. Everything o.k.? The camera pushes in with small amplitude at slow speed. Elaine (S1) says: [English] Good morning. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I know this ain't very romantic, but\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I know this ain't very romantic, but what can I say? I don't want to explode.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: I know this ain't very romantic, but what can I say? I don't want to explode., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I know this ain't very romantic, but what can I say? I don't want to explode. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: They run through the green tangle. FLASH trips over a vine.\nSetting: derived from Flash Gordon\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. FOREST - FLASH AND TIGER MAN\nACTION: They run through the green tangle. FLASH trips over a vine. TIGER MAN picks him up.\nDIALOGUE — Tiger Man: Thank you for freeing me. 48.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Flash Gordon: They run through the green tangle. FLASH trips over a vine. TIGER MAN picks him up., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They run through the green tangle. FLASH trips over a vine. TIGER MAN picks him up. The camera pushes in with small amplitude at slow speed. Tiger Man (S1) says: [English] Thank you for freeing me. 48. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: CUT TO A SDIREN BLARING and a fire engine racing through the streets of\nSetting: derived from \"Jacob'S Ladder\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. U.S. ARMY RECRUITING HEADQUARTERS - NIGHT\nACTION: CUT TO A SDIREN BLARING and a fire engine racing through the streets of lower MANHATTAN. A CROWD is forming. Banks of lights and television\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Jacob'S Ladder\": CUT TO A SDIREN BLARING and a fire engine racing through the streets of lower MANHATTAN. A CROWD is forming. Banks of lights and television, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CUT TO A SDIREN BLARING and a fire engine racing through the streets of lower MANHATTAN. A CROWD is forming. Banks of lights and television The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Give me a pack of those Turkish\nSetting: derived from Sunset Boulevard\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Give me a pack of those Turkish cigarettes -- Melachrinos.\nDIALOGUE — Artie'S Voice: Stick 'em up, Gillis, or I'll let you have it!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Sunset Boulevard: Give me a pack of those Turkish cigarettes -- Melachrinos., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Give me a pack of those Turkish cigarettes -- Melachrinos. The camera pushes in with small amplitude at slow speed. Artie'S Voice (S1) says: [English] Stick 'em up, Gillis, or I'll let you have it! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Tony, Natasha, and Smart Hulk hash it out.\nSetting: derived from Avengers: Endgame\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. AVENGERS COMPOUND, OFFICE - DAY, EARLY\nACTION: Tony, Natasha, and Smart Hulk hash it out. So this `Time Stone' guy...what\nDIALOGUE — Natasha: So this `Time Stone' guy...what kind of doctor was he?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Avengers: Endgame: Tony, Natasha, and Smart Hulk hash it out. So this `Time Stone' guy...what, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Tony, Natasha, and Smart Hulk hash it out. So this `Time Stone' guy...what The camera pushes in with small amplitude at slow speed. Natasha (S1) says: [English] So this `Time Stone' guy...what kind of doctor was he? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Untitled\n\nSCREENPLAY EXCERPT:\nBrock\nNurse Joy.\n\nNurse Joy\nWhere is this place? Why am I here?\n\nMewtwo\nTo make you take care of me, I took you from the Pok�mon Center.\nA doctor who is familiar with pok�mon is convenient. You were\nquite useful. You don't remember anything though.\n\nBrock\nWhat did you say?\n\nMewtwo\nHumans can be manipulated in any way with my power.\n\nMisty\nWith what power?\n\nPikachu\nPIKA!\n\t(Meanwhile TR are in the castle and they find a strange\nroom and they enter it.)\n\n\nMeowth\nWhat is this place?\n\t(They look and see a Charizard, Venusaur and a Blastoise in\nBig Glass Test Tubes sleeping.)\n\nMeowth\nCute!\n\nJames\nIs it really?\n\nJessie\nThese aren't like any treasures I've seen.\n\t(Jessie walks away and sits on some button and a computer\nturns on with a very fuzzy screen.)\n\nComputer\n...our study... to continue in our research ... we would have to\ncollect samples from our test subjects...\n\t(All of a sudden something grabs Meowth and throws him in a\nmachine)\n\nMeowth\nHELP! HELP!\n\nJessie\nMeowth!\n\t(Both Jessie and James start pull to get Meowth out and\nthey do get him out but not before the machine got 3 hairs from\noff of his tail.)\n\nMeowth\nOUCH! My hair!\n\t(on the computer screen a figure of a Meowth shows up and\nthe computer starts saying some weird stuff then in an empty test\ntube next to Team Rocket, a Meowth falls down from a pipe into\nit)\n\nAll of TR\nMEOWTH?!\n\nMeowth\nI am Meowth.\n\nJames\nThen this is.... Meowth's copy?"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Where is this place? Why am I here? To make you take care of me, I took you from the Pok�mon Center.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: Where is this place? Why am I here? To make you take care of me, I took you from the Pok�mon Center., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Where is this place? Why am I here? To make you take care of me, I took you from the Pok�mon Center. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Good News, Bad News\n\nSCREENPLAY EXCERPT:\nJERRY\n If my father was moving this he'd had\n to have a cigarette in his mouth the\n whole way. (talks from now on like he\n has a cigarette in his mouth) Have you\n got your end?...Your end's got to come\n down first, easy now, drop it down...drop\n it down, your end's got to come down.\n\n\n\n GEORGE\n Ya know, I can't believe you're bringin'\n in an extra bed for woman, that wants\n to sleep with you. Why don't you bring\n in an extra guy too? (sits down)\n\n\n JERRY\n (hands George a beer) Look, it's a very\n awkward situation, I, I don't wanna\n be presumptuous.\n\n\n GEORGE\n All right, all right, one more time,\n one more time! What was the EXACT phrasing\n of the request?\n\n\n JERRY\n All right, she said she couldn't find\n a decent hotel- room...\n\n\n GEORGE\n A decent hotel-room...\n\n\n JERRY\n Yeah, a decent hotel-room, would it\n be terribly inconvenient if she stayed\n at my place.\n\n\n GEORGE\n You can't be serious. This is New York\n city: there must be eleven million decent\n hotel-rooms! Whatta ya need? A flag?\n (waves with his handker"}, {"role": "assistant", "content": "## SCENE 1 — INT. SITS DOWN\nACTION: If my father was moving this he'd had to have a cigarette in his mouth the\nDIALOGUE — George: Ya know, I can't believe you're bringin' in an extra bed for woman, that wants to sleep with you. Why don't you bring in an extra guy too? (sits down)\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Good News, Bad News: If my father was moving this he'd had to have a cigarette in his mouth the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. If my father was moving this he'd had to have a cigarette in his mouth the The camera pushes in with small amplitude at slow speed. George (S1) says: [English] Ya know, I can't believe you're bringin' in an extra bed for woman, that wants to sleep with you. Why don't you bring in an extra guy too? (sits down) The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Hellboy enters to find Abe, bleeding but alive, leaning\nSetting: derived from \"Hellboy\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ABANDONED SHOWER ROOM (SET) - DAY\nACTION: Hellboy enters to find Abe, bleeding but alive, leaning against the tile, blue blood all around him. Hellboy tries\nDIALOGUE — Hellboy: We need an ambulance. Now!! Over!!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hellboy\": Hellboy enters to find Abe, bleeding but alive, leaning against the tile, blue blood all around him. Hellboy tries, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hellboy enters to find Abe, bleeding but alive, leaning against the tile, blue blood all around him. Hellboy tries The camera pushes in with small amplitude at slow speed. Hellboy (S1) says: [English] We need an ambulance. Now!! Over!! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Coraline goes to the little door, still ajar. She peeks\nSetting: derived from Coraline\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LIVING ROOM - SAME\nACTION: Coraline goes to the little door, still ajar. She peeks behind and finds the WALL OF SOLID BRICKS. Yes, it was\nDIALOGUE — Coraline: Huh?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Coraline: Coraline goes to the little door, still ajar. She peeks behind and finds the WALL OF SOLID BRICKS. Yes, it was, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Coraline goes to the little door, still ajar. She peeks behind and finds the WALL OF SOLID BRICKS. Yes, it was The camera pushes in with small amplitude at slow speed. Coraline (S1) says: [English] Huh? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nC'mon, Lenny - we'll take a look down\n\t\tthere together. Then you'll know. You'll\n\t\tknow what you really are.\n\nLeonard glances fearfully at the door, then looks at Teddy.\nHe THRUSTS the barrel of his gun into Teddy's mouth and WE\nARE AT THE SHOT FROM THE END OF THE OPENING SEQUENCE. Teddy\npanics, shaking his head, trying to talk around the metal,\nbut GAGS just as Leonard pulls the trigger. A SHOT rings out\nas we -\n\n\t\t\t\t\t\t\tCUT TO:\n\n5 INT. MOTEL ROOM 21 - DAY ##BLACK AND WHITE SEQUENCE##\t 5\n\nLeonard lies on the queen-sized bed. He lifts his head.\n\n\t\t\t\tLEONARD (V.O.)\n\t\tSo you're in some motel room...\n\nHe gets up, surveys the room as if for the first time. He\nwears BOXERS and a PLAID WORK SHIRT.\n\n\t\t\t\tLEONARD (cont'd)\n\t\t... you don't know how long you've been\n\t\tthere, or how you got there...\n\nThere is a room key on the dresser. The plastic tag\nidentifies it as the key to ROOM 21. Leonard opens drawers in\nthe room."}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: C'mon, Lenny - we'll take a look down there together. Then you'll know. You'll\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: C'mon, Lenny - we'll take a look down there together. Then you'll know. You'll, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. C'mon, Lenny - we'll take a look down there together. Then you'll know. You'll The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Fred's POV: A BIG FIST slams into Fred's face, knocking him\nSetting: derived from As Credits Roll\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. POLICE STATION - INTERROGATION ROOM - NIGHT\nACTION: Fred's POV: A BIG FIST slams into Fred's face, knocking him back in a chair. Ed and Al are standing in front of Fred.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from As Credits Roll: Fred's POV: A BIG FIST slams into Fred's face, knocking him back in a chair. Ed and Al are standing in front of Fred., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Fred's POV: A BIG FIST slams into Fred's face, knocking him back in a chair. Ed and Al are standing in front of Fred. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Marine Biologist\n\nSCREENPLAY EXCERPT:\nGEORGE\n Diane DeConn? You saw Diane DeConn!\n\n\n\n JERRY\n Something huh?\n\n\n GEORGE\n Yeah! How'd she look.\n\n\n JERRY\n She looked great. She asked about you.\n\n\n\n GEORGE\n She did! What did she say?\n\n\n JERRY\n \"How's George?\"\n\n\n GEORGE\n George! She said George? She remembered\n my name. Diane DeConn\n\n\n remembered my name. She was the \"it\" girl!\n\n\n JERRY\n Yeah she asked for your number, I think\n she's gonna get in touch with\n\n\n you.\n\n\n GEORGE\n OK, I'm tellin' you right now if your\n kiddin' around I'm not gonna\n\n\n be able to be friends with you anymore. I'm serious about that.\n You\n\n\n got that.\n\n\n JERRY\n I got no problem with that.\n\n\n GEORGE\n Good. Cause if this is a lie, if this\n is a joke, if this is your\n\n\n idea of some cute little game...we're finished!\n\n\n JERRY\n Expect a call.\n\n\n GEORGE\n Oh my god he's not kidding.\n\n\n JERRY\n Now I should tell you at this point\n she's under the imp"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Diane DeConn? You saw Diane DeConn! Yeah! How'd she look.\nDIALOGUE — Jerry: Something huh?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Marine Biologist: Diane DeConn? You saw Diane DeConn! Yeah! How'd she look., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Diane DeConn? You saw Diane DeConn! Yeah! How'd she look. The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] Something huh? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Matrix\n\nSCREENPLAY EXCERPT:\nINT. SHRINE (MATRIX) - DAY\n\n\tNeo enters nervously. Beneath his feet is a path of the\n\tzodiac leading to marbled stairs that rise to a dais and\n\ta three-legged throne.\n\n\tThe throne is empty.\n\n\t\t\t\t\tNEO\n\t\t\tHello?\n\n\tA DISTANT FEMALE VOICE calls to him and he follows it up\n\tthe stairs. At the top of the dais, he smells something\n\tcooking.\n\n\tFollowing the scent, he moves behind the pillars where he\n\tfinds an open door.\n\n\t\t\t\t\tVOICE (O.S.)\n\t\t\tJust come on in.\n\n\tHe walks through a vestibule where he sees a fabulous\n\tmoonstone headdress and velvet robes. A second door\n\tleads into --"}, {"role": "assistant", "content": "## SCENE 1 — INT. SHRINE (MATRIX) - DAY\nACTION: Neo enters nervously. Beneath his feet is a path of the zodiac leading to marbled stairs that rise to a dais and\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Matrix: Neo enters nervously. Beneath his feet is a path of the zodiac leading to marbled stairs that rise to a dais and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Neo enters nervously. Beneath his feet is a path of the zodiac leading to marbled stairs that rise to a dais and The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Heavy Metal Movie\n\nSCREENPLAY EXCERPT:\nINT. HARRY'S APARTMENT\n\n This futuristic bachelor pad is constructed entirely of brightly\n coloured molden plastic. Bowling trophies sit on a shelf\n underneath a neon \"Buweiser\" clock. The place is a mess.\n\n HARRY carries the GIRL in, pushes the light button to \"On\",\n and lays her down on the sofa. As he takes off his jacket\n he looks down at the unconscious figure - her clothes torn\n in all the right places. She moans and starts to come to.\n\n\n GIRL\n\n Where am I?\n\n\n HARRY\n\n My place ...\n\n\n Walking over to a hi-tech plastic bar, HARRY pushes a button\n and a machine pours and mixes a drink. He hands the GIRL the\n glass. She sits up and drinks.\n\n\n GIRL\n (looking at Harry)\n\n They'd have killed me if it\n wasn't for you.\n\n\n HARRY\n\n What'd they want to snuff you\n for?\n\n\n - - -"}, {"role": "assistant", "content": "## SCENE 1 — INT. HARRY'S APARTMENT\nACTION: This futuristic bachelor pad is constructed entirely of brightly coloured molden plastic. Bowling trophies sit on a shelf\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Heavy Metal Movie: This futuristic bachelor pad is constructed entirely of brightly coloured molden plastic. Bowling trophies sit on a shelf, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. This futuristic bachelor pad is constructed entirely of brightly coloured molden plastic. Bowling trophies sit on a shelf The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The room is dark except for one burning candle. Kimberly tosses and\nSetting: derived from Final Destination 2\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. KIMBERLY'S ROOM - NIGHT\nACTION: The room is dark except for one burning candle. Kimberly tosses and turns. The wind sends a single branch scraping against her window. A\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Final Destination 2: The room is dark except for one burning candle. Kimberly tosses and turns. The wind sends a single branch scraping against her window. A, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The room is dark except for one burning candle. Kimberly tosses and turns. The wind sends a single branch scraping against her window. A The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DrakeAndJosh illustrating the trope 'GeekyAnalogy'. Inverted in Drake & Josh when Josh teaches Drake about atoms and molecules by comparing it to guitar chords.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying GeekyAnalogy.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DrakeAndJosh, GeekyAnalogy, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for GeekyAnalogy. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: L A B Y R I N T H\n\nSCREENPLAY EXCERPT:\nL A B Y R I N T H\n\n by\n Laura Phillips\n and\n Terry Jones\n\n Story\n by\n Dennis Lee\n\n-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\n\nFADE IN ON:\n\n1\tEXT: SKY - DAY\n\nA WHITE BIRD soaring. The sky is a glorious explosion of blue and\nmauve and lavender. The setting sun washes the clouds with a delicate\npink tint. The bird swoops and spirals and we are right there with\nhim. Then suddenly, below us, an extraordinary sight appears.\n\n2\tEXT: LABYRINTH - DAY\n\nIt is the labyrinth, an enormous maze of incredible mandala like\nintricacy. From our magnificent vantage point, WE ARE BARELY ABLE TO\nMAKE OUT its details: the twisting walls interrupted here and there\nby lush forest, the complex web of waterways, the forbidding castle\nat the core. It is vast and magical, frightening and compelling at\nthe same time. We would love to linger, perhaps fly lower and get a\nbetter look at this wondrous place, but the bird flies on\n\n3\tEXT: GLADE - DAY\n\nIt is an idyllic setting, warm and inviting. The late afternoon sun\nwashes everything with a pink glow. Between the water and a line of\nlush fruit trees is a flower-strewn glade. A FIGURE emerges from the\ntrees and walks toward the water. It is a YOUNG WOMAN dressed in a\nflowing white dress that swirls around her in the breeze. The pink\nlight burnishes her hair and causes a gold circlet she wears on her\nhead to glow. She is b"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- A WHITE BIRD soaring. The sky is a glorious explosion of blue and\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from L A B Y R I N T H: -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- A WHITE BIRD soaring. The sky is a glorious explosion of blue and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- A WHITE BIRD soaring. The sky is a glorious explosion of blue and The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Black Screen\n\nSCREENPLAY EXCERPT:\nEXT. SLOPES OF AMON HEN -- DAY\n\nFrodo is walking beneath the trees...lost in thought. His\nfeet hit the rough edge of an ancient stone slab...his eyes\nfollow an overgrown path towards stone stairs leading to the\nsummit of Amon Hen...the seeing seat. A CRACKLING SOUND!\nFrodo freezes....\n\n BOROMIR\n (quietly)\n None of us should wander alone; you least\n of all. So much depends on you...Frodo?\n\nFrodo turns slowly...he stares at Boromir, tense, cautious.\n\n BOROMIR (CONT'D)\n I know why you seek solitude. You\n suffer, I see it day by day. Are you\n sure you do not suffer needlessly?\n\nFrodo stands silent for a moment..the murmur of the Wind in\nthe trees and the distant roar of the falls of Rauros can be\nheard\n\n BOROMIR (CONT'D)\n Let me help you. There are other ways,\n Frodo..other paths that we might take.\n\n FRODO\n I know what you would say, and it would\n seem like wisdom but for the warning of\n my heart.\n\n BOROMIR\n Warning? Against what?\n\nBoromir has started forward towards Frodo, he pulls himself\nup.\n\n\n\n\n (CONTINUED)\n\n 109.\nCONTINUED:\n\n\n BOROMIR (CONT'D)\n We are all afraid, Frodo. But to let that\n fear drive us to destroy what hope we\n have...don't you see that is madness?\n\n FRODO\n There is no other"}, {"role": "assistant", "content": "## SCENE 1 — EXT. SLOPES OF AMON HEN -- DAY\nACTION: Frodo is walking beneath the trees...lost in thought. His feet hit the rough edge of an ancient stone slab...his eyes\nDIALOGUE — Boromir: None of us should wander alone; you least of all. So much depends on you...Frodo?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: Frodo is walking beneath the trees...lost in thought. His feet hit the rough edge of an ancient stone slab...his eyes, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Frodo is walking beneath the trees...lost in thought. His feet hit the rough edge of an ancient stone slab...his eyes The camera pushes in with small amplitude at slow speed. Boromir (S1) says: [English] None of us should wander alone; you least of all. So much depends on you...Frodo? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I could have killed myself.\nSetting: derived from Joker\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I could have killed myself. CLOSE ON THE LAST LINE, he crosses out \"could\"...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Joker: I could have killed myself. CLOSE ON THE LAST LINE, he crosses out \"could\"..., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I could have killed myself. CLOSE ON THE LAST LINE, he crosses out \"could\"... The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jennifer'S Body\n\nSCREENPLAY EXCERPT:\nEXT. KETTLE HIGH CAMPUS - CONTINUOUS\n Needy and Chip walk to an alley behind the school.\n Melting snow drips off the eaves.\n Needy pulls out a large black binder and opens it. It's\n full of copies from the library and Internet printouts.\n The spine reads \"JENNIFER RESEARCH\"\n\n NEEDY\n Look at this.\n She thrusts a piece of paper into CHIP's hands.\n\n\n (CONTINUED)\n\n 84.\n CONTINUED:\n\n CHIP\n (CONFUSED)\n Demon...demonic...\n NEEDY\n Demonic transference. It's\n something that happens when you\n try to sacrifice a virgin to Satan\n without using an actual virgin!\n CHIP stares at her.\n NEEDY (CONT'D)\n I think that's what happened to\n Jennifer. Those guys from that\n band were trying to sacrifice her\n in the woods. But what they didn't\n realize is that she hasn't been a\n virgin since junior high!\n Jennifer's a total cum-dumpster!\n It all makes sense now! Read\n this...\n She grabs the piece of paper from him.\n NEEDY (CONT'D)\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. KETTLE HIGH CAMPUS - CONTINUOUS\nACTION: Needy and Chip walk to an alley behind the school. Melting snow drips off the eaves.\nDIALOGUE — Needy: Look at this. She thrusts a piece of paper into CHIP's hands.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jennifer'S Body: Needy and Chip walk to an alley behind the school. Melting snow drips off the eaves., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Needy and Chip walk to an alley behind the school. Melting snow drips off the eaves. The camera pushes in with small amplitude at slow speed. Needy (S1) says: [English] Look at this. She thrusts a piece of paper into CHIP's hands. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Les Miserables\n\nSCREENPLAY EXCERPT:\nDo you hear the people sing\n\n GAVROCHE/COURFEYRAC\n Singing the song of angry men\n\n STUDENTS/CITIZENS"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Do you hear the people sing Singing the song of angry men\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Les Miserables: Do you hear the people sing Singing the song of angry men, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Do you hear the people sing Singing the song of angry men The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \" T E R M I N A T O R \"\n\nSCREENPLAY EXCERPT:\nHold on, goddamnit!\n\n He rouses slightly, gripping her tighter.\n\n HIGH ANGLE - MOVING WITH BOTH VEHICLES as Sarah starts to\n ZIGZAG desperately across all four lanes. The truck stays\n with her, closing, its trailer WHIPLASHING VIOLENTLY.\n\n CUT TO:\n\n\n241 EXT./INT. TUNNEL - NIGHT 241\n\n The truck is right behind them as then enter a TUNNEL.\n A half-mile of exitless concrete and strobing fluorescent\n lights.\n\n M.C.U. - SARAH AND KYLE (PROCESS SHOT) - He blinks and looks\n back at a SOLID WALL OF METAL AND LIGHTS looming behind them.\n Sarah hunches down. They hit eighty.\n\n FULL SHOT - The leviathan dwarfs them, its big tires ROARING\n like the hubs of Hell.\n\n CUT TO:\n\n\n242 EXT. FREEWAY - NIGHT 242\n\n The tanker is twenty feet behind them as they clear the\n tunnel. Sarah dodges to one side and LOCKS THE BRAKES.\n The bike slides, fish-tailing.\n The truck roars past, hitting the air-brakes.\n The trailers force her closer and closer to the guardrail\n as Terminator tries to sandwich her.\n The bike slides to a stop.\n The rearmost set of trailer wheels slams into the guardrail\n right in front of Sarah.\n\n Sarah emerges from a cloud of tire smoke, cutting across\n all four lances behind the stopped semi.\n\n CUT TO:\n\n\n243 "}, {"role": "assistant", "content": "## SCENE 1 — INT. PROCESS SHOT\nACTION: He rouses slightly, gripping her tighter. HIGH ANGLE - MOVING WITH BOTH VEHICLES as Sarah starts to\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \" T E R M I N A T O R \": He rouses slightly, gripping her tighter. HIGH ANGLE - MOVING WITH BOTH VEHICLES as Sarah starts to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He rouses slightly, gripping her tighter. HIGH ANGLE - MOVING WITH BOTH VEHICLES as Sarah starts to The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Hunched down, she moves as quickly as she can, tearing\nSetting: derived from Coraline\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PASSAGEWAY � CONTINUOUS\nACTION: Hunched down, she moves as quickly as she can, tearing through the sticky cobwebs.\nDIALOGUE — Tall Other Mother: Coraline!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Coraline: Hunched down, she moves as quickly as she can, tearing through the sticky cobwebs., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hunched down, she moves as quickly as she can, tearing through the sticky cobwebs. The camera pushes in with small amplitude at slow speed. Tall Other Mother (S1) says: [English] Coraline! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Untitled\n\nSCREENPLAY EXCERPT:\nUmio\nThat's my Gyarados. I rode him across the sea. Those kind of\nwaves are nothing for it.\n\nAsh\nGyarados is a brutal pok�mon isn't it?\n\nUmio\nYeah, but if you can handle it, no one is more reliable.\n\nSweet\nMy pok�mon are just the same over there.\n\t(points to her Blastoise, Wigglytuff, and more. Then all\nthe lights turn off, there is a bright beam of light coming down\nthe center of a huge spiral staircase.)\n\nLady\nThat you for waiting everyone. The strongest pok�mon trainer is\ncoming.\n\t(All of a sudden Mewtwo starts floating down and all the\npok�mon get worried.)\n\n\nLady\nYes, this is the strongest pok�mon trainer and strongest pok�mon.\nMr. Mewtwo."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: That's my Gyarados. I rode him across the sea. Those kind of waves are nothing for it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: That's my Gyarados. I rode him across the sea. Those kind of waves are nothing for it., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. That's my Gyarados. I rode him across the sea. Those kind of waves are nothing for it. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: 1 Ext. Cityscape Sunset 1\n\nSCREENPLAY EXCERPT:\nYes ... No ... send that back ...\n wrong department ... of course ... of\n course not ... yes ... no ... maybe.\n\n CUT to SAM watching this caravanserai with awe as it\n starts pass him.\n\n EXPEDITER 4\n ... about these requisitions ...\n\n EXPEDITER 5\n Mr Warren ... EX/27 has 15 suspects\n still outstanding.\n\n EXPEDITER 6\n ... a decision, Mr Warren ...\n\n WARREN\n ... cancel that ... okay ... put half\n as terrorists, the rest as victims\n ... yes ... yes ... no ... definitely\n no ...\n\n SAM doesn't have the nerve to jump into this. The scrum\n sweeps by and fades away along the corridor, and finally\n disappearing around the corner at the other end. SAM\n follows. Silence has descended again.\n\n\n61 INT. CORRIDOR DAY 61\n\n SAM gets to the corner of the corridor and finds a similar\n corridor at right angles. He hesitates and continues. Then\n he starts to hear the sound again. It is coming up from\n behind. MR WARREN has circumnavigated the building. The\n same sort of business is being enacted at the same pace.\n As the scrum reaches SAM he gathers his nerve and jumps\n right in beside WARREN, and keeps going.\n\n SAM"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Yes ... No ... send that back ... wrong department ... of course ... of\nDIALOGUE — Expediter 4: ... about these requisitions ...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from 1 Ext. Cityscape Sunset 1: Yes ... No ... send that back ... wrong department ... of course ... of, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Yes ... No ... send that back ... wrong department ... of course ... of The camera pushes in with small amplitude at slow speed. Expediter 4 (S1) says: [English] ... about these requisitions ... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Eric Rogers, Ken Keeler & David X. Cohen\nSetting: derived from Futurama\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Eric Rogers, Ken Keeler & David X. Cohen Transcribed by Dave, The Neutral Planet\nDIALOGUE — Farnsworth: Good news, everyone! I've invented a device that allows you to operate equipment from great distances. I call it \"the fing-longer\". Observe. And, here we go....\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Futurama: Eric Rogers, Ken Keeler & David X. Cohen Transcribed by Dave, The Neutral Planet, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Eric Rogers, Ken Keeler & David X. Cohen Transcribed by Dave, The Neutral Planet The camera pushes in with small amplitude at slow speed. Farnsworth (S1) says: [English] Good news, everyone! I've invented a device that allows you to operate equipment from great distances. I call it \"the fing-longer\". Observe. And, here we go.... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BykerGrove illustrating the trope 'VerySpecialEpisode'. \nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying VerySpecialEpisode.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BykerGrove, VerySpecialEpisode, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for VerySpecialEpisode. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Jacob'S Ladder\"\n\nSCREENPLAY EXCERPT:\nEXT. U.S. ARMY RECRUITING HEADQUARTERS - NIGHT\n\nCUT TO A SDIREN BLARING and a fire engine racing through the streets of\nlower MANHATTAN. A CROWD is forming. Banks of lights and television\ncameras amass in the cold night air. Police cars and mobile units rush\nto the scene.\n\nCUT TO JACOB. In one hand he is holding a brightly lit torch. In the\nother he is holding a container of gasoline and pouring it on the steps\nof the U.S. ARMY RECRUITING HEADQUARTERS. The volatile liquid splashes\nagainst his pants and shoes and runs down the pavement. A five gallon\ncontainer lies emptying nearby. Gasoline belches from it insistently\nand pours onto the street. Bystanders back away as the gasoline snakes\ntoward them.\n\nTelevision cameras and microphones are pointing in JACOB's direction,\nbut at a safe distance. He is yelling at them, his teeth chattering\nfrom the cold.\n\n\t\t\t\tJACOB\n\t\tListen to me. There were four\n\t\tcompanies in our batallion. Five\n\t\thundred men. Seven of us were left\n\t\twhen it was over. Seven! Four\n\t\tcompanies engaged in an enemy\n\t\toffensive that not one of us who\n\t\tsurvived can remember fighting.\n\n\t\t\t\tBYSTANDERS\n\t\tUse the torch!\n\n\t\t\t\tONLOOKER\n\t\tShut up! Let him talk!\n\nPOLICE AMBULANCES are arriving at the scene. FIREMEN ready hoses at\nnearby hydrants. T.V. CAMERAS are rolling.\n\n\t\t\t\tJACOB\n\t\t\t(shouting)\n\t\tYou don't forget a battle where 500\n\t\tmen were killed. They did something\n\t\tto us. I want to know the truth, the\n\t\tgoddamn truth. We have a right to\n\t\tknow.\n\t\t\t(he yells toward the\n\t\t\tcameras)\n\t\tAre you getting all this? I want this\n\t\ton national T.V. I want the whole\n\t\tcountr"}, {"role": "assistant", "content": "## SCENE 1 — EXT. U.S. ARMY RECRUITING HEADQUARTERS - NIGHT\nACTION: CUT TO A SDIREN BLARING and a fire engine racing through the streets of lower MANHATTAN. A CROWD is forming. Banks of lights and television\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Jacob'S Ladder\": CUT TO A SDIREN BLARING and a fire engine racing through the streets of lower MANHATTAN. A CROWD is forming. Banks of lights and television, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CUT TO A SDIREN BLARING and a fire engine racing through the streets of lower MANHATTAN. A CROWD is forming. Banks of lights and television The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The MAC Mothers are on a stage, preparing two electric\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. USO SHOW - NIGHT\nACTION: The MAC Mothers are on a stage, preparing two electric Something must be done!\nDIALOGUE — Kyle'S Mother: Something must be done! Tomorrow it will be! We're going to execute this threat to our democracy!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: The MAC Mothers are on a stage, preparing two electric Something must be done!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The MAC Mothers are on a stage, preparing two electric Something must be done! The camera pushes in with small amplitude at slow speed. Kyle'S Mother (S1) says: [English] Something must be done! Tomorrow it will be! We're going to execute this threat to our democracy! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Did you see that whale thing on TV last\nSetting: derived from The Marine Biologist\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. GEORGE LOOKS TO THE TABLE\nACTION: Did you see that whale thing on TV last I'm such a Huge whale fan. These marine\nDIALOGUE — Jerry: No.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Marine Biologist: Did you see that whale thing on TV last I'm such a Huge whale fan. These marine, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Did you see that whale thing on TV last I'm such a Huge whale fan. These marine The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] No. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dark City\n\nSCREENPLAY EXCERPT:\nINT. HOSPITAL ROOM\n\n Bumstead enters darkness - pneumatic MACHINE NOISE fills the\n gloom. He walks quietly to his mother, encased in a metallic\n contraption, something like an iron lung.\n\n BUMSTEAD (whisper)\n Brought you these.\n\n He places flowers on the bedside table. His mother just stares\n expressionless into space, in a coma. He sits on a metal\n chair, looks at the woman. Her life-sustaining machines hiss\n and breathe."}, {"role": "assistant", "content": "## SCENE 1 — INT. HOSPITAL ROOM\nACTION: Bumstead enters darkness - pneumatic MACHINE NOISE fills the gloom. He walks quietly to his mother, encased in a metallic\nDIALOGUE — Bumstead: Brought you these.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dark City: Bumstead enters darkness - pneumatic MACHINE NOISE fills the gloom. He walks quietly to his mother, encased in a metallic, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Bumstead enters darkness - pneumatic MACHINE NOISE fills the gloom. He walks quietly to his mother, encased in a metallic The camera pushes in with small amplitude at slow speed. Bumstead (S1) says: [English] Brought you these. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Dollhouse illustrating the trope 'BetterOnDVD'. Dollhouse, partly because shows by Tim Minear and Joss Whedon already tend to do this anyway, partly because it's a dense, fast-paced Myth Arc-heavy show with steady plot and character arc progression from one episode to the next, and partly because watching the episodes more than once means you catch more of the little things.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying BetterOnDVD.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Dollhouse, BetterOnDVD, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for BetterOnDVD. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The MCP blasted it... There are very\nSetting: derived from T R O N\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The MCP blasted it... There are very few domains left with any power at all.\nDIALOGUE — Flynn: Lousy utility companies ... they're the same wherever you go.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T R O N: The MCP blasted it... There are very few domains left with any power at all., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The MCP blasted it... There are very few domains left with any power at all. The camera pushes in with small amplitude at slow speed. Flynn (S1) says: [English] Lousy utility companies ... they're the same wherever you go. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Blade\n\nSCREENPLAY EXCERPT:\nEXT. BLADE'S OLDS/ABANDONED FACTORY - NIGHT\n\n\tA gate grinds open.\n\n\tWe follow the Olds as it cruises around the back of the building,\n\theading down a concrete loading ramp. At the bottom of the ramp, a\n\theavy iron door rises. Blade's Olds disappears into the darkness."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BLADE'S OLDS/ABANDONED FACTORY - NIGHT\nACTION: We follow the Olds as it cruises around the back of the building, heading down a concrete loading ramp. At the bottom of the ramp, a\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade: We follow the Olds as it cruises around the back of the building, heading down a concrete loading ramp. At the bottom of the ramp, a, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We follow the Olds as it cruises around the back of the building, heading down a concrete loading ramp. At the bottom of the ramp, a The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Buckaroo Banzai\n\nSCREENPLAY EXCERPT:\nINT. LIZARDO'S ROOM - NIGHT\t104\n\nNearly half a century later, the elderly Lizardo regards with\nsmoldering evil the turn of the key in his own door and the appearance\nof a GUARD who comes in and unplugs his TV set.\n\nGUARD\nCheer up, Lizardo, it's Friday. I come for your TV. You been using\ntoo damn much juice...ten thousand kilowatts again this month. Beats\nme how one old homicidal loony could use that much power.\n\nThe guard leaving with Lizardo's TV...Lizardo trying to control\nhimself, his eyes rolling up at the ceiling where yet another UNUSUAL\nEQUATION has been scrawled in an angry hand.\n\nDOCTOR LIZARDO\nI want my TV...don't take my TV...I want it. I want the Overthruster.\nI want it."}, {"role": "assistant", "content": "## SCENE 1 — INT. LIZARDO'S ROOM - NIGHT\t104\nACTION: Nearly half a century later, the elderly Lizardo regards with smoldering evil the turn of the key in his own door and the appearance\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Buckaroo Banzai: Nearly half a century later, the elderly Lizardo regards with smoldering evil the turn of the key in his own door and the appearance, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Nearly half a century later, the elderly Lizardo regards with smoldering evil the turn of the key in his own door and the appearance The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The cruiser PUNCHES through the burning \"Paradise\" sign,\nSetting: derived from Legion\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HIGHWAY - CONTINUOUS\nACTION: The cruiser PUNCHES through the burning \"Paradise\" sign, smashing it into fiery pieces!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Legion: The cruiser PUNCHES through the burning \"Paradise\" sign, smashing it into fiery pieces!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The cruiser PUNCHES through the burning \"Paradise\" sign, smashing it into fiery pieces! The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Flash Gordon\n\nSCREENPLAY EXCERPT:\nEXT. FOREST -FEATURE BARIN\n (BARIN, TREE MAN)\n\n ANGLE - BARIN advancing by himself, HEARING the triumphant\n SHOUTS of his men.\n\n TREE MAN (O.S.)\n Prince Barin! We have him!\n\n BARIN\n Not him! Bring him back for sport:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. FOREST -FEATURE BARIN\nACTION: ANGLE - BARIN advancing by himself, HEARING the triumphant Prince Barin! We have him!\nDIALOGUE — Tree Man: Prince Barin! We have him!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Flash Gordon: ANGLE - BARIN advancing by himself, HEARING the triumphant Prince Barin! We have him!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ANGLE - BARIN advancing by himself, HEARING the triumphant Prince Barin! We have him! The camera pushes in with small amplitude at slow speed. Tree Man (S1) says: [English] Prince Barin! We have him! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Alderaan looms behind the Death Star battlestation.\nSetting: derived from Star Wars\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. DEATH STAR\nACTION: Alderaan looms behind the Death Star battlestation.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Star Wars: Alderaan looms behind the Death Star battlestation., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Alderaan looms behind the Death Star battlestation. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Anthony and Bob go out the door. Dignan is walking over to a\nSetting: derived from Bottle Rocket\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HAMBURGER PLACE. NIGHT\nACTION: Anthony and Bob go out the door. Dignan is walking over to a station wagon. Little Richard is getting in.\nDIALOGUE — Dignan: Little Richard. Wait a second. Hang on. I didn't mean to --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bottle Rocket: Anthony and Bob go out the door. Dignan is walking over to a station wagon. Little Richard is getting in., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Anthony and Bob go out the door. Dignan is walking over to a station wagon. Little Richard is getting in. The camera pushes in with small amplitude at slow speed. Dignan (S1) says: [English] Little Richard. Wait a second. Hang on. I didn't mean to -- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Seven\n\nSCREENPLAY EXCERPT:\nINT. LUXURY APARTMENT/CRIME SCENE, LIVING ROOM -- NIGHT\n\n A MALE FORENSIC uses a magnifying glass to study a very clear\n fingerprint in black powder on the wall.\n\n FORENSIC\n Oh, boy...\n\n MILLS (O.S.)\n Talk to me.\n\n The forensic bites his lip, still studying.\n\n Mills and Somerset watch the forensic who works offscreen.\n\n MILLS (CONT)\n (to Somerset)\n Listen, honestly... have you ever seen\n anything like this? Been involved in\n anything remotely like this?\n\n SOMERSET\n No. I have not.\n\n FORENSIC (O.S.)\n Well, I can tell you this, detectives...\n\n The forensic steps down from a stool. Behind him, where the\n painting once was, there are fingerprints, clear and distinct.\n The prints have been left side by side, to form letters which\n form the words: HELP ME.\n\n FORENSIC (CONT)\n ...just by studying the underloop...\n these are not the victim's prints."}, {"role": "assistant", "content": "## SCENE 1 — INT. LUXURY APARTMENT/CRIME SCENE, LIVING ROOM -- NIGHT\nACTION: A MALE FORENSIC uses a magnifying glass to study a very clear fingerprint in black powder on the wall.\nDIALOGUE — Forensic: Oh, boy...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Seven: A MALE FORENSIC uses a magnifying glass to study a very clear fingerprint in black powder on the wall., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A MALE FORENSIC uses a magnifying glass to study a very clear fingerprint in black powder on the wall. The camera pushes in with small amplitude at slow speed. Forensic (S1) says: [English] Oh, boy... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Les Miserables\n\nSCREENPLAY EXCERPT:\n69.\n\n GRANTAIRE (cont'd)\n And here he comes like Don Juan!\n It is better than an opera!\n\nThey burst into laughter. But Enjolras isn't smiling.\n\n ENJOLRAS\n It is time for us all\n To decide who we are.\n Do we fight for the right\n To a night at the opera now?\n Have you asked yourselves\n What's the price you might pay?\n Is this simply a game\n For rich young boys to play?\n The colours of the world\n Are changing day by day -\n\n Red - the blood of angry men!\n Black - the dark of ages past!\n Red - a world about to dawn!\n Black - the night that ends at\n last!\n\n MARIUS\n Had you seen her today\n You might know how it feels\n To be struck to the bone\n By a moment of breathless delight!\n\n Had you been there today\n You might also have known\n How the world may be changed\n In just one burst of light\n And what was right seems wrong\n And what was wrong seems right!\n\n GRANTAIRE\n Red!\n\n MARIUS\n I feel my soul on fire!\n\n GRANTAIRE\n Black!\n\n MARIUS\n My world if she's not there!\n\n STUDENTS\n Red!\n\n MARIUS\n The colour of desire!\n\n STUDENTS\n Black!\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: And here he comes like Don Juan! It is better than an opera!\nDIALOGUE — Grantaire: And here he comes like Don Juan! It is better than an opera!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Les Miserables: And here he comes like Don Juan! It is better than an opera!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. And here he comes like Don Juan! It is better than an opera! The camera pushes in with small amplitude at slow speed. Grantaire (S1) says: [English] And here he comes like Don Juan! It is better than an opera! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Still in slow motion, the YOUNGER DUKE is trying to eat a\nSetting: derived from Fear & Loathing In Las Vegas\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. MATRIX MEN'S ROOM - NIGHT\nACTION: Still in slow motion, the YOUNGER DUKE is trying to eat a HUGE SPANSULE OF ACID. With difficulty.\nDIALOGUE — Duke: I decided to eat only half at first. Good thinking. But I spilled the rest on the sleeve of my red Pendleton shirt.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fear & Loathing In Las Vegas: Still in slow motion, the YOUNGER DUKE is trying to eat a HUGE SPANSULE OF ACID. With difficulty., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Still in slow motion, the YOUNGER DUKE is trying to eat a HUGE SPANSULE OF ACID. With difficulty. The camera pushes in with small amplitude at slow speed. Duke (S1) says: [English] I decided to eat only half at first. Good thinking. But I spilled the rest on the sleeve of my red Pendleton shirt. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Pleasantville\n\nSCREENPLAY EXCERPT:\nINT. DOCTOR'S OFFICE. DAY.\n\n\tIt is a typical '50s family practice right out of Norman\n\tRockwell. There is a jar of tongue depressors on the counter\n\tand a jar of lollipops beside them. Lisa Anne (Mary Sue's\n\tbest friend) is being examined by Dr. Henderson. Her mother\n\tsits at her side.\n\n\t\t\t\tDR. HENDERSON\n\t\tLet me see it again.\n\n\tLisa Anne opens her mouth and sticks out a BRIGHT RED TONGUE.\n\tEverything else in the frame is Black and White, but her\n\ttongue literally gleams with color.\n\n\t\t\t\tDR. HENDERSON (CONT)\n\t\t\t(examining it)\n\t\tWell ... I don't think it's anything to\n\t\tworry about ... It'll probably just\n\t\tclear up by itself.\n\t\t\t(to Lisa Anne)\n\t\tCut down on greasy foods and chocolate.\n\t\tNo french fries, that kind of thing.\n\t\t\t(aside/\n\t\t\tto the Mother)\n\t\tIt's just a \"teenage\" thing."}, {"role": "assistant", "content": "## SCENE 1 — INT. DOCTOR'S OFFICE. DAY.\nACTION: It is a typical '50s family practice right out of Norman Rockwell. There is a jar of tongue depressors on the counter\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Pleasantville: It is a typical '50s family practice right out of Norman Rockwell. There is a jar of tongue depressors on the counter, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. It is a typical '50s family practice right out of Norman Rockwell. There is a jar of tongue depressors on the counter The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Mills sits down at his desk. He starts looking through police\nSetting: derived from Seven\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. MILLS' APARTMENT, LIVING ROOM -- NIGHT\nACTION: Mills sits down at his desk. He starts looking through police paperwork. The RADIO in the other room goes OFF in mid song.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Seven: Mills sits down at his desk. He starts looking through police paperwork. The RADIO in the other room goes OFF in mid song., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Mills sits down at his desk. He starts looking through police paperwork. The RADIO in the other room goes OFF in mid song. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Stantz tries to start the car, but the engine won't turn over.\nSetting: derived from Ext. Manhattan Island - Day\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ECTO-1 - DAY (CONTINUOUS ACTION)\nACTION: Stantz tries to start the car, but the engine won't turn over. Give it up, Ray. You're living in the past.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ext. Manhattan Island - Day: Stantz tries to start the car, but the engine won't turn over. Give it up, Ray. You're living in the past., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Stantz tries to start the car, but the engine won't turn over. Give it up, Ray. You're living in the past. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Coraline\n\nSCREENPLAY EXCERPT:\nINT. OTHER HALLWAY - CONTINUOUS\n\n\n\n\n OTHER MOTHER\n You could stay here forever... if you\n want to.\n\n\n\n CORALINE\n Really?\n\n\n\n OTHER FATHER\n Sure...we'll sing and play games, and\n Mother will cook your favorite meals.\n\n\n Other Mother rests her hand on Coraline's shoulder.\n\n\n\n OTHER MOTHER\n There's one tiny little thing we need to\n do.\n\n\n\n CORALINE\n What's that?\n\n\n\n OTHER FATHER\n Well, it's a surprise."}, {"role": "assistant", "content": "## SCENE 1 — INT. OTHER HALLWAY - CONTINUOUS\nACTION: You could stay here forever... if you Sure...we'll sing and play games, and\nDIALOGUE — Other Mother: You could stay here forever... if you want to.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Coraline: You could stay here forever... if you Sure...we'll sing and play games, and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You could stay here forever... if you Sure...we'll sing and play games, and The camera pushes in with small amplitude at slow speed. Other Mother (S1) says: [English] You could stay here forever... if you want to. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Sheriff, may I present a pair of\nSetting: derived from Tombstone\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Sheriff, may I present a pair of Fellow sophisticates, Turkey\nDIALOGUE — White: 'Fraid I'll have to have those guns.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Tombstone: Sheriff, may I present a pair of Fellow sophisticates, Turkey, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sheriff, may I present a pair of Fellow sophisticates, Turkey The camera pushes in with small amplitude at slow speed. White (S1) says: [English] 'Fraid I'll have to have those guns. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Legion\n\nSCREENPLAY EXCERPT:\nEXT. RED ROCK PARK - CONTINUOUS\n\n The Cruiser spins out of control, swerving off the road and\n careening into a large rock formation.\n This ride is over."}, {"role": "assistant", "content": "## SCENE 1 — EXT. RED ROCK PARK - CONTINUOUS\nACTION: The Cruiser spins out of control, swerving off the road and careening into a large rock formation.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Legion: The Cruiser spins out of control, swerving off the road and careening into a large rock formation., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The Cruiser spins out of control, swerving off the road and careening into a large rock formation. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Blade\n\nSCREENPLAY EXCERPT:\nINT. THE BLACK PEARL, ELEVATOR - NIGHT\n\n\tThe elevator descends. Then a tone CHIMES, signaling the end of the\n\tride. The doors hiss open --"}, {"role": "assistant", "content": "## SCENE 1 — INT. THE BLACK PEARL, ELEVATOR - NIGHT\nACTION: The elevator descends. Then a tone CHIMES, signaling the end of the ride. The doors hiss open --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade: The elevator descends. Then a tone CHIMES, signaling the end of the ride. The doors hiss open --, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The elevator descends. Then a tone CHIMES, signaling the end of the ride. The doors hiss open -- The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Bottle Rocket\n\nSCREENPLAY EXCERPT:\nEXT. MOTEL HALLWAY. NIGHT\n\nOutside the room. Dignan closes the door. There's three\nhousekeeping carts in the hall.\n\n DIGNAN\n What the fuck is going on here?\n\n ANTHONY\n What. What's the matter?\n\n DIGNAN\n Anthony, we're on the run from the\n law here. Did you tell these people\n your real name?\n\n ANTHONY\n No. I didn't. Dignan, they don't\n speak English.\n\n DIGNAN\n They don't?\n\n ANTHONY\n No. Not really. Inez speaks a little.\n\n DIGNAN\n Which one was that?\n\n ANTHONY\n On the left.\n\nDignan cracks open the door and looks inside.\n\n ANTHONY\n She's from Cuba.\n\n DIGNAN\n No kidding.\n\n BOB\n Dignan.\n\n DIGNAN\n (closes the door)\n Anthony. Bad news."}, {"role": "assistant", "content": "## SCENE 1 — EXT. MOTEL HALLWAY. NIGHT\nACTION: Outside the room. Dignan closes the door. There's three housekeeping carts in the hall.\nDIALOGUE — Dignan: What the fuck is going on here?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bottle Rocket: Outside the room. Dignan closes the door. There's three housekeeping carts in the hall., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Outside the room. Dignan closes the door. There's three housekeeping carts in the hall. The camera pushes in with small amplitude at slow speed. Dignan (S1) says: [English] What the fuck is going on here? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Jack enters. Space Monkeys render fat and make soap. They\nSetting: derived from F I G H T C L U B\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. KITCHEN - NIGHT\nACTION: Jack enters. Space Monkeys render fat and make soap. They pinch HERBS, adding them to the mix. They add VODKA. Off\nDIALOGUE — Freckled Space Monkey: \"You are not a beautiful and unique snowflake. You are the same decaying organic matter as everything else. We are all part of the same compost heap.\"\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from F I G H T C L U B: Jack enters. Space Monkeys render fat and make soap. They pinch HERBS, adding them to the mix. They add VODKA. Off, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack enters. Space Monkeys render fat and make soap. They pinch HERBS, adding them to the mix. They add VODKA. Off The camera pushes in with small amplitude at slow speed. Freckled Space Monkey (S1) says: [English] \"You are not a beautiful and unique snowflake. You are the same decaying organic matter as everything else. We are all part of the same compost heap.\" The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I shall report this indignity the my\nSetting: derived from Mrs. Teasdale\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I shall report this indignity the my Mrs. Teasdale, I feel this\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Mrs. Teasdale: I shall report this indignity the my Mrs. Teasdale, I feel this, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I shall report this indignity the my Mrs. Teasdale, I feel this The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Bull illustrating the trope 'ObfuscatingInsanity'. On Bull, Hazel Diaz has been a major drug lord for almost 30 years. She brilliantly acts in public like she's mentally ill, even talking to thin air and wild acts. Thus, any time she's arrested, her lawyer argues she's not fit for trial, she goes to a mental ward for a bit then back on the streets. When he's called in for her latest case, Bull is both impressed and amazed at how Diaz has been pull...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying ObfuscatingInsanity.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Bull, ObfuscatingInsanity, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for ObfuscatingInsanity. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Batman\n\nSCREENPLAY EXCERPT:\nINT. WAYNE MANOR - NIGHT\n\n BRUCE and VICKI enter. He takes her coat, drops it on a\n chair by the door. VICKI is giddy, all champagned up.\n\n VICKI\n -- but it's not fair. I'm half\n drunk and you're not even --\n\n BRUCE\n Would you like me to take you home?\n\n VICKI\n God. You would.\n (sidling up to him)\n Come on, Bruce. I just want to get\n two drinks in you. As an\n experiment.\n\n BRUCE\n Maybe we should just kiss.\n\n VICKI\n ... We could try that.\n\n WIDER ANGLE\n\n BRUCE embracing VICKI in the vastness of the darkened entry\n hall, framed by long semicircular STAIRWAYS on opposite\n walls. A SUDDEN FLASH OF LIGHTNING transports us to:"}, {"role": "assistant", "content": "## SCENE 1 — INT. WAYNE MANOR - NIGHT\nACTION: BRUCE and VICKI enter. He takes her coat, drops it on a chair by the door. VICKI is giddy, all champagned up.\nDIALOGUE — Vicki: -- but it's not fair. I'm half drunk and you're not even --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman: BRUCE and VICKI enter. He takes her coat, drops it on a chair by the door. VICKI is giddy, all champagned up., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. BRUCE and VICKI enter. He takes her coat, drops it on a chair by the door. VICKI is giddy, all champagned up. The camera pushes in with small amplitude at slow speed. Vicki (S1) says: [English] -- but it's not fair. I'm half drunk and you're not even -- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The most exquisite hands you've ever\nSetting: derived from The Puffy Shirt\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CLEARS THROAT\nACTION: The most exquisite hands you've ever seen.. Oh, he had it all.\nDIALOGUE — Photographer: He was it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Puffy Shirt: The most exquisite hands you've ever seen.. Oh, he had it all., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The most exquisite hands you've ever seen.. Oh, he had it all. The camera pushes in with small amplitude at slow speed. Photographer (S1) says: [English] He was it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DadsArmy illustrating the trope 'YouGotMurder'. Dad's Army. The platoon is given a mission to test security at an army base, and decide to infiltrate disguised as firemen. As usual things go wrong and they end up fighting a real fire elsewhere. A high-ranking officer arrives and tells Mainwaring not to worry as his 'other plan' worked. Turns out Wilson sent a fake bomb in the mail.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying YouGotMurder.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DadsArmy, YouGotMurder, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for YouGotMurder. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages":[{"role":"system","content":"You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT, DURATION 5."},{"role":"user","content":"Premise: A train passenger folds a letter and looks out at city lights in the rain.\nWrite SCENE 1 now."},{"role":"assistant","content":"## SCENE 1 — INT. TRAIN CARRIAGE - NIGHT\nACTION: A young woman by a rain-covered window lowers a letter and turns her gaze to the passing city lights as the camera trucks slightly.\nDIALOGUE — WOMAN: I get off at the next station.\nSHOT: medium close-up, truck right with small amplitude at slow speed\nSTORYBOARD_PROMPT: young woman at rain-covered train window, night city lights outside, folded letter in hands, carriage interior, cinematic live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, the young woman shown in Picture 1 remains beside the rain-covered train window, preserving her appearance, clothing, seat position, and the carriage layout. The camera trucks right with small amplitude at slow speed as she lifts her gaze from the folded letter toward the passing city lights. The quiet, breathy young woman (S1) says: [English] I get off at the next station. She folds the letter along its crease as the frame settles into Picture 2.\n\noverall_soundscape: Train wheels produce a steady metallic rhythm beneath a low ventilation hum. Rain ticks against the window; paper rustles softly.\n\nnon_diegetic_music: Sustained cello notes at a slow tempo with widely spaced piano tones, gradually decreasing in volume.\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The door to the Empress's room opens into the courtyard. Lots of creatures from Fantasia await help from the Empress. The Empress's head servant comes out of her room and speaks to them all.\nSetting: derived from The Neverending Story\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. VIEW- IVORY TOWER- THE EMPRESSES COURT\nACTION: The door to the Empress's room opens into the courtyard. Lots of creatures from Fantasia await help from the Empress. The Empress's head servant comes out of her room and speaks to them all. Friends, I know why you are all here. The nothing is destroying our world. I also know.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Neverending Story: The door to the Empress's room opens into the courtyard. Lots of creatures from Fantasia await help from the Empress. The Empress's head servant comes out of her room and speaks to them all. Friends, , detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The door to the Empress's room opens into the courtyard. Lots of creatures from Fantasia await help from the Empress. The Empress's head servant comes out of her room and speaks to them all. Friends, I know why you are all here. The nothing is destroying our world. I also know. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: T R O N\n\nSCREENPLAY EXCERPT:\nWait a minute Yori! YORI!! LOOK!\n\n 575 FULL SHOT 575\n\n as Yori runs to Flynn's side and-peers down.\n\n YORI\n TRON!!\n\n 576 CLOSE UP - YORI 576\n\n tears of joy streaming down her face.\n\n 577 HER POV 577\n\n the fight continuing.\n\n\n 577A FULL SHOT - THE BRIDGE 577\n\n\n YORI\n Flynn, we've got to help him!\n\n\n FLYNN\n Let's get some power here.\n\n\n His hands, laying on the control console, begin to send forth great\n pulses of glowing energy into the carrier.\n\n CUT TO:\n\n\n 578 CLOSE UP 578\n\n\n Sark's face.\n\n\n SARK\n You are very persistent, Tron.\n\n\n 579 MED. SHOT 579\n\n\n Tron's disk circles Sark, attacking twice. Sark deflects it and\n immediately throws his own disk.\n\n\n 580 SHOT OF TWO WEAPONS 580\n\n\n racing almost side by side through the air.\n\n\n 581 SHOT OF TRON 581\n\n\n as the two disks come at him.\n\n\n TRON\n I'm also bette"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Wait a minute Yori! YORI!! LOOK! 575 FULL SHOT 575\nDIALOGUE — Yori: TRON!!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T R O N: Wait a minute Yori! YORI!! LOOK! 575 FULL SHOT 575, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Wait a minute Yori! YORI!! LOOK! 575 FULL SHOT 575 The camera pushes in with small amplitude at slow speed. Yori (S1) says: [English] TRON!! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DirtyJobs illustrating the trope 'OfficialFanSubmittedContent'. Dirty Jobs relies on viewers to submit suggestions for jobs to cover. Every episode ends with a clip of Mike Rowe (usually covered in muck) asking for viewers to send in suggestions.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying OfficialFanSubmittedContent.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DirtyJobs, OfficialFanSubmittedContent, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for OfficialFanSubmittedContent. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: What do you want me to say?\nSetting: derived from Black Screen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BACKSTAGE\nACTION: What do you want me to say? Just don't lie to me.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: What do you want me to say? Just don't lie to me., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What do you want me to say? Just don't lie to me. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Hey! What are you doing here? I thought\nSetting: derived from He Phone Message\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Hey! What are you doing here? I thought I was supposed to call you\nDIALOGUE — George: I, I couldn't wait. I was too anxious to see you.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from He Phone Message: Hey! What are you doing here? I thought I was supposed to call you, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hey! What are you doing here? I thought I was supposed to call you The camera pushes in with small amplitude at slow speed. George (S1) says: [English] I, I couldn't wait. I was too anxious to see you. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ------------------------------------------------------------------------\nSetting: derived from \"Hunter\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/21/86 15\nDIALOGUE — Schaefer: Billy's on it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hunter\": ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/21/86 15, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/21/86 15 The camera pushes in with small amplitude at slow speed. Schaefer (S1) says: [English] Billy's on it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dark City\n\nSCREENPLAY EXCERPT:\nEXT. FLYING P.O.V. - NIGHT\n\n A FLYING PERSPECTIVE past buildings. An INSECT-LIKE BUZZING\n O.S. Way down BELOW, in a canyon of silent buildings, a LONE\n FIGURE walks."}, {"role": "assistant", "content": "## SCENE 1 — EXT. FLYING P.O.V. - NIGHT\nACTION: A FLYING PERSPECTIVE past buildings. An INSECT-LIKE BUZZING O.S. Way down BELOW, in a canyon of silent buildings, a LONE\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dark City: A FLYING PERSPECTIVE past buildings. An INSECT-LIKE BUZZING O.S. Way down BELOW, in a canyon of silent buildings, a LONE, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A FLYING PERSPECTIVE past buildings. An INSECT-LIKE BUZZING O.S. Way down BELOW, in a canyon of silent buildings, a LONE The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DoctorWho illustrating the trope 'ClimacticBattleResurrection'. The Tenth Doctor in full Physical God mode at the end of the Doctor Who episode \"Last of The Time Lords\", bringing down the Master, turning time back an entire year, and saving at least a few hundred million people.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying ClimacticBattleResurrection.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DoctorWho, ClimacticBattleResurrection, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for ClimacticBattleResurrection. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Army Of Darkness\n\nSCREENPLAY EXCERPT:\nKeep you damn filthy bones outta\n\t\t\t my mouth.\n\t SNAP! He breaks the boney arm in two and runs away from the sea\n\t of limbs. A bone arm rips from the grave and reaches for the\n\t Book of the Dead but Ash scoops it up first. He leaps over\n\t another set of groping arms that rip from the ground!\n419\tASH CLIMBS\t\t\t\t\t\t\t\t\t\t 419\n\t atop the horse. He glances back in fear to see...\n420\tEXT. CEMETARY - OLD OAK TREE\t\t\t\t\t\t\t 420\n\t The burial site of Evil Ash. A bolt of lightning strikes the\n\t grave marker.\n421\tEVIL ASH'S BURIAL MARKER\t\t\t\t\t\t\t\t421\n\t is thrust from the ground. A hand breaks the surface of the\n\t earth.\n422\tEVIL ASH'S BODY PARTS\t\t\t\t\t\t\t\t 422\n\t fly up from the grave and assemble themselves into a lopsided,\n\t decayed version of EVIL ASH!\n423\tEXT. MILL - NIGHT\t\t\t\t\t\t\t\t\t 423\n\t ASH\n\t stares in horror at his evil self. All around, skeletons rip\n\t from the earth and shriek as they come back to life!\n424\tA FEARFUL ASH\t\t\t\t\t\t\t\t\t\t 424\n\t kicks the horse and rides off.\n425\tEXT. MILL - LONG SHOT - NIGHT\t\t\t\t\t\t\t425\n\t ASH ON HORSEBACK\n\t galloping back the way he came. In the distance we see the\n\t cemetery. More bodies arise from the ground.\n426\tEXT. WOODS - ASH - NIGHT\t\t\t\t\t\t\t\t426\n\t Now far from the danger but still he rides hard.\n\t\t\t\t\t ASH\n\t\t\t I'm through bein' their garbage\n\t\t\t boy. I did my part of the\n\t\t\t bargain.\n\t He pats the saddlebag, where the book is and grins.\n\t\t\t Now they owe me. Like in the\n\t\t\t deal. I want back.\n\t He whips his horse...\n\t\t\t HA! GIIDDUP NOW!!\n\t ...and rides off into the darkness.\n\t\t\t\t\t\t\t\t\t DISSO"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Keep you damn filthy bones outta SNAP! He breaks the boney arm in two and runs away from the sea\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Army Of Darkness: Keep you damn filthy bones outta SNAP! He breaks the boney arm in two and runs away from the sea, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Keep you damn filthy bones outta SNAP! He breaks the boney arm in two and runs away from the sea The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ANGLE ON: THEODEN looks at EOWYN who stands as if turned to\nSetting: derived from The Lord Of The Rings: The Return Of The King\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. THEODEN'S TENT, DUNHARROW PLATEAU - DAWN\nACTION: ANGLE ON: THEODEN looks at EOWYN who stands as if turned to stone ... she is looking up at a LIGHTENING SKY ... The QUIET\nDIALOGUE — Theoden: I have left instruction: the people are to follow your rule in my stead ... take up my seat in the Golden Hall ... long may you defend Edoras if the battle...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Return Of The King: ANGLE ON: THEODEN looks at EOWYN who stands as if turned to stone ... she is looking up at a LIGHTENING SKY ... The QUIET, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ANGLE ON: THEODEN looks at EOWYN who stands as if turned to stone ... she is looking up at a LIGHTENING SKY ... The QUIET The camera pushes in with small amplitude at slow speed. Theoden (S1) says: [English] I have left instruction: the people are to follow your rule in my stead ... take up my seat in the Golden Hall ... long may you defend Edoras if the battle... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Untitled\n\nSCREENPLAY EXCERPT:\nWhat's that? A celebrity cooking show for\nthe new season? :\n\n\nHARLAN -\nNo. It's my favourite episode of Videodrome.\nThe one that stars Max Renn. —\n\n\nMax looks at the cassette as it lies on Harlan's desk. Is it\nbreathing? Max looks away. He doesn't want it to be breathing.\n\n\nMAX\nThat show was transmitted? You brought\nit in on the dish?\n\n\nHARLAN\nNot exactly. After all, I wouldn't have\n\n\nrecognized you in that kinky outfit\n` were wearing, now would I? ae you\n\n\nGr 37\n\n\ni MAX\nIt wasn't an accident that we pulled in\nVideodrome. You made it happen. They\nSent you to me. You've been with them all\nong. .\n\n\nHarlan Picks up the cassette and moves around from behind the\ndesk towards Max. The cassette definitely contracts when\nlan touches it. Max flattens against the wall, paralysed.\n\n\nHARLAN\nWe didn't even pull it in, Max.\nIt was playing through an Accumicon -\nplayback module we installed in the\nfurnace room next to the lab. It's\ntoo dangerous to transmit. Much too >\ndangerous. .. for now. 7 - `\n\n\nMAX\nBut why? Why go through all that?\n\n\nHARLAN -\nWe needed a guiried pig. Somebody\nwho would respond to Videodrome.\n- And we want Channel 83. It's small,\n\n\n-but it's a start.\n\n\nHarlan is very close, now. Very close. He hefts the weight of\nthe cassette in his hand. It stretches and pulses impatiently.\nMax can hardly breathe. :\n\n\nMAX\nWhat do you...what do you want from me?\n\n\nHarlan starts to unbutton Max's shirt. Max is frozen against .\nthe wall, splayed out as though riding a centrifuge spinning\n\nat top velocity. His hair seems to be blowing back. Video dandruff\n\nseems to be raining down on bot"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: What's that? A celebrity cooking show for No. It's my favourite episode of Videodrome.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: What's that? A celebrity cooking show for No. It's my favourite episode of Videodrome., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What's that? A celebrity cooking show for No. It's my favourite episode of Videodrome. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Enough of that, or I'll forget to\nSetting: derived from Les Miserables\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Enough of that, or I'll forget to You heard me ask for something,\nDIALOGUE — Customer 1: Mine host Thenardier He was there, so they say, At the field of Waterloo.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Les Miserables: Enough of that, or I'll forget to You heard me ask for something,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Enough of that, or I'll forget to You heard me ask for something, The camera pushes in with small amplitude at slow speed. Customer 1 (S1) says: [English] Mine host Thenardier He was there, so they say, At the field of Waterloo. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: As Credits Roll\n\nSCREENPLAY EXCERPT:\nINT. MR. EDDY'S 600 PULLMAN - DAY\n\nMr. Eddy rotates his neck - shaking out some adrenaline\nremnants. Pete is visibly shaken by what he has seen.\n\n\t\t\t\tMR. EDDY\n\t\tSorry about that, Pete, but tailgating is\n\t\tone thing I can't tolerate.\n\n\t\t\t\tPETE\n\t\tI can see that.\n\n\t\t\t\tMR. EDDY\n\t\tI'll bet you know how many car lengths it\n\t\ttakes to stop at... say 45 m.p.h.\n\n\t\t\t\tPETE\n\t\tEight, nine car lengths. A hundred and\n\t\tsixty-two feet.\n\n\t\t\t\tMR. EDDY\n\t\tAt sixty?\n\n\t\t\t\tPETE\n\t\tFifteen car lengths. About two hundred\n\t\tand seventy feet.\n\n\t\t\t\tMR. EDDY\n\t\t\t(to his companions in the\n\t\t\treariew)\n\t\tWhat'd I tell ya.\n\nMr. Eddy smiles and they continue along their way.\n\n\t\t\t\t\t\t\t\tCUT TO:\n\nEXT./INT. ARNIE'S GARAGE - DAY\n\nThe Pullman pulls off the street into Arnie's. Mr. Eddy and\nhis assistants stay in the car. Pete opens his door.\n\n\t\t\t\tMR. EDDY\n\t\tWait a minute ...\n\nMr. Eddy takes out a couple of one hundred dollar bills and\nputs them into Pete's hand.\n\n\t\t\t\tPETE\n\t\tThanks, Mr. Eddy.\n\n\t\t\t\tMR. EDDY\n\t\tNo... Thank you!... I'll be bringin' the\n\t\t\"Caddy\" by tomorrow.\n\nMr. Eddy reaches into the glove compartment and takes out a\nvideotape. He holds it out to Pete.\n\n\t\t\t\tMR. EDDY (CON'T)\n\t\tYou like pornos?\n\n\t\t\t\tPETE\n\t\tPornos?\n\n\t\t\t\tMR. EDDY\n\t\tYeah. Give ya a boner.\n\n\t\t\t\tPETE\n\t\tNo thanks, Mr. Eddy.\n\nMr. Eddy tosses the tape back in the glove compartment.\n\n\t\t\t\tMR. EDDY\n\t\tSuit yourself, champ.\n\n\t\t\t\tPETE\n\t\tOkay... Well, I'll see ya then.\n\n\t\t\t\tMR. EDDY\n\t\tYou will.\n\n\t\t\t\t\t\t\t\tCUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — INT. MR. EDDY'S 600 PULLMAN - DAY\nACTION: Mr. Eddy rotates his neck - shaking out some adrenaline remnants. Pete is visibly shaken by what he has seen.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from As Credits Roll: Mr. Eddy rotates his neck - shaking out some adrenaline remnants. Pete is visibly shaken by what he has seen., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Mr. Eddy rotates his neck - shaking out some adrenaline remnants. Pete is visibly shaken by what he has seen. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade Out:\n\nSCREENPLAY EXCERPT:\nInt. Tower Room -- Close on the hour glass on the table --\nCAMERA PULLS back to reveal Dorothy standing by the table as\nshe watches the sand run through the glass -- she looks about\ndesperately --\n\n LAP DISSOLVE TO:\n\nELS -- The Tin Man, Scarecrow and Lion making their way over\nrocky hillside -- Toto barks as he waits for them\n\n LAP DISSOLVE TO:\n\nCU -- Toto makes his way over top of the rocks -- barks --\nexits out to right -- MS -- The Lion, Tin Man and Scarecrow\nstruggling up the rocks -- the Tin Man slips -- ELS -- The\nTin Man slips off the side of the rocks -- hangs to the\nLion's tail to keep from falling -- the Scarecrow tries to\nhelp him --\n\n LION\n Oh -- Oh -- Oh -- I....\n\nMS -- the Tin Man hanging to Lion's tail -- they speak --\nScarecrow tries to get up to help them both --\n\n LION\n ...I -- I -- I hope my strength holds\n out.\n\n TIN MAN\n I hope your tail holds out. Oh --\n\n LAP DISSOLVE TO:\n\n\n (CONTINUED)\n\n 99.\nCONTINUED: (7)\n\n\nCU - The Hour Glass - more sand in the bottom - MCS - In.\nTower Room - Dorothy watching the hour glass - CAMERA PANS\nwith her as she runs to the door, tries it - then runs to\nanother door - finds that locked, also - CAMERA TRUCKS\nforward slightly as she sobs, then turns and comes slowly\nforward as the CAMERA PULLS back"}, {"role": "assistant", "content": "## SCENE 1 — INT. TOWER ROOM -- CLOSE ON THE HOUR GLASS ON THE TABLE --\nACTION: CAMERA PULLS back to reveal Dorothy standing by the table as she watches the sand run through the glass -- she looks about\nDIALOGUE — Lion: Oh -- Oh -- Oh -- I....\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade Out:: CAMERA PULLS back to reveal Dorothy standing by the table as she watches the sand run through the glass -- she looks about, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CAMERA PULLS back to reveal Dorothy standing by the table as she watches the sand run through the glass -- she looks about The camera pushes in with small amplitude at slow speed. Lion (S1) says: [English] Oh -- Oh -- Oh -- I.... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Suddenly the wall behind Alia and the spinning Baron\nSetting: derived from Dune\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. EMPEROR'S SPACE SHIP - GREEN THRONE ROOM - DAY\nACTION: Suddenly the wall behind Alia and the spinning Baron is completely blown away by a BLAST.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dune: Suddenly the wall behind Alia and the spinning Baron is completely blown away by a BLAST., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Suddenly the wall behind Alia and the spinning Baron is completely blown away by a BLAST. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in ArtemisFowl illustrating the trope 'TooDumbToLive'. Goblins in general. In The Arctic Incident , our heroes are saved by a group of goblin gangsters pulling a Klingon Promotion and killing themselves off to become leader during a combat mission . In The Opal Deception , a goblin attempts to escape capture by flying into a pressure chute with a jetpack moments before a magma flare.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TooDumbToLive.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for ArtemisFowl, TooDumbToLive, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TooDumbToLive. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in HyakujuuSentaiGaoranger illustrating the trope 'AnimalMotifTeam'. Super Sentai and its American adaptation Power Rangers have a lot of these, and we do mean a lot. Hell, if there's a Power Rangers parody, almost always expect them to have an Animal Motif:\n Dengeki Sentai Changeman: Has a team with mythological animals as motif, consisting of Dragon (Hiry Tsurugi/Red), Griffin (Sh Hayate/Black), Pegasus (Yma zora/Blue), Mermaid (Sayaka Nagisa/White), and Phoenix ...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying AnimalMotifTeam.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for HyakujuuSentaiGaoranger, AnimalMotifTeam, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for AnimalMotifTeam. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Seth eats a burger and chili fries at the same time.\nSetting: derived from American History X\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BURGER SHOP - CAMHERON AND SETH\nACTION: Seth eats a burger and chili fries at the same time. Cameron smokes across from him and watches Seth eat like\nDIALOGUE — Cammeron: I think I need to go to the hospital.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American History X: Seth eats a burger and chili fries at the same time. Cameron smokes across from him and watches Seth eat like, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Seth eats a burger and chili fries at the same time. Cameron smokes across from him and watches Seth eat like The camera pushes in with small amplitude at slow speed. Cammeron (S1) says: [English] I think I need to go to the hospital. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Good News, Bad News\n\nSCREENPLAY EXCERPT:\nKRAMER\n (enters) Are you up?\n\n\n JERRY\n (To Kramer) Yeah...(in the phone) Yeah,\n people do move! Have you ever seen the\n big trucks out on the street? Yeah,\n no problem (hangs up the phone).\n\n\n KRAMER\n Boy, the Mets blew it tonight, huh?\n\n\n\n JERRY\n (upset) Ooohhhh, what are you doing?\n Kramer, it's a tape! (actually I can't\n understand the name; maybe \"Kramer\"\n wasn't named \"Kramer\" in the pilot!)\n I taped the game, it's one o'clock in\n the morning! I avoided human contact\n all night to watch this.\n\n\n KRAMER\n Hey, I'm sorry, I...ya know, I, I thought\n you knew...(takes two loaves of bread\n out of his pockets) You got any meat?\n\n\n\n JERRY\n (a little irritated) Meat? I don't,\n I don't know, go...hunt! (Kramer walks\n to the refrigerator and sticks his head\n in) Well what, what happened in the\n game anyway?\n\n\n KRAMER"}, {"role": "assistant", "content": "## SCENE 1 — INT. ENTERS\nACTION: (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the\nDIALOGUE — Jerry: (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the big trucks out on the street? Yeah, no problem (hangs up the phone).\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Good News, Bad News: (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] (To Kramer) Yeah...(in the phone) Yeah, people do move! Have you ever seen the big trucks out on the street? Yeah, no problem (hangs up the phone). The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Heavenly Creatures\n\nSCREENPLAY EXCERPT:\nEXT. BOROVNIA CASTLE COURTYARD - DAY. . . the courtyard. A f�te is in progress. The courtyard is full of PLASTICINE PEOPLE, laughing and singing amidst stalls and sideshows.\n\nPauline pushes her way through the crowd. A Plasticine finger taps her shoulder . . . she spins around and gasps!\n\nPAULINE\nCharles!\n\nThe PLASTICINE KING smiles.\n\nCHARLES (James Mason voice) Hello, Gina . . . great to see you here!\n\nHe turns away. Pauline scans the crowd. She gets a fleeting glimpse of the back of a ROBED WOMAN.\n\nPAULINE\n(Calling) Deborah!\n\nDeborah does not hear, and Pauline desperately pushes between Plasticine people to try and reach her.\n\nA blur of Plasticine faces slide past . . .\n\nCRASH CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. BOROVNIA CASTLE COURTYARD - DAY. . . THE COURTYARD. A F�TE IS IN PROGRESS. THE COURTYARD IS FULL OF PLASTICINE PEOPLE, LAUGHING AND SINGING AMIDST STALLS AND SIDESHOWS.\nACTION: Pauline pushes her way through the crowd. A Plasticine finger taps her shoulder . . . she spins around and gasps! The PLASTICINE KING smiles.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Heavenly Creatures: Pauline pushes her way through the crowd. A Plasticine finger taps her shoulder . . . she spins around and gasps! The PLASTICINE KING smiles., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Pauline pushes her way through the crowd. A Plasticine finger taps her shoulder . . . she spins around and gasps! The PLASTICINE KING smiles. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Hustle illustrating the trope 'SexForServices'. \nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying SexForServices.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Hustle, SexForServices, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for SexForServices. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The both of you should go home. We\nSetting: derived from Black Screen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. X-FORCE WAR ROOM\nACTION: The both of you should go home. We will keep you abreast.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: The both of you should go home. We will keep you abreast., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The both of you should go home. We will keep you abreast. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Natural Born Killers\"\n\nSCREENPLAY EXCERPT:\nINT. GRACE'S HOME - NIGHT\n\n CAMERA is LOW ANGLE as Tim backs up after his attack. We notice\n the High School Ring he wears on his right fist as he steps into\n a fighting stance, ready for Mickey to drop.\n\n Mickey, however, spits out a stream of blood, and smiles at him.\n\n CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — INT. GRACE'S HOME - NIGHT\nACTION: CAMERA is LOW ANGLE as Tim backs up after his attack. We notice the High School Ring he wears on his right fist as he steps into\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Natural Born Killers\": CAMERA is LOW ANGLE as Tim backs up after his attack. We notice the High School Ring he wears on his right fist as he steps into, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CAMERA is LOW ANGLE as Tim backs up after his attack. We notice the High School Ring he wears on his right fist as he steps into The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Angel illustrating the trope 'TortureTechnician'. Buffy the Vampire Slayer:\n Faith was quite fond of torture - at one point, Angel complimented her on how well she knew how to do it. One wonders how a teenager knew so much about the five basic torture groups.\n Angelus himself, the soulless version of Angel, an aficionado of physical and psychological torture (as he put it \"The last time I tortured someone, they didn't even have chainsaws.\"). Ange...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TortureTechnician.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Angel, TortureTechnician, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TortureTechnician. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: As Credits Roll\n\nSCREENPLAY EXCERPT:\nEXT. DESERT HIGHWAY - DAY\n\nFred is behind the wheel of his car. Behind him are the\ncops.\n\nCLOSE UP ON FRED'S FACE - through the front windshield. The\nshot is wide enough to reveal the cop car in hot pursuit.\n\nFred's face begins to change again, grotesquely contorting as\nhe races into the vortex. His tormented SCREAM blends with\nthe howling SIREN of the police car gaining on him.\n\n\n\n\t\t\t\tTHE END"}, {"role": "assistant", "content": "## SCENE 1 — EXT. DESERT HIGHWAY - DAY\nACTION: Fred is behind the wheel of his car. Behind him are the CLOSE UP ON FRED'S FACE - through the front windshield. The\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from As Credits Roll: Fred is behind the wheel of his car. Behind him are the CLOSE UP ON FRED'S FACE - through the front windshield. The, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Fred is behind the wheel of his car. Behind him are the CLOSE UP ON FRED'S FACE - through the front windshield. The The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Aye. Cry as you may, cry as you\nSetting: derived from Bride Of Leprechaun\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Aye. Cry as you may, cry as you might. It's going to be one hell of\nDIALOGUE — Cody: In your dreams, buddy.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bride Of Leprechaun: Aye. Cry as you may, cry as you might. It's going to be one hell of, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Aye. Cry as you may, cry as you might. It's going to be one hell of The camera pushes in with small amplitude at slow speed. Cody (S1) says: [English] In your dreams, buddy. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Meanwhile in a grimy narrow control booth high above the factory floor,\nSetting: derived from Buckaroo Banzai\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SPEED OPERATOR'S PULPIT (YOYODYNE) - NIGHT\t318\nACTION: Meanwhile in a grimy narrow control booth high above the factory floor, a terrified Penny Priddy sits BOUND and GAGGED, John O'Connor SPREADING\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Buckaroo Banzai: Meanwhile in a grimy narrow control booth high above the factory floor, a terrified Penny Priddy sits BOUND and GAGGED, John O'Connor SPREADING, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Meanwhile in a grimy narrow control booth high above the factory floor, a terrified Penny Priddy sits BOUND and GAGGED, John O'Connor SPREADING The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Austin Powers: International Man Of Mystery\n\nSCREENPLAY EXCERPT:\nINT. TEXAN'S STALL\n\n TEXAN\n That's right! Show that turd who's\n boss!"}, {"role": "assistant", "content": "## SCENE 1 — INT. TEXAN'S STALL\nACTION: That's right! Show that turd who's\nDIALOGUE — Texan: That's right! Show that turd who's boss!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Austin Powers: International Man Of Mystery: That's right! Show that turd who's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. That's right! Show that turd who's The camera pushes in with small amplitude at slow speed. Texan (S1) says: [English] That's right! Show that turd who's boss! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Black Screen\n\nSCREENPLAY EXCERPT:\nINT. TENEMENT BUILDING - NIGHT\n\nManray screws out the bare light bulb which is the sole\nsource of illumination and lays down onto his mattress on\nthe floor."}, {"role": "assistant", "content": "## SCENE 1 — INT. TENEMENT BUILDING - NIGHT\nACTION: Manray screws out the bare light bulb which is the sole source of illumination and lays down onto his mattress on\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: Manray screws out the bare light bulb which is the sole source of illumination and lays down onto his mattress on, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Manray screws out the bare light bulb which is the sole source of illumination and lays down onto his mattress on The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Event Horizon\"\n\nSCREENPLAY EXCERPT:\nINT. EVENT HORIZON - SECOND CONTAINMENT (FOR VIDEO)\n\n Second Containment. A lone engineer finishes his check of\n the Core. He turns to the camera and gives a self-conscious\n \"thumbs-up.\"\n\n KILPACK (O.S.)\n When you get this message, God\n willing, we will reach the solar\n system of Proxima Centauri..."}, {"role": "assistant", "content": "## SCENE 1 — INT. EVENT HORIZON - SECOND CONTAINMENT (FOR VIDEO)\nACTION: Second Containment. A lone engineer finishes his check of the Core. He turns to the camera and gives a self-conscious\nDIALOGUE — Kilpack: When you get this message, God willing, we will reach the solar system of Proxima Centauri...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Event Horizon\": Second Containment. A lone engineer finishes his check of the Core. He turns to the camera and gives a self-conscious, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Second Containment. A lone engineer finishes his check of the Core. He turns to the camera and gives a self-conscious The camera pushes in with small amplitude at slow speed. Kilpack (S1) says: [English] When you get this message, God willing, we will reach the solar system of Proxima Centauri... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Crank\n\nSCREENPLAY EXCERPT:\nEXT. DOWN THE BLOCK - DAY\n 83 83\n A POLICE MOTORCYCLE is tipped over on a lawn, coughing up\n black smoke, spewing oil. A crowd of Mexicans are gathered\n around, gaping.\n\n lames start to SHOOT UP from the motorcycle; THE CROWD\n SCATTERS, ducking for cover.\n\n EXT L.A. CHINATOWN, 8 MINUTES LATER\n 4 84\n MONTAGE: Chinatown is bustling with activity. Vendors haggle\n ... workers hustle down the sidewalk with baskets of\n chickens, sides of meat ... tourists wander ... motorists\n argue and punch their horns ... and the lunch hour crowd\n converges on a hundred eateries ...\n\n e hear a million HEARTBEATS, old, young ... even fast\n ticking chicken heartbeats, all overlapping, blending\n together in a swelling din of live things.\n\n The voyeuristic CAMERA picks CHEV and EVE up through the\n crowd. CHEV is wearing DARK GLASSES to go with his 80's hair\n E\n and jogger.\n\n VE\n You're embarrassing.\n\n CHEV\n You know, I didn't have a lot of time to\n pick this out ...\n\n EVE\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. DOWN THE BLOCK - DAY\nACTION: 83 83 A POLICE MOTORCYCLE is tipped over on a lawn, coughing up\nDIALOGUE — Ve: You're embarrassing.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Crank: 83 83 A POLICE MOTORCYCLE is tipped over on a lawn, coughing up, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. 83 83 A POLICE MOTORCYCLE is tipped over on a lawn, coughing up The camera pushes in with small amplitude at slow speed. Ve (S1) says: [English] You're embarrassing. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Star Wars\n\nSCREENPLAY EXCERPT:\nEXT. DEATH STAR\n\n Alderaan looms behind the Death Star battlestation."}, {"role": "assistant", "content": "## SCENE 1 — EXT. DEATH STAR\nACTION: Alderaan looms behind the Death Star battlestation.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Star Wars: Alderaan looms behind the Death Star battlestation., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Alderaan looms behind the Death Star battlestation. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jay And Silent Bob Strike Back\n\nSCREENPLAY EXCERPT:\nEXT. BALCONY--DAY\n\n A monkey dressed like Mussolini addresses a huge crowd of\n apes, who wave fists in the air.\n\n JAY (V.O.)\n Then all it'll take is one little\n monkey in a spiffy suit to whip the\n dumber chimps into a frenzy, until\n they go all ape-shit and start\n demanding more bananas, better pay,\n and human flesh!"}, {"role": "assistant", "content": "## SCENE 1 — EXT. BALCONY--DAY\nACTION: A monkey dressed like Mussolini addresses a huge crowd of apes, who wave fists in the air.\nDIALOGUE — Jay: Then all it'll take is one little monkey in a spiffy suit to whip the dumber chimps into a frenzy, until they go all ape-shit and start demanding more bananas,...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jay And Silent Bob Strike Back: A monkey dressed like Mussolini addresses a huge crowd of apes, who wave fists in the air., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A monkey dressed like Mussolini addresses a huge crowd of apes, who wave fists in the air. The camera pushes in with small amplitude at slow speed. Jay (S1) says: [English] Then all it'll take is one little monkey in a spiffy suit to whip the dumber chimps into a frenzy, until they go all ape-shit and start demanding more bananas,... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Airplane\n\nSCREENPLAY EXCERPT:\nINT. PASSENGER CABIN - ANOTHER AREA - NIGHT\n\n Jack is seated across the aisle from Mrs. Hammen. He falls\n asleep and slumps over.\n\n MRS. HAMMEN\n (hysterically)\n He's dead. He's dead!\n\n Jack wakes up. Disgustedly:\n\n JACK\n No, I'm not dead."}, {"role": "assistant", "content": "## SCENE 1 — INT. PASSENGER CABIN - ANOTHER AREA - NIGHT\nACTION: Jack is seated across the aisle from Mrs. Hammen. He falls asleep and slumps over.\nDIALOGUE — Mrs. Hammen: He's dead. He's dead!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Airplane: Jack is seated across the aisle from Mrs. Hammen. He falls asleep and slumps over., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack is seated across the aisle from Mrs. Hammen. He falls asleep and slumps over. The camera pushes in with small amplitude at slow speed. Mrs. Hammen (S1) says: [English] He's dead. He's dead! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Ninja Assassin\n\nSCREENPLAY EXCERPT:\nGOLDENROD REVISED 2nd DRAFT 6/4/08 43.\n47 CONTINUED: (2) 47\n\n MIKA\n No...not right now.\n\n ZABRANSKI\n Has Agent Maslow seemed different\n lately, in any way? Has he done\n or said anything that struck you\n as unusual?\n\n MIKA\n No why. Has Agent Maslow done\n something?\n\n He smiles in the way a piranha might smile if it could.\n\n ZABRANSKI\n Just routine. The agency cares\n about its field operatives. We\n monitor for signs of stress,\n fatigue, mental exhaustion. Agent\n Maslow has a serious case load.\n That much work can wear on a man.\n\n He leaves.\n\n She gets up and closes her door. She grabs her phone and\n calls Maslow.\n\n MASLOW (V.O.)\n This is Ryan Maslow. Leave a\n number-\n\n She disconnects then texts.\n\n TEXT: 911. CALL ME. M\n\n\n48 INT. PARKING GARAGE - NIGHT 48\n\n It is later. The garage is a gyre of shadows. Mika\n walks towards her car.\n\n The fluorescent light flickers and goes out. She stands\n in the dark.\n\n MIKA\n Come on...you're kidding me.\n\n It flickers back on. She continues t"}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: GOLDENROD REVISED 2nd DRAFT 6/4/08 43. 47 CONTINUED: (2) 47\nDIALOGUE — Mika: No...not right now.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ninja Assassin: GOLDENROD REVISED 2nd DRAFT 6/4/08 43. 47 CONTINUED: (2) 47, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GOLDENROD REVISED 2nd DRAFT 6/4/08 43. 47 CONTINUED: (2) 47 The camera pushes in with small amplitude at slow speed. Mika (S1) says: [English] No...not right now. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Rear Window\"\n\nSCREENPLAY EXCERPT:\nEXT. NEIGHBORHOOD - DAY - SEMI-LONG SHOT\n\n The man enters the bedroom. We can see a woman lying on the\n far bed. Near her, a small table is covered with medicine\n bottles, spoons, boxes of pills, a water pitcher and the\n other impedimenta of the chronically ill. The woman sits up\n as the man enters. She takes a wet cloth off her forehead.\n Before the man even reaches her, she begins talking, somewhat\n vigorously. Pointing to a wristwatch, she seems to be saying\n something such as \"You should have been home two hours ago!\n I could be lying here dying for all you'd know -- or care!\"\n The man stops short of the bed, makes gestures of trying to\n placate her, but she goes on scolding. His attitude changes\n to weary patience, then irritation, then anger.\n\n He shouts back at her, turns and goes out of the room.\n\n Back in the living room, he picks up his hat, throws it\n against the wall in anger, and leaves the apartment, slamming\n the door behind him."}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEIGHBORHOOD - DAY - SEMI-LONG SHOT\nACTION: The man enters the bedroom. We can see a woman lying on the far bed. Near her, a small table is covered with medicine\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Rear Window\": The man enters the bedroom. We can see a woman lying on the far bed. Near her, a small table is covered with medicine, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The man enters the bedroom. We can see a woman lying on the far bed. Near her, a small table is covered with medicine The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Pauline pushes her way through the crowd. A Plasticine finger taps her shoulder . . . she spins around and gasps!\nSetting: derived from Heavenly Creatures\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BOROVNIA CASTLE COURTYARD - DAY. . . THE COURTYARD. A F�TE IS IN PROGRESS. THE COURTYARD IS FULL OF PLASTICINE PEOPLE, LAUGHING AND SINGING AMIDST STALLS AND SIDESHOWS.\nACTION: Pauline pushes her way through the crowd. A Plasticine finger taps her shoulder . . . she spins around and gasps! The PLASTICINE KING smiles.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Heavenly Creatures: Pauline pushes her way through the crowd. A Plasticine finger taps her shoulder . . . she spins around and gasps! The PLASTICINE KING smiles., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Pauline pushes her way through the crowd. A Plasticine finger taps her shoulder . . . she spins around and gasps! The PLASTICINE KING smiles. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: S T R A N G E D A Y S\n\nSCREENPLAY EXCERPT:\nEXT. CORAL LOUNGE/PARKING LOT - NIGHT\n\n Lenny and Iris come out the front door.\n\n IRIS\n It's a bad situation... if they get\n me I know they're going to -- OH\n SHIT!\n\n She bridles like a startled horse. Lenny looks at what\n she sees: a red beam, sweeping the parking lot. Must be\n the cops. He turns back...\n\n She's gone. What the hell?\n\n He looks back at the red light... and edges around the\n corner to see what the cops are up to...\n\n It's not a cop car. It's a tow-truck, with Lenny's BMW on\n the hook. Lenny runs to the TOW DRIVER, who looks like a\n biker only meaner.\n\n LENNY\n Hang on, that's my car--\n\n TOW DRIVER\n Not anymore. Belongs to the bank.\n\n LENNY\n Hey wait a second...\n\n Ignoring him, the driver has started the hydraulic lift,\n and Lenny makes the mistake of grabbing his beefy arm.\n The guy whips around, putting the muzzle of a .38 in\n Lenny's face.\n\n LENNY\n Oh, yeah, that's the answer! Two\n million years of human evolution and\n that's the best idea you can come up\n with?\n (driver continues\n with the hoist)\n Okay. Look, whattya get to repo a\n car? Two hundred? Two fifty? I'll\n pay you three fifty, right now. All\n you gotta do is drop it off the hook\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. CORAL LOUNGE/PARKING LOT - NIGHT\nACTION: Lenny and Iris come out the front door. It's a bad situation... if they get\nDIALOGUE — Iris: It's a bad situation... if they get me I know they're going to -- OH SHIT!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from S T R A N G E D A Y S: Lenny and Iris come out the front door. It's a bad situation... if they get, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Lenny and Iris come out the front door. It's a bad situation... if they get The camera pushes in with small amplitude at slow speed. Iris (S1) says: [English] It's a bad situation... if they get me I know they're going to -- OH SHIT! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The kids take off towards the bathroom, running through the rain. Grant\nSetting: derived from Jurassic Park\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The kids take off towards the bathroom, running through the rain. Grant gets out of his cruiser, strides to Regis. He indicates the fence.\nDIALOGUE — Grant: Is that still the T-Rex paddock?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jurassic Park: The kids take off towards the bathroom, running through the rain. Grant gets out of his cruiser, strides to Regis. He indicates the fence., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The kids take off towards the bathroom, running through the rain. Grant gets out of his cruiser, strides to Regis. He indicates the fence. The camera pushes in with small amplitude at slow speed. Grant (S1) says: [English] Is that still the T-Rex paddock? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nWhat the hell you want to go there for?\n\n\t\t\t\tLEONARD\n\t\tYou know it?\n\n\t\t\t\tTEDDY\n\t\tYeah, it's just this fucked-up building.\t\t *\n\t\tWhy are we going there?\n\n\t\t\t\tLEONARD"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: What the hell you want to go there for? Yeah, it's just this fucked-up building.\t\t *\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: What the hell you want to go there for? Yeah, it's just this fucked-up building.\t\t *, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What the hell you want to go there for? Yeah, it's just this fucked-up building.\t\t * The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Event Horizon\"\n\nSCREENPLAY EXCERPT:\nINT. LEWIS AND CLARK - BRIDGE\n\n They stare at Justin's monitor.\n\n SMITH\n What the hell is that?\n\n WEIR\n That's the Core: the gravity drive.\n The heart of the ship."}, {"role": "assistant", "content": "## SCENE 1 — INT. LEWIS AND CLARK - BRIDGE\nACTION: They stare at Justin's monitor. What the hell is that?\nDIALOGUE — Smith: What the hell is that?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Event Horizon\": They stare at Justin's monitor. What the hell is that?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They stare at Justin's monitor. What the hell is that? The camera pushes in with small amplitude at slow speed. Smith (S1) says: [English] What the hell is that? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Shaken, but oddly transcendent, Catherine watches as\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CAMPBELL CENTER - LABORATORY\nACTION: Shaken, but oddly transcendent, Catherine watches as Stargher's body is wheeled out of the lab on a gurney. She\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: Shaken, but oddly transcendent, Catherine watches as Stargher's body is wheeled out of the lab on a gurney. She, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Shaken, but oddly transcendent, Catherine watches as Stargher's body is wheeled out of the lab on a gurney. She The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Avengers: Endgame\n\nSCREENPLAY EXCERPT:\nINT. AVENGERS COMPOUND, OFFICE - DAY, EARLY\n\nTony, Natasha, and Smart Hulk hash it out.\n\n NATASHA\n So this `Time Stone' guy...what\n kind of doctor was he?\n\n SMART HULK\n Strange?\n\n TONY\n Cross between ear-nose-and-throat\n and rabbit-from-hat.\n\n SMART HULK\n Had a nice place in the Village.\n\n TONY\n Yeah. Sullivan Street.\n 55\n\n\n SMART HULK\n Bleecker.\n\n TONY\n Sullivan and Bleecker. They're\n cross streets.\n\n NATASHA\n Wait, he lived in New York? Guys,\n pick the right year and there were\n three stones in New York.\n\n SMART HULK\n (beat)\n Shut the front door..."}, {"role": "assistant", "content": "## SCENE 1 — INT. AVENGERS COMPOUND, OFFICE - DAY, EARLY\nACTION: Tony, Natasha, and Smart Hulk hash it out. So this `Time Stone' guy...what\nDIALOGUE — Natasha: So this `Time Stone' guy...what kind of doctor was he?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Avengers: Endgame: Tony, Natasha, and Smart Hulk hash it out. So this `Time Stone' guy...what, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Tony, Natasha, and Smart Hulk hash it out. So this `Time Stone' guy...what The camera pushes in with small amplitude at slow speed. Natasha (S1) says: [English] So this `Time Stone' guy...what kind of doctor was he? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages":[{"role":"system","content":"You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment + integrated_multimodal_description + overall_soundscape + non_diegetic_music), DURATION 5."},{"role":"user","content":"Premise: A baker opens a street bakery before sunrise and places the first loaf on the counter.\nTone: warm, quiet\nWrite SCENE 1 now."},{"role":"assistant","content":"## SCENE 1 — EXT./INT. STREET BAKERY - PRE-DAWN\nACTION: A middle-aged baker opens the shutters, then the camera moves inside as they set a warm loaf on the wooden counter.\nDIALOGUE — BAKER: First batch of the morning.\nSHOT: medium-wide, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: pre-dawn street bakery, baker opening wooden shutters, warm interior light, fresh bread, live-action cinematic, quiet mood\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium-wide shot frames a baker opening the shutters of a small street bakery before sunrise, matching Picture 1. The camera pushes in with small amplitude at slow speed as the middle-aged baker with a calm, slightly raspy voice (S1) places a fresh loaf on the wooden counter and says: [English] First batch of the morning. Steam rises and the composition settles into the close warm counter view of Picture 2.\n\noverall_soundscape: Wooden shutters scrape open over a quiet street; trays clink softly; bread crust crackles as it is set down.\n\nnon_diegetic_music: Soft acoustic-guitar pattern at a moderate tempo with sparse upright-bass notes and a gentle fade.\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The police car is parked right behind Ecto-2. Spengler stops at the\nSetting: derived from Ext. Manhattan Island - Day\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ARMAND'S RESTAURANT - NIGHT (CONTINUOUS ACTION)\nACTION: The police car is parked right behind Ecto-2. Spengler stops at the police car and confronts the cops.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ext. Manhattan Island - Day: The police car is parked right behind Ecto-2. Spengler stops at the police car and confronts the cops., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The police car is parked right behind Ecto-2. Spengler stops at the police car and confronts the cops. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: A Nightmare On Elm Street\n\nSCREENPLAY EXCERPT:\nYou believe it's possible to dream\n about what's going to happen?\n\n GLEN\n No.\n\n NANCY\n You believe in the Boogey Man?\n\n GLEN\n One two, Freddie's coming\n for you? No. Rod killed Tina.\n he's a fruitcake and you know it.\n\n NANCY\n You believe in anything?\n\n GLEN\n I believe in you, me, and\n Rock and Roll. And I'm not\n too sure about you lately.\n\n NANCY thinks.\n\n NANCY\n Listen, I got a crazy favor\n to ask.\n\n GLEN\n Uh-oh...\n\n NANCY\n It's nothing too hard or anything."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You believe it's possible to dream about what's going to happen?\nDIALOGUE — Glen: No.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from A Nightmare On Elm Street: You believe it's possible to dream about what's going to happen?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You believe it's possible to dream about what's going to happen? The camera pushes in with small amplitude at slow speed. Glen (S1) says: [English] No. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Now both Spreg and Engelman are stalking Mace through the\nSetting: derived from S T R A N G E D A Y S\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. STREET - NIGHT\nACTION: Now both Spreg and Engelman are stalking Mace through the crowd. Spreg catches a glimpse of her running and he\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from S T R A N G E D A Y S: Now both Spreg and Engelman are stalking Mace through the crowd. Spreg catches a glimpse of her running and he, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Now both Spreg and Engelman are stalking Mace through the crowd. Spreg catches a glimpse of her running and he The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Total Recall\"\n\nSCREENPLAY EXCERPT:\nWhat I'm saying is... I don't\n remember you. I don't remember\n us. I don't remember me --\n\n Melina's expression grows dead hard.\n\n MELINA\n [And I thought...]\n\n QUAIL\n Who is us?\n\n MELINA\n (ignoring the\n question)\n Memory erasure is what they\n use on agents. Go away.\n\n QUAIL\n Wait... I was an agent -- I'm\n not now -- just tell me who\n I am! How did we know each\n other? Why are they trying\n to kill me?!\n\n He takes a step toward her, nearly frantic.\n\n MELINA\n I'm not trusting you again.\n\n QUAIL\n You loved me once, you must\n have --\n\n MELINA\n That you was a liar. Who you\n are now I don't even know.\n\n QUAIL\n Please, Melina --\n\n MELINA"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: What I'm saying is... I don't remember you. I don't remember\nDIALOGUE — Melina: [And I thought...]\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Total Recall\": What I'm saying is... I don't remember you. I don't remember, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What I'm saying is... I don't remember you. I don't remember The camera pushes in with small amplitude at slow speed. Melina (S1) says: [English] [And I thought...] The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Total Recall\"\n\nSCREENPLAY EXCERPT:\nHow could you do it? After\n eight years!\n\n KIRSTEN\n I'm not your wife, Quail.\n\n QUAIL\n Not my wife! You are out of\n your mind.\n\n KIRSTEN\n (indifferent;\n nursing her\n arm)\n It's a false memory implant.\n I never saw you before six\n weeks ago.\n\n Quail is totally disoriented.\n\n QUAIL\n Why are you lying like this?\n\n KIRSTEN\n No, Quail. It's true. You\n work for E.I.O. So do I.\n\n As she speaks, she walks to a picture on the wall and from\n behind it pulls out a small wallet. She flips it open to\n show him her E.I.O. badge. He looks down uncomprehendingly\n at the holographic lettering \"Earth Intelligence Organisation\".\n\n QUAIL\n (indicating their\n surroundings)\n But why all of this?\n\n KIRSTEN"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: How could you do it? After I'm not your wife, Quail.\nDIALOGUE — Kirsten: I'm not your wife, Quail.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Total Recall\": How could you do it? After I'm not your wife, Quail., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. How could you do it? After I'm not your wife, Quail. The camera pushes in with small amplitude at slow speed. Kirsten (S1) says: [English] I'm not your wife, Quail. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Michael reacts to the sound. An urgency comes over him.\nSetting: derived from Legion\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PARADISE FALLS GAS 'N' GRUB - CONTINUOUS\nACTION: Michael reacts to the sound. An urgency comes over him. Audrey arrives with more steaming water.\nDIALOGUE — Audrey: What is that?!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Legion: Michael reacts to the sound. An urgency comes over him. Audrey arrives with more steaming water., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Michael reacts to the sound. An urgency comes over him. Audrey arrives with more steaming water. The camera pushes in with small amplitude at slow speed. Audrey (S1) says: [English] What is that?! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: T R O N\n\nSCREENPLAY EXCERPT:\nThe MCP blasted it... There are very\n few domains left with any power at all.\n\n\n FLYNN\n Lousy utility companies ...\n they're the same wherever you go.\n\n\n 456-461 OMIT 456-461\n\n\n 462 INT. DIMLY LIT CELL 462\n\n\n Where we see Dumont plugged into energy foot sockets. Two memory\n guards stand by his side holding their staffs against his chest.\n Dumont's face is wracked with pain, his robe torn, but his\n expression is full of determination. As we watch, another blast of\n torturous energy comes from the boots, and his face contorts. CAMERA\n PULLS BACK AND ANGLES UP to the second level where we see Sark\n looking down into the pit that is Dumont's cell.\n\n\n 463 CLOSE UP SARK 463\n\n\n 464 MED. SHOT - DUMONT 464\n\n\n He looks up.\n\n\n DUMONT\n What do you want? I'm busy.\n\n\n 465 SHOT OVER DUMONT'S SHOULDER 465\n\n\n of Sark.\n\n\n SARK"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The MCP blasted it... There are very few domains left with any power at all.\nDIALOGUE — Flynn: Lousy utility companies ... they're the same wherever you go.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T R O N: The MCP blasted it... There are very few domains left with any power at all., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The MCP blasted it... There are very few domains left with any power at all. The camera pushes in with small amplitude at slow speed. Flynn (S1) says: [English] Lousy utility companies ... they're the same wherever you go. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: A Nightmare On Elm Street\n\nSCREENPLAY EXCERPT:\nWho the fuck you think you are,\n whoever you are?\n\n\n24. EXT. TINA'S BACK YARD. NIGHT. 24.\n\n WIDE ANGLE ON THE REAR OF THE HOUSE. A LIGHT COMES ON. TINA\n appears in the doorway.\n\n TINA"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Who the fuck you think you are, 24. EXT. TINA'S BACK YARD. NIGHT. 24.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from A Nightmare On Elm Street: Who the fuck you think you are, 24. EXT. TINA'S BACK YARD. NIGHT. 24., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Who the fuck you think you are, 24. EXT. TINA'S BACK YARD. NIGHT. 24. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: THE YOUNG MAN from the bar grabs JACOB under the arms and drags him\nSetting: derived from \"Jacob'S Ladder\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BAR - DAY\nACTION: THE YOUNG MAN from the bar grabs JACOB under the arms and drags him Where am I? Who are you?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Jacob'S Ladder\": THE YOUNG MAN from the bar grabs JACOB under the arms and drags him Where am I? Who are you?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. THE YOUNG MAN from the bar grabs JACOB under the arms and drags him Where am I? Who are you? The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Bottle Rocket\n\nSCREENPLAY EXCERPT:\nINT/EXT. PLANNING SEQUENCE. DAY\n\nWe see the team assembled around a table. They're drinking\ncoffee and looking at pictures, charts, maps, diagrams,\ntools. Dignan passes out some walkie talkies.\n\nThey test out some smoke cannisters on the roof. Dignan\nthrows one down. They all run for cover.\n\nApplejack and Dignan drive fast through a slalom course in a\ndeserted parking lot.\n\nWe see Kumar working on a safe. He wears a stethoscope and\nlistens to the tumblers. He turns the dial. Dignan is timing\nhim with a stopwatch.\n\nKumar opens the safe. He looks at Dignan. Dignan nods with a\nserious expression on his face."}, {"role": "assistant", "content": "## SCENE 1 — INT/EXT. PLANNING SEQUENCE. DAY\nACTION: We see the team assembled around a table. They're drinking coffee and looking at pictures, charts, maps, diagrams,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bottle Rocket: We see the team assembled around a table. They're drinking coffee and looking at pictures, charts, maps, diagrams,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We see the team assembled around a table. They're drinking coffee and looking at pictures, charts, maps, diagrams, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jay And Silent Bob Strike Back\n\nSCREENPLAY EXCERPT:\nINT. SOUNDSTAGE--SAME\n\n Jay and Silent Bob rush in to see--\n\n Wes Craven getting ready to direct a scene with a familiar-\n looking GHOSTFACE KILLER and SHANNEN DOHERTY. The\n Clapper/Loader's clapboard reads: Scream 4\n\n CLAPPER/LOADER\n Scream four, scene thirty-seven,\n take one.\n (claps it and rushes\n off)\n\n WES CRAVEN\n Action!\n\n The Killer chases Shannen around the room, falling over stuff,\n until she hits him with a lamp, knocking him out.\n\n SHANNEN DOHERTY\n Alright, you bastard! Let's see who\n you really are!\n\n Shannen pulls the mask off the short performer to reveal\n SUZANNE.\n\n Jay and Silent Bob go wide-eyed.\n\n SHANNEN DOHERTY\n Fucking Miramax--\n (getting up)\n CUT!\n\n Shannen heads over to Wes, holding the mask.\n\n WES CRAVEN\n Shannen, usually I say \"cut.\"\n\n SHANNEN DOHERTY\n A monkey? Jesus, you guys aren't\n even trying anymore, are you?\n\n WES CRAVEN\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. SOUNDSTAGE--SAME\nACTION: Jay and Silent Bob rush in to see-- Wes Craven getting ready to direct a scene with a familiar-\nDIALOGUE — Wes Craven: Action!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jay And Silent Bob Strike Back: Jay and Silent Bob rush in to see-- Wes Craven getting ready to direct a scene with a familiar-, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jay and Silent Bob rush in to see-- Wes Craven getting ready to direct a scene with a familiar- The camera pushes in with small amplitude at slow speed. Wes Craven (S1) says: [English] Action! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Minority Report\"\n\nSCREENPLAY EXCERPT:\nEXT. THE SPRAWL\n\n Precrime transports zoom through the city\n\n CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. THE SPRAWL\nACTION: Precrime transports zoom through the city\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Minority Report\": Precrime transports zoom through the city, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Precrime transports zoom through the city The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nThat's got her. Thanks. I'm Steve\n Christy. Welcome to Camp Crystal Lake.\n You got some grubby clothes? Climb into\n 'em. Alice, see if Bill has cleaned out\n the boathouse. I want him to start with\n the canoes. What happened to Brenda?\n\n ALICE\n You told her to sweep the courts.\n\n STEVE sets off.\n\n STEVE\n I'd rather she painted. Let's go, folks.\n\n As soon as he's gone, NED turns to ALICE.\n\n NED\n I thought we had two weeks...\n\n ALICE shrugs.\n\n ALICE\n You can changed in three.\n\n CUT TO:\n\n\n 33 EXT. EDGE OF LAKE - DAY\n\n The CAMERA looks down a long line of canoes, each with its hull\n up. A brush slaps into view. BILL is sweating while finishing\n a coat of paint on the last canoe.\n\n BILL is a thin, ascetic-looking college sophomore. He rubs his\n long thin rose with the back of his wrist as he hears someone\n call his name OS. He turns to see ALICE at the bottom of the\n trail.\n\n ALICE\n Steve said for you to start on the boats.\n\n BILL\n I finished the boats.\n\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: That's got her. Thanks. I'm Steve Christy. Welcome to Camp Crystal Lake.\nDIALOGUE — Alice: You told her to sweep the courts.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: That's got her. Thanks. I'm Steve Christy. Welcome to Camp Crystal Lake., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. That's got her. Thanks. I'm Steve Christy. Welcome to Camp Crystal Lake. The camera pushes in with small amplitude at slow speed. Alice (S1) says: [English] You told her to sweep the courts. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Tombstone\n\nSCREENPLAY EXCERPT:\nOh, Jesus, Johnny...\n\nAllie's eyes flare, Louisa stifles a sob. The Cowboys\nsnicker. Wyatt clenches his teeth, still staring straight\nahead, and drives on. A few beats then Curly Bill turns to\nIke:\n\n CURLY BILL\n Take Frank and finish it.\n\nEXT - TRAIN STATION, TUCSON - NIGHT\n\nA train pours steam onto the platform in puffs and clouds\nwhile 2 porters load Morgan's coffin into a boxcar. Up ahead,\nMattie and Louisa fumble with luggage while Allie boosts\nVirgil into a passenger car. As we HEAR the conductor's voice\nshouting, \" 'Board!\", Ike and Stillwell emerge from the\nshadows near the scales, crouched behind shotguns. They\nexchange nods and start forward, cocking their weapons,\nsquinting through the steam:\n\n STILLWELL\n That's Virgil with the women. But\n Where the hell's Wyatt?\n\n O.S. VOICE\n Right behind you, Stillwell.\n\nThey spin around. Wyatt stands behind them, looking down the\nbarrels of Virgil's big 10 gauge. Stillwell raises his\nshotgun, Wyatt FIRES. Stillwell hits the floor in a crumpled\nheap, his torso a smoking bundle of bloody rags. Screams\n??????????\nPage 87 cut off right side\nPage 88 missing\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n WAGONMASTER\n Rest of us're 3 months out of\n Independece but we're from\n Carthage, Illinois originally.\n\n WYATT\n Carthage, really? I'm from Monmouth.\n\n WAGONMASTER\n Another Illinois man! Practical\n Neighbors! Step down, Marshal.\n Have "}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Allie's eyes flare, Louisa stifles a sob. The Cowboys snicker. Wyatt clenches his teeth, still staring straight\nDIALOGUE — Curly Bill: Take Frank and finish it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Tombstone: Allie's eyes flare, Louisa stifles a sob. The Cowboys snicker. Wyatt clenches his teeth, still staring straight, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Allie's eyes flare, Louisa stifles a sob. The Cowboys snicker. Wyatt clenches his teeth, still staring straight The camera pushes in with small amplitude at slow speed. Curly Bill (S1) says: [English] Take Frank and finish it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: And beyond on a plateau overlooking the grounds, a\nSetting: derived from Blade Runner\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. PLATEAU - DUSK 85\nACTION: And beyond on a plateau overlooking the grounds, a figure stands watching, waiting like a bird of prey.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade Runner: And beyond on a plateau overlooking the grounds, a figure stands watching, waiting like a bird of prey., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. And beyond on a plateau overlooking the grounds, a figure stands watching, waiting like a bird of prey. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The officer reluctantly steps aside.\nSetting: derived from Dawn Of\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. NOW DISTRAUGHT...ALMOST PLEADING\nACTION: The officer reluctantly steps aside. 15\t The young man and Fan move down a crowded hall and into a small\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dawn Of: The officer reluctantly steps aside. 15\t The young man and Fan move down a crowded hall and into a small, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The officer reluctantly steps aside. 15\t The young man and Fan move down a crowded hall and into a small The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Natural Born Killers\"\n\nSCREENPLAY EXCERPT:\nEXT. CIRCLE K - DAY\n\n This sequence is filmed in 16MM COLOR, cinema verite a'la 'COPS.'\n\n The CAMERA runs behind three blue windbreaker clad COPS, as they\n run up to the Circle K, shouting obscenities.\n\n In front of the store three windbreaker cops have Mickey on the\n ground, beating him with nightsticks. One COP lies on the\n ground near them, holding his hands over his face, screaming.\n\n To the left Mallory is slugging it out with a windbreaker COP,\n matching each other blow for blow.\n\n Meanwhile, the three cops we ran with reach the action and join\n in.\n\n CUT TO:\n\n MONTAGE\n\n NEWSPAPER-- Newspaper or magazine COLOR AD for a 7/11 store.\n\n NEWSPAPER-- Newspaper or magazine COLOR AD for a gas station.\n\n WAYNE (V.O.)\n They started off robbing 7/11 type\n stores and gas stations and later\n graduated to banks and the big time."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CIRCLE K - DAY\nACTION: This sequence is filmed in 16MM COLOR, cinema verite a'la 'COPS.' The CAMERA runs behind three blue windbreaker clad COPS, as they\nDIALOGUE — Wayne: They started off robbing 7/11 type stores and gas stations and later graduated to banks and the big time.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Natural Born Killers\": This sequence is filmed in 16MM COLOR, cinema verite a'la 'COPS.' The CAMERA runs behind three blue windbreaker clad COPS, as they, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. This sequence is filmed in 16MM COLOR, cinema verite a'la 'COPS.' The CAMERA runs behind three blue windbreaker clad COPS, as they The camera pushes in with small amplitude at slow speed. Wayne (S1) says: [English] They started off robbing 7/11 type stores and gas stations and later graduated to banks and the big time. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in KamenRiderGaim illustrating the trope 'TheUnChosenOne'. Kamen Rider:\n Gentarou, of Kamen Rider Fourze. He was never meant to be Fourze by anyone, instead coming out of nowhere and actually stealing the Fourze driver to become the Kamen Rider.\n This trope is a point of much frustration for Shroud in Kamen Rider Double. Shoutarou transformed with Philip for the first time pretty much out of necessity, with no idea what he was doing. Shroud spends the maj...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TheUnChosenOne.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for KamenRiderGaim, TheUnChosenOne, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TheUnChosenOne. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: T R O N\n\nSCREENPLAY EXCERPT:\nOkay!\n\n 196 SARK'S AIRCRAFT CARRIER - THE BRIDGE 196\n\n High above the game we see Sark watching.\n\n\n 197 FLYNN 197\n\n\n racing for another catch. He makes it and hurls the pellet again.\n We see that Crom's platform is disappearing fast.\n\n\n 198 OMIT 198\n\n 199 THE GAME IN PROGRESS 199\n\n\n FLYNN\n Here's an easy one...\n\n\n Flynn throws again. The pellet hits the mirror platform and heads\n straight at Crom, who misses it. The pellet destroys the ring he\n was standing on, and Crom has to grab frantically at the edge of\n the next ring to keep from falling.\n\n\n 200-201 OMIT 200-201\n\n\n 202 FLYNN 202\n\n\n He is sent another pellet from above and catches it, holding it\n ready, waiting for his opponent to get back up on the ring.\n\n\n 203 SARK 203\n\n\n His face lit from below by the glow of the Game Grid... angry\n\n\n 204 FLYNN 204\n\n\n He puts down his cesta. Suddenly a voice rings out from above.\n Flynn looks up. There, appearing in the mirror disk above, is Sark's\n face, furious, huge.\n\n\n SARK\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: 196 SARK'S AIRCRAFT CARRIER - THE BRIDGE 196 High above the game we see Sark watching.\nDIALOGUE — Flynn: Here's an easy one...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T R O N: 196 SARK'S AIRCRAFT CARRIER - THE BRIDGE 196 High above the game we see Sark watching., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. 196 SARK'S AIRCRAFT CARRIER - THE BRIDGE 196 High above the game we see Sark watching. The camera pushes in with small amplitude at slow speed. Flynn (S1) says: [English] Here's an easy one... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: T R O N\n\nSCREENPLAY EXCERPT:\nJump.\n\n\n Yori looks up.\n\n\n YORI\n But... won't you be de-rezzed?\n\n\n FLYNN"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: But... won't you be de-rezzed?\nDIALOGUE — Yori: But... won't you be de-rezzed?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T R O N: But... won't you be de-rezzed?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. But... won't you be de-rezzed? The camera pushes in with small amplitude at slow speed. Yori (S1) says: [English] But... won't you be de-rezzed? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Deckard looks around the corner of the door down the\nSetting: derived from Blade Runner\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CORRIDOR - TENTH FLOOR - NIGHT 106\nACTION: Deckard looks around the corner of the door down the hall. Batty's at the other end. Except for jockstrap\nDIALOGUE — Batty: You wanna play?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade Runner: Deckard looks around the corner of the door down the hall. Batty's at the other end. Except for jockstrap, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Deckard looks around the corner of the door down the hall. Batty's at the other end. Except for jockstrap The camera pushes in with small amplitude at slow speed. Batty (S1) says: [English] You wanna play? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: MARIAN and RAQUEL, two exquisite young women, are looking\nSetting: derived from As Credits Roll\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LINGERIE SHOP - DAY\nACTION: MARIAN and RAQUEL, two exquisite young women, are looking over the merchandise.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from As Credits Roll: MARIAN and RAQUEL, two exquisite young women, are looking over the merchandise., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. MARIAN and RAQUEL, two exquisite young women, are looking over the merchandise. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: T W I N P E A K S\n\nSCREENPLAY EXCERPT:\nOkay, this test is going to be about the\n theorems I told you about last week. You\n remember the...\n\n LAURA\n Don't tell me now. Tell me right before\n the test. I won't be able to remember\n long enough.\n\n DONNA\n You graduating this year will be proof\n that miracles happen.\n\n LAURA\n Thanks.\n\n Laura tweaks Donna's cheek.\n\n DONNA\n James called me last night looking for\n you.\n\n LAURA\n When?\n\n DONNA\n The usual, 9:15.\n\n LAURA\n He probably wanted to drive over.\n\n DONNA\n Were you with Bobby? Or are you two\n still fighting?\n\n LAURA\n No, and yes. I don't know what I'm going\n to do about Bobby. I know he is seeing\n someone else and that's okay with me,\n and he thinks I'm seeing someone else\n and that's not okay with him.\n\n DONNA\n Are you going to tell him about that"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Okay, this test is going to be about the theorems I told you about last week. You\nDIALOGUE — Laura: Don't tell me now. Tell me right before the test. I won't be able to remember long enough.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T W I N P E A K S: Okay, this test is going to be about the theorems I told you about last week. You, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Okay, this test is going to be about the theorems I told you about last week. You The camera pushes in with small amplitude at slow speed. Laura (S1) says: [English] Don't tell me now. Tell me right before the test. I won't be able to remember long enough. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: EVE looks up into CHEV'S eyes, absolutely vulnerable,\nSetting: derived from Crank\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CHOCOLATE'S APARTMENT - EARLIER\nACTION: EVE looks up into CHEV'S eyes, absolutely vulnerable, absolutely STUNNING. Pale sunlight punches through the half\nDIALOGUE — Eve: Do you promise?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Crank: EVE looks up into CHEV'S eyes, absolutely vulnerable, absolutely STUNNING. Pale sunlight punches through the half, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. EVE looks up into CHEV'S eyes, absolutely vulnerable, absolutely STUNNING. Pale sunlight punches through the half The camera pushes in with small amplitude at slow speed. Eve (S1) says: [English] Do you promise? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: If my father was moving this he'd had\nSetting: derived from Good News, Bad News\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SITS DOWN\nACTION: If my father was moving this he'd had to have a cigarette in his mouth the\nDIALOGUE — George: Ya know, I can't believe you're bringin' in an extra bed for woman, that wants to sleep with you. Why don't you bring in an extra guy too? (sits down)\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Good News, Bad News: If my father was moving this he'd had to have a cigarette in his mouth the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. If my father was moving this he'd had to have a cigarette in his mouth the The camera pushes in with small amplitude at slow speed. George (S1) says: [English] Ya know, I can't believe you're bringin' in an extra bed for woman, that wants to sleep with you. Why don't you bring in an extra guy too? (sits down) The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Right out of Nazi Germany. Barbed wire, guards, the whole\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. INTERNMENT CAMP - BEHIND THE USO SHOW - DUSK\nACTION: Right out of Nazi Germany. Barbed wire, guards, the whole They are on their backs, shimmying along the ground\nDIALOGUE — The Mole: Be careful not to touch this wire-\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: Right out of Nazi Germany. Barbed wire, guards, the whole They are on their backs, shimmying along the ground, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Right out of Nazi Germany. Barbed wire, guards, the whole They are on their backs, shimmying along the ground The camera pushes in with small amplitude at slow speed. The Mole (S1) says: [English] Be careful not to touch this wire- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fantastic Mr. Fox\n\nSCREENPLAY EXCERPT:\nINT. FOX'S TREE. NIGHT\n\n Mrs. Fox licks the stump of Fox's tail and mends it with\n gauze and medical tape. She looks furious. Kylie and the Fox\n cubs watch, concerned. Ash says uncertainly:\n\n ASH\n It'll grow back, won't it?\n\n KYLIE\n (shaking his head)\n Tails don't grow back, except for\n lizards.\n\n FOX\n\n (MISERABLY)\n Tails don't grow back. I'm going to be\n tail-less for the rest of my life.\n\n ASH\n\n (NERVOUSLY)\n Well, anyway, it's not half as bad as\n double-pneumonia, right?\n (pointing to Kristofferson)\n His dad's got one foot in the grave and\n one foot on a banana peel. That's a lot\n worse than --\n Kristofferson hurls an acorn violently onto the floor. It\n ricochets off a wall and into a teacup. Everyone falls\n silent. Kristofferson turns away.\n\n\n\n\n\n\n\n\n\n 35.\n\n KRISTOFFERSON\n Excuse me, everyone. I'm going to go\n meditate for half an hour.\n Kristofferson walks out of the room. Mrs. Fox looks at Ash\n and says angrily:\n\n MRS. FOX\n You've got twenty-nine minutes to come up\n with a proper apology.\n Ash crosses his arms in front of his chest and stares\n straight ahead into space grimly. Fox says suddenly:\n\n FOX\n What's with the crazy ou"}, {"role": "assistant", "content": "## SCENE 1 — INT. FOX'S TREE. NIGHT\nACTION: Mrs. Fox licks the stump of Fox's tail and mends it with gauze and medical tape. She looks furious. Kylie and the Fox\nDIALOGUE — Ash: It'll grow back, won't it?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fantastic Mr. Fox: Mrs. Fox licks the stump of Fox's tail and mends it with gauze and medical tape. She looks furious. Kylie and the Fox, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Mrs. Fox licks the stump of Fox's tail and mends it with gauze and medical tape. She looks furious. Kylie and the Fox The camera pushes in with small amplitude at slow speed. Ash (S1) says: [English] It'll grow back, won't it? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Bodyguard\n\nSCREENPLAY EXCERPT:\nGuess again, whore! Fuck you and\n fuck Miami. I'm coming for you.\n I know where you are and I'm\n coming for you...\n\n 95.\n\n\n Rachel is paralyzed by fear, unable to hang up the phone.\n\n146 INT. PRESIDENTIAL SUITE - LIVING ROOM - NIGHT 146\n Frank comes in the front door. He's had it. Devaney,\n who has been sitting alone in the living room, stubs out\n a cigarette and stands to meet him. Frank doesn't pause.\n FRANK\n I'm through. I'll get you back to\n L.A. and that's it. The guy with\n Fletcher can cover you till you get\n a replacement.\n\n DEVANEY\n She got another call, while you\n were downstairs and she answered it\n herself. It was him, Frank. Same\n guy.\n FRANK\n (moving toward\n balcony)\n I don't care.\n DEVANEY\n Farmer, it really shook her up.\n She thought it was going to be\n Fletcher calling her from L.A. I\n think she'll be reasonable now.\n FRANK"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Guess again, whore! Fuck you and fuck Miami. I'm coming for you.\nDIALOGUE — Frank: I'm through. I'll get you back to L.A. and that's it. The guy with Fletcher can cover you till you get a replacement.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Bodyguard: Guess again, whore! Fuck you and fuck Miami. I'm coming for you., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Guess again, whore! Fuck you and fuck Miami. I'm coming for you. The camera pushes in with small amplitude at slow speed. Frank (S1) says: [English] I'm through. I'll get you back to L.A. and that's it. The guy with Fletcher can cover you till you get a replacement. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: GOLDENROD REVISED 2nd DRAFT 6/4/08 48A.\nSetting: derived from Ninja Assassin\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: GOLDENROD REVISED 2nd DRAFT 6/4/08 48A. 54 CONTINUED: (2) 54\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ninja Assassin: GOLDENROD REVISED 2nd DRAFT 6/4/08 48A. 54 CONTINUED: (2) 54, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GOLDENROD REVISED 2nd DRAFT 6/4/08 48A. 54 CONTINUED: (2) 54 The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Hackers\n\nSCREENPLAY EXCERPT:\nINT. COMPUTER ROOM, ELLINGSON MINERAL.\n\n PLAGUE\n Come on you son of a bitch, is that all you\n got, huh? Come on, let's see what else you\n can do! You talking to me? Huh? Hahahaha.\n Are you nuts? Come at me!\n\n Margo and Duke are awed by Plague's imminent failure and\n total unprofessionalism.\n\n HAL\n They're in the kernal.\n\n DA VINCI VIRUS\n Help... me...\n\n The Da Vinci virus dies. The tankers stop capsizing and\n right themselves.\n\n WOMAN\n (off screen)\n The tankers have stopped capsizing.\n\n SOMEONE ELSE\n (off screen)\n Ballast tanks are emptying. It'll be okay.\n\n The main screen lights up:\n\n ARF ARF!\n WE GOTCHA!\n\n MESS WITH THE BEST\n DIE LIKE THE REST\n\n PLAGUE\n Little pissant!\n\n The Gibson dies in a flash of light."}, {"role": "assistant", "content": "## SCENE 1 — INT. COMPUTER ROOM, ELLINGSON MINERAL.\nACTION: Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you\nDIALOGUE — Plague: Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you can do! You talking to me? Huh? Hahahaha. Are you nuts? Come at me!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hackers: Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you The camera pushes in with small amplitude at slow speed. Plague (S1) says: [English] Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you can do! You talking to me? Huh? Hahahaha. Are you nuts? Come at me! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: What the hell you want to go there for?\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: What the hell you want to go there for? Yeah, it's just this fucked-up building.\t\t *\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: What the hell you want to go there for? Yeah, it's just this fucked-up building.\t\t *, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What the hell you want to go there for? Yeah, it's just this fucked-up building.\t\t * The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BuffytheVampireSlayer illustrating the trope 'WitchHunt'. Buffyverse:\n In the Buffy episode \"Gingerbread\", Giles notes that a certain breed of demon prefers to turn humans against each other rather than attacking them directly, from Salem all the way back to Hansel & Gretel. One such demon convinces Buffy's own mother to lead a witch hunt that almost gets Buffy, Willow and their friend Amy burned at the stake.\n An interesting example in the Angel episode...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying WitchHunt.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BuffytheVampireSlayer, WitchHunt, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for WitchHunt. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Who the fuck you think you are,\nSetting: derived from A Nightmare On Elm Street\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Who the fuck you think you are, 24. EXT. TINA'S BACK YARD. NIGHT. 24.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from A Nightmare On Elm Street: Who the fuck you think you are, 24. EXT. TINA'S BACK YARD. NIGHT. 24., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Who the fuck you think you are, 24. EXT. TINA'S BACK YARD. NIGHT. 24. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Barbie\n\nSCREENPLAY EXCERPT:\nEXT. BARBIE LAND. DAY.\n\nGloria and Sasha are totally impressed by the beauty of\nBARBIE LAND as they drive by the ocean in Barbie's\nconvertible.\n\n BARBIE MARGOT\n (soaking it all in)\n I can feel my heels lifting\n already. Yes, this is what I was\n supposed to do. Bring you back\n here!\n\n GLORIA\n (so happy)\n It feels right!\n\n BARBIE MARGOT\n It does!\n\nThey pass the BEACH. Kens are playing volleyball a la Top Gun\nand the Barbies are cheering (reverse of beginning).\n\n BARBIE MARGOT\n That's strange...\n\nAlso Kens race on the beach and hug triumphantly in the surf\n(a la Rocky III). Barbie Issa hands a Ken a beer.\n\n BARBIE ISSA\n Incoming brewski beer--\n (she makes plane sounds)\n\n BARBIE MARGOT\n Uh, so that's our president with\n the beer. And the cheering squad...\n is the supreme court?\n\n BARBIE ISSA\n This is so much better than being\n President!!\n\n BARBIE MARGOT\n Something's weird today.\n 67.\n\n\nA Ken Mermaid leaps out of the surf and waves. You don't\nthink that's a thing? Check this out:\n\n KEN MERMAID\n Hi Barbie!\n\n BARBIE MARGOT\n (slightly taken aback))\n Oh... OK. Hi Ken.\n\nBarbie Mermaid emerges and hands the Ken Mermaid a beer.\n\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. BARBIE LAND. DAY.\nACTION: Gloria and Sasha are totally impressed by the beauty of BARBIE LAND as they drive by the ocean in Barbie's\nDIALOGUE — Barbie Margot: I can feel my heels lifting already. Yes, this is what I was supposed to do. Bring you back here!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Barbie: Gloria and Sasha are totally impressed by the beauty of BARBIE LAND as they drive by the ocean in Barbie's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Gloria and Sasha are totally impressed by the beauty of BARBIE LAND as they drive by the ocean in Barbie's The camera pushes in with small amplitude at slow speed. Barbie Margot (S1) says: [English] I can feel my heels lifting already. Yes, this is what I was supposed to do. Bring you back here! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nEXT. INTERNMENT CAMP - BEHIND THE USO SHOW - DUSK\n\n Right out of Nazi Germany. Barbed wire, guards, the whole\n bit.\n\n ANGLE ON BOYS\n\n They are on their backs, shimmying along the ground\n underneath the blanket of razor-wire that is set up around\n the camp.\n\n The Mole uses the bolt-cutters to cut the wire as he goes,\n clearing a path for Stan, Kyle and Cartman.\n\n THE MOLE\n Be careful not to touch this wire-\n\n A wire catches on Cartman's arm and SNAPS against his face.\n\n CARTMAN\n OW! MOTHERFUCKER!\n\n BZZAAAAT!!!\n\n CARTMAN\n FUC-\n\n Stan shoves his hand over Cartman's mouth, leaving Cartman to\n cuss bloody murder into Stan's glove.\n\n CARTMAN\n MPHMPH MPHMPHMPH!!!\n\n BZAAT!\n\n CARTMAN\n MMMPPPHHH!!!\n\n Finally they clear the fence and lie down in view of the\n camp.\n\n THE MOLE\n It sounds like the USO show has started.\n We have precious little time...\n\n Kyle looks at the horrible concentration camp.\n\n KYLE\n Oh my God...\n\n THE MOLE\n God? ...Let me tell you something about\n God... He is ze biggest bitch of zem all.\n\n Kyle looks at The Mole, oddly.\n\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. INTERNMENT CAMP - BEHIND THE USO SHOW - DUSK\nACTION: Right out of Nazi Germany. Barbed wire, guards, the whole They are on their backs, shimmying along the ground\nDIALOGUE — The Mole: Be careful not to touch this wire-\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: Right out of Nazi Germany. Barbed wire, guards, the whole They are on their backs, shimmying along the ground, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Right out of Nazi Germany. Barbed wire, guards, the whole They are on their backs, shimmying along the ground The camera pushes in with small amplitude at slow speed. The Mole (S1) says: [English] Be careful not to touch this wire- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: T W I N P E A K S\n\nSCREENPLAY EXCERPT:\nNo, I am uptight. I hate it... I don't want\n to be this way, but Laura I don't... I\n mean... I'm your friend no matter what\n way you are.\n\n LAURA\n You know, even when I think about your\n face I get happier.\n\n They go in and sit down on the couch.\n\n DONNA\n Do you want to talk?\n\n LAURA\n No, I want to smoke.\n\n She digs for a cigarette in her purse.\n\n DONNA\n I'm in a mess today, too."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: No, I am uptight. I hate it... I don't want to be this way, but Laura I don't... I\nDIALOGUE — Laura: You know, even when I think about your face I get happier.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T W I N P E A K S: No, I am uptight. I hate it... I don't want to be this way, but Laura I don't... I, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. No, I am uptight. I hate it... I don't want to be this way, but Laura I don't... I The camera pushes in with small amplitude at slow speed. Laura (S1) says: [English] You know, even when I think about your face I get happier. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Musty and cramped. A portable shower, a dressing table\nSetting: derived from Blade Runner\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. DRESSING ROOM - NIGHT 64\nACTION: Musty and cramped. A portable shower, a dressing table and not much else. Salome takes the snake from around\nDIALOGUE — Deckard: It that mother real?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade Runner: Musty and cramped. A portable shower, a dressing table and not much else. Salome takes the snake from around, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Musty and cramped. A portable shower, a dressing table and not much else. Salome takes the snake from around The camera pushes in with small amplitude at slow speed. Deckard (S1) says: [English] It that mother real? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: In the darkened procedure room, Miriam finds Catherine\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CAMPBELL CENTER - LABORATORY - DAY\nACTION: In the darkened procedure room, Miriam finds Catherine standing between the two suspension devices, alone.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: In the darkened procedure room, Miriam finds Catherine standing between the two suspension devices, alone., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. In the darkened procedure room, Miriam finds Catherine standing between the two suspension devices, alone. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Chuck illustrating the trope 'ObliviousGuiltSlinging'. Chuck:\n A good example of this as Casey is supposed to kill Chuck once he's no longer useful, but especially in one of the more recent episodes, Chuck acts in a friendly way which makes Casey have reservations about doing this.\n Another good example occurs when Sarah is told to bring in Chuck in the episode \"Chuck Versus the First Kill\", and Chuck notes how she's the only one he believes he can tr...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying ObliviousGuiltSlinging.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Chuck, ObliviousGuiltSlinging, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for ObliviousGuiltSlinging. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: We get a better view of the salesman waiting while his\nSetting: derived from \"Rear Window\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEIGHBORHOOD - NIGHT - MEDIUM SHOT\nACTION: We get a better view of the salesman waiting while his connection is being made. Whoever he has called answers. And\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Rear Window\": We get a better view of the salesman waiting while his connection is being made. Whoever he has called answers. And, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We get a better view of the salesman waiting while his connection is being made. Whoever he has called answers. And The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Int: Bedroom. Early-Morning Hours\n\nSCREENPLAY EXCERPT:\nCan we talk about this later?\n\n\t\t\t\tRANDAL\n\t\tThe jizz-mopper's job is to clean\n\t\tup the booths afterward, because\n\t\tpractically everybody shoots a load\n\t\tagainst the window, and I don't\n\t\tknow if you know or not, but cum\n\t\tleaves streaks if you don't clean\n\t\tit right away.\n\n\t\t\t\tCUSTOMER"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Can we talk about this later? The jizz-mopper's job is to clean\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int: Bedroom. Early-Morning Hours: Can we talk about this later? The jizz-mopper's job is to clean, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Can we talk about this later? The jizz-mopper's job is to clean The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I can't let them kill any more children.\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: I can't let them kill any more children. She wades through table after table, pushing them out of her\nDIALOGUE — Prowler.: When MRS. VOORHEES is too quick for her, shoving a table across her path like a barred door. Before ALICE can think twice, MRS. VOORHEES has shoved a second...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: I can't let them kill any more children. She wades through table after table, pushing them out of her, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I can't let them kill any more children. She wades through table after table, pushing them out of her The camera pushes in with small amplitude at slow speed. Prowler. (S1) says: [English] When MRS. VOORHEES is too quick for her, shoving a table across her path like a barred door. Before ALICE can think twice, MRS. VOORHEES has shoved a second... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: WIDER, ON NANCY alone in the room. She turns and looks at the\nSetting: derived from A Nightmare On Elm Street\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: WIDER, ON NANCY alone in the room. She turns and looks at the bed. MUSIC slips in and builds. The bed has changed color.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from A Nightmare On Elm Street: WIDER, ON NANCY alone in the room. She turns and looks at the bed. MUSIC slips in and builds. The bed has changed color., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. WIDER, ON NANCY alone in the room. She turns and looks at the bed. MUSIC slips in and builds. The bed has changed color. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: It's Martine. I think he's dying. He\nSetting: derived from The Suicide\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: It's Martine. I think he's dying. He tried to kill himself with pills.\nDIALOGUE — Jerry: What?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Suicide: It's Martine. I think he's dying. He tried to kill himself with pills., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. It's Martine. I think he's dying. He tried to kill himself with pills. The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] What? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Guess again, whore! Fuck you and\nSetting: derived from The Bodyguard\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Guess again, whore! Fuck you and fuck Miami. I'm coming for you.\nDIALOGUE — Frank: I'm through. I'll get you back to L.A. and that's it. The guy with Fletcher can cover you till you get a replacement.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Bodyguard: Guess again, whore! Fuck you and fuck Miami. I'm coming for you., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Guess again, whore! Fuck you and fuck Miami. I'm coming for you. The camera pushes in with small amplitude at slow speed. Frank (S1) says: [English] I'm through. I'll get you back to L.A. and that's it. The guy with Fletcher can cover you till you get a replacement. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A speeder banks through and away from the legs of a walker. Two other\nSetting: derived from The Empire Strikes Back\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HOTH - ICE PLAIN - BATTLEFIELD\nACTION: A speeder banks through and away from the legs of a walker. Two other speeders pass the first speeder from the opposite direction. Other\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Empire Strikes Back: A speeder banks through and away from the legs of a walker. Two other speeders pass the first speeder from the opposite direction. Other, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A speeder banks through and away from the legs of a walker. Two other speeders pass the first speeder from the opposite direction. Other The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Thanks again for last night!\nSetting: derived from The Sponge\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Thanks again for last night! Hey, I didn't even use one.\nDIALOGUE — Elaine: What?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Sponge: Thanks again for last night! Hey, I didn't even use one., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Thanks again for last night! Hey, I didn't even use one. The camera pushes in with small amplitude at slow speed. Elaine (S1) says: [English] What? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: That's my Gyarados. I rode him across the sea. Those kind of\nSetting: derived from Untitled\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: That's my Gyarados. I rode him across the sea. Those kind of waves are nothing for it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: That's my Gyarados. I rode him across the sea. Those kind of waves are nothing for it., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. That's my Gyarados. I rode him across the sea. Those kind of waves are nothing for it. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DoctorWho illustrating the trope 'TooMuchForManToHandle'. Doctor Who:\n \"The Long Game\" has a mention of futuristic computers activated by linking to people's brains having safeguards to prevent overload.\n \"The Parting of the Ways\": Rose absorbs the Time Vortex to become the Bad Wolf and can't control its powers effectively.\n \"Journey's End\": Doctor Donna's brain starts burning out from all of the Doctor's Time Lord experience, leading to a serious Down...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TooMuchForManToHandle.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DoctorWho, TooMuchForManToHandle, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TooMuchForManToHandle. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Elaine and automatic pilot are relaxed in their seats. Both\nSetting: derived from Airplane\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. AIRPLANE - THROUGH COCKPIT WINDOW - NIGHT\nACTION: Elaine and automatic pilot are relaxed in their seats. Both are smoking cigarettes.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Airplane: Elaine and automatic pilot are relaxed in their seats. Both are smoking cigarettes., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Elaine and automatic pilot are relaxed in their seats. Both are smoking cigarettes. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Nick tears desperately up the stairway -- he hits the fourth\nSetting: derived from Basic Instinct\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. THE STAIRWAY\nACTION: Nick tears desperately up the stairway -- he hits the fourth floor door. It explodes open.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Basic Instinct: Nick tears desperately up the stairway -- he hits the fourth floor door. It explodes open., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Nick tears desperately up the stairway -- he hits the fourth floor door. It explodes open. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Star Wars\n\nSCREENPLAY EXCERPT:\nEXT. TATOOINE - LARS HOMESTEAD\n\n The giant twin suns of Tatooine slowly disappear behind a\n distant dune range. Luke stands watching them for a few\n moments, then reluctantly enters the doomed entrance to the\n homestead."}, {"role": "assistant", "content": "## SCENE 1 — EXT. TATOOINE - LARS HOMESTEAD\nACTION: The giant twin suns of Tatooine slowly disappear behind a distant dune range. Luke stands watching them for a few\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Star Wars: The giant twin suns of Tatooine slowly disappear behind a distant dune range. Luke stands watching them for a few, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The giant twin suns of Tatooine slowly disappear behind a distant dune range. Luke stands watching them for a few The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Jack is seated across the aisle from Mrs. Hammen. He falls\nSetting: derived from Airplane\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PASSENGER CABIN - ANOTHER AREA - NIGHT\nACTION: Jack is seated across the aisle from Mrs. Hammen. He falls asleep and slumps over.\nDIALOGUE — Mrs. Hammen: He's dead. He's dead!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Airplane: Jack is seated across the aisle from Mrs. Hammen. He falls asleep and slumps over., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack is seated across the aisle from Mrs. Hammen. He falls asleep and slumps over. The camera pushes in with small amplitude at slow speed. Mrs. Hammen (S1) says: [English] He's dead. He's dead! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Untitled\n\nSCREENPLAY EXCERPT:\nYes, I am, Max.\n\n\nMAX\nI want to see Moses and Rafe. And\nMasha. I think that will help me.\nI£ I see that they're alive. Right\nnow, I'm sure that they're dead.\n\n\nHARLAN"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I want to see Moses and Rafe. And Masha. I think that will help me.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: I want to see Moses and Rafe. And Masha. I think that will help me., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I want to see Moses and Rafe. And Masha. I think that will help me. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Joker\n\nSCREENPLAY EXCERPT:\nI could have killed myself.\n\n CLOSE ON THE LAST LINE, he crosses out \"could\"...\n\n Writes... \"should\".\n\n JOKER"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I could have killed myself. CLOSE ON THE LAST LINE, he crosses out \"could\"...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Joker: I could have killed myself. CLOSE ON THE LAST LINE, he crosses out \"could\"..., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I could have killed myself. CLOSE ON THE LAST LINE, he crosses out \"could\"... The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Lando stops ten feet from Han. The two men eye each other carefully.\nSetting: derived from The Empire Strikes Back\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CLOUD CITY - LANDING PLATFORM\nACTION: Lando stops ten feet from Han. The two men eye each other carefully. Lando shakes his head.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Empire Strikes Back: Lando stops ten feet from Han. The two men eye each other carefully. Lando shakes his head., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Lando stops ten feet from Han. The two men eye each other carefully. Lando shakes his head. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: We will be cruel to the Germans,\nSetting: derived from Inglourious Basterds\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HITLERS PRIVATE LAIR\nACTION: We will be cruel to the Germans, and through our cruelty, they will\nDIALOGUE — Lt. Aldo: Sound good? They all say;\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Inglourious Basterds: We will be cruel to the Germans, and through our cruelty, they will, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We will be cruel to the Germans, and through our cruelty, they will The camera pushes in with small amplitude at slow speed. Lt. Aldo (S1) says: [English] Sound good? They all say; The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Blackadder illustrating the trope 'Luvvies'. Actors Keanrick and Mossop act this way. Apart from calling each other 'dear' and 'love' they're also incredibly pompous and are more interested in patronage, self-promotion, and the rather hideous play they wrote.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying Luvvies.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Blackadder, Luvvies, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for Luvvies. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: That's got her. Thanks. I'm Steve\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: That's got her. Thanks. I'm Steve Christy. Welcome to Camp Crystal Lake.\nDIALOGUE — Alice: You told her to sweep the courts.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: That's got her. Thanks. I'm Steve Christy. Welcome to Camp Crystal Lake., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. That's got her. Thanks. I'm Steve Christy. Welcome to Camp Crystal Lake. The camera pushes in with small amplitude at slow speed. Alice (S1) says: [English] You told her to sweep the courts. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Jay and Silent Bob rush in to see--\nSetting: derived from Jay And Silent Bob Strike Back\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SOUNDSTAGE--SAME\nACTION: Jay and Silent Bob rush in to see-- Wes Craven getting ready to direct a scene with a familiar-\nDIALOGUE — Wes Craven: Action!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jay And Silent Bob Strike Back: Jay and Silent Bob rush in to see-- Wes Craven getting ready to direct a scene with a familiar-, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jay and Silent Bob rush in to see-- Wes Craven getting ready to direct a scene with a familiar- The camera pushes in with small amplitude at slow speed. Wes Craven (S1) says: [English] Action! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Asteroid City\n\nSCREENPLAY EXCERPT:\nEXT. HIGHWAY. DAY\n\nA warning bell rings at a railroad crossing.\n\nThe train slowly rattles over a blacktop interstate which\ndescends, sidewinding, from a distant plateau down to a concise\ndesert town circumvented by towering cactus, yucca, and\nsagebrush. Dry winds blow red dust and tangled scrub. A\nroadrunner darts and sprints. In the remote background: a range\nof faraway mountains encircling the vast, sandy arena. In the\nclose foreground: a decorative covered-wagon roadside sign which\nreads \"Asteroid City, Pop. 87\" in carved orange letters.\n\nThe town itself: a caf�, a motel, a gas station. A parked\nChrysler sedan, a Ford with a running board. White, wooden,\nranch-style fencing punctuated by telephone poles, electric\nwires, and palm trees in tight clusters. A few barns and\nbungalows dot the outskirts, along with a billboard advertising\nan Indian reservation casino 250 miles ahead.\n\nOn the edge of the town: the ruins of the uncompleted/chopped-\noff elevated highway-spur with an official roadblock warning:\n\n Route-Calculations Error\n RAMP CLOSED INDEFINITELY\n Department of Roads and Rails\n 5.\n\n\nA hundred yards into the desert: a modestly-sized astronomical\nobservatory; a field of small radio telescopes; a chainlink\nfence with regional-monument/point-of-interest marker; a flight\nof steep, skinny wooden steps which descends into a wide crater\nwith, at center, ensconced in a rebar display-cage: a pock-\nmarked metallic rock the size of a cantaloupe."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HIGHWAY. DAY\nACTION: A warning bell rings at a railroad crossing. The train slowly rattles over a blacktop interstate which\nDIALOGUE — Ramp Closed Indefinitely: Department of Roads and Rails 5.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Asteroid City: A warning bell rings at a railroad crossing. The train slowly rattles over a blacktop interstate which, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A warning bell rings at a railroad crossing. The train slowly rattles over a blacktop interstate which The camera pushes in with small amplitude at slow speed. Ramp Closed Indefinitely (S1) says: [English] Department of Roads and Rails 5. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Don�t...Don�t worry, I�ll bring someone\nSetting: derived from Saw\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. THINKING\nACTION: Don�t...Don�t worry, I�ll bring someone back, I promise. (He continues crawling\nDIALOGUE — Adam: (thinking) Key...key...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Saw: Don�t...Don�t worry, I�ll bring someone back, I promise. (He continues crawling, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Don�t...Don�t worry, I�ll bring someone back, I promise. (He continues crawling The camera pushes in with small amplitude at slow speed. Adam (S1) says: [English] (thinking) Key...key... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nI know this ain't very romantic, but\n what can I say? I don't want to explode.\n\n She grabs a flashlight, scoots to the door and is gone.\n\n JACK lies back and puts his hands under his hand.\n\n CUT TO:\n\n\n 71 EXT. CAMP GROUNDS - NIGHT\n\n MARCIE runs across the now-wet central area towards the\n bathroom facility. She tries to skip between the raindrops,\n but she knows she'll get a drenching.\n\n CUT TO:\n\n\n 72 INT. BOY'S CABIN - NIGHT\n\n JACK rolls to his side, picks up a small hand-rolled cigarette\n and lights it. He inhales deeply, rolls on his back.\n Something overhead catches his eye.\n\n Across the way from the row of toilet stalls is a row of shower\n stalls, and further down is a row of several sinks and mirrors.\n We hear the toilet flush.\n\n We CUT INSIDE the toilet booth as MARCIE stands reading the\n graffiti to herself. We can only see her from the chest up.\n\n MARCIE"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I know this ain't very romantic, but what can I say? I don't want to explode.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: I know this ain't very romantic, but what can I say? I don't want to explode., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I know this ain't very romantic, but what can I say? I don't want to explode. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: - you gotta agree it's funny! These two,\nSetting: derived from Jurassic Park\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: - you gotta agree it's funny! These two, they dig up dinosaurs! It's wonderful,\nDIALOGUE — Nedry: Dr. Bones, you're going to love this place.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jurassic Park: - you gotta agree it's funny! These two, they dig up dinosaurs! It's wonderful,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. - you gotta agree it's funny! These two, they dig up dinosaurs! It's wonderful, The camera pushes in with small amplitude at slow speed. Nedry (S1) says: [English] Dr. Bones, you're going to love this place. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Matrix\n\nSCREENPLAY EXCERPT:\nINT. APPLIANCE STORE (MATRIX) - DAY\n\n\tThe PHONE begins to RING as the others crawl in.\n\n\t\t\t\t\tSWITCH\n\t\t\tGod, I love that sound."}, {"role": "assistant", "content": "## SCENE 1 — INT. APPLIANCE STORE (MATRIX) - DAY\nACTION: The PHONE begins to RING as the others crawl in. God, I love that sound.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Matrix: The PHONE begins to RING as the others crawl in. God, I love that sound., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The PHONE begins to RING as the others crawl in. God, I love that sound. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Final Destination 2\n\nSCREENPLAY EXCERPT:\nINT. DELIVERY ROOM - CONTINUOUS\n\nIsabella is pushing again.\n\n OBSTETRICIAN\n Almost there. I see the head...\n\nThe NURSE takes a pair of SCISSORS off a tray."}, {"role": "assistant", "content": "## SCENE 1 — INT. DELIVERY ROOM - CONTINUOUS\nACTION: Isabella is pushing again. Almost there. I see the head...\nDIALOGUE — Obstetrician: Almost there. I see the head...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Final Destination 2: Isabella is pushing again. Almost there. I see the head..., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Isabella is pushing again. Almost there. I see the head... The camera pushes in with small amplitude at slow speed. Obstetrician (S1) says: [English] Almost there. I see the head... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in HowIMetYourMother illustrating the trope 'IllBeInMyBunk'. In an episode of How I Met Your Mother, Marshall imagines going back in time to stop his 13-year-old self from smoking his first cigarette by giving him a picture of Lily, to which Young Marshall comments, \"Wow, she's hot.\" Marshall tells his younger self that he will marry her in the future. Surprised and excited, Young Marshall runs off, saying \"I'll be in my tent.\"\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying IllBeInMyBunk.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for HowIMetYourMother, IllBeInMyBunk, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for IllBeInMyBunk. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: a chunk of the ceiling collapses upon Clear's artworks, including the canvas incorporating a jagged piece of debris from Flight 180. The metal SLAMS to the floor, on top of...\nSetting: derived from \"Final Destination\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. GARAGE - CLEAR RIVER'S HOUSE - NIGHT\nACTION: a chunk of the ceiling collapses upon Clear's artworks, including the canvas incorporating a jagged piece of debris from Flight 180. The metal SLAMS to the floor, on top of... which causes the flammable liquid to stream out, toward the driveway.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Final Destination\": a chunk of the ceiling collapses upon Clear's artworks, including the canvas incorporating a jagged piece of debris from Flight 180. The metal SLAMS to the floor, on top of... which causes the flammab, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. a chunk of the ceiling collapses upon Clear's artworks, including the canvas incorporating a jagged piece of debris from Flight 180. The metal SLAMS to the floor, on top of... which causes the flammable liquid to stream out, toward the driveway. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Sponge\n\nSCREENPLAY EXCERPT:\nSUSAN\n I'm going to the ladies room.\n\n\n LENA\n I'll go with you. (They leave.)\n\n\n GEORGE\n What are you looking at me like that\n for?\n\n\n JERRY\n Why'd you have to mention 'unlisted\n number'?\n\n\n GEORGE\n What are you talking about?\n\n\n JERRY\n Alright, I gotta tell you something,\n but you cannot tell Susan."}, {"role": "assistant", "content": "## SCENE 1 — INT. THEY LEAVE.\nACTION: I'm going to the ladies room. I'll go with you. (They leave.)\nDIALOGUE — Lena: I'll go with you. (They leave.)\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Sponge: I'm going to the ladies room. I'll go with you. (They leave.), detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm going to the ladies room. I'll go with you. (They leave.) The camera pushes in with small amplitude at slow speed. Lena (S1) says: [English] I'll go with you. (They leave.) The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BronBroen illustrating the trope 'WellIntentionedExtremist'. Subverted with Jens ; double subverted with the eco terrorists, who are seemingly revealed to be Unwitting Pawns in a money-making scheme, until Gertrud turns out to be a true believer .\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying WellIntentionedExtremist.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BronBroen, WellIntentionedExtremist, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for WellIntentionedExtremist. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Ninja Assassin\n\nSCREENPLAY EXCERPT:\nGOLDENROD REVISED 2nd DRAFT 6/4/08 8A.\n5A CONTINUED: 5A\n\n MASLOW *\n Does that evidence include photos *\n of the Loch Ness monster as well? *\n\n He chuckles. She's not laughing. *\n\n MIKA *\n You've trusted my research before, *\n what's the problem now Ryan? *\n\n MASLOW *\n What's the problem? Come on Mika, *\n it's like one of those questions *\n on an IQ test, which of these *\n doesn't belong; laptops, space *\n shuttles, nanotechnology, ninjas. *\n\n MIKA *\n All I'm asking for is an hour. *\n That's all. Give me an hour and *\n if you still think I'm chasing *\n UFO's, I'll walk out and you'll *\n never hear another thing about *\n them. *\n\n He eyes her. "}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: GOLDENROD REVISED 2nd DRAFT 6/4/08 8A. 5A CONTINUED: 5A\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ninja Assassin: GOLDENROD REVISED 2nd DRAFT 6/4/08 8A. 5A CONTINUED: 5A, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GOLDENROD REVISED 2nd DRAFT 6/4/08 8A. 5A CONTINUED: 5A The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Sunset Boulevard\n\nSCREENPLAY EXCERPT:\nNow we need a topcoat. Let's see\n what you have in camel's hair.\n\n The salesman leaves.\n\n NORMA\n How about some evening clothes?\n\n GILLIS\n I don't need a tuxedo.\n\n NORMA\n Of course you do. A tuxedo and\n tails.\n\n GILLIS\n Tails. That's ridiculous.\n\n NORMA\n You'll need them for parties.\n You'll need them for New Year's\n Eve."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Now we need a topcoat. Let's see what you have in camel's hair.\nDIALOGUE — Norma: How about some evening clothes?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Sunset Boulevard: Now we need a topcoat. Let's see what you have in camel's hair., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Now we need a topcoat. Let's see what you have in camel's hair. The camera pushes in with small amplitude at slow speed. Norma (S1) says: [English] How about some evening clothes? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: FARAMIR and a few survivors flee WEST OSGILIATH, desperately\nSetting: derived from The Lord Of The Rings: The Return Of The King\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. PELENNOR FIELDS - DAY\nACTION: FARAMIR and a few survivors flee WEST OSGILIATH, desperately riding toward MINAS TIRITH . . . in the SKY above DARK SHAPES\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Return Of The King: FARAMIR and a few survivors flee WEST OSGILIATH, desperately riding toward MINAS TIRITH . . . in the SKY above DARK SHAPES, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. FARAMIR and a few survivors flee WEST OSGILIATH, desperately riding toward MINAS TIRITH . . . in the SKY above DARK SHAPES The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Dorothy in the basket of balloon -- Tin Man, Scarecrow and\nSetting: derived from Fade Out:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. PUBLIC SQUARE, EMERALD CITY -- LS -- THE WIZARD AND\nACTION: Dorothy in the basket of balloon -- Tin Man, Scarecrow and Lion standing on platform with them -- people of Oz grouped\nDIALOGUE — Wizard: Good people of Oz, this is positively the finest exhibition ever to be shown --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade Out:: Dorothy in the basket of balloon -- Tin Man, Scarecrow and Lion standing on platform with them -- people of Oz grouped, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Dorothy in the basket of balloon -- Tin Man, Scarecrow and Lion standing on platform with them -- people of Oz grouped The camera pushes in with small amplitude at slow speed. Wizard (S1) says: [English] Good people of Oz, this is positively the finest exhibition ever to be shown -- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Sachiko locks the door behind her and steps to the middle of\nSetting: derived from Dark City\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BATHROOM\nACTION: Sachiko locks the door behind her and steps to the middle of the room, lifts a small round rug off the tiled floor.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dark City: Sachiko locks the door behind her and steps to the middle of the room, lifts a small round rug off the tiled floor., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sachiko locks the door behind her and steps to the middle of the room, lifts a small round rug off the tiled floor. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Peter Jackson, Fran Walsh and Philippa Boyens.\nSetting: derived from The Lord Of The Rings: The Two Towers\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Peter Jackson, Fran Walsh and Philippa Boyens. Based on \"The Lord of The Rings\" trilogy by\nDIALOGUE — Gandalf: You cannot pass!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Two Towers: Peter Jackson, Fran Walsh and Philippa Boyens. Based on \"The Lord of The Rings\" trilogy by, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Peter Jackson, Fran Walsh and Philippa Boyens. Based on \"The Lord of The Rings\" trilogy by The camera pushes in with small amplitude at slow speed. Gandalf (S1) says: [English] You cannot pass! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade Out:\n\nSCREENPLAY EXCERPT:\nInt. Corridor -- MS -- CAMERA TRUCKS back ahead of the\nScarecrow, Lion, Dorothy and Tin Man as they come forward --\nthey stop at intervals, keep the Lion from running back --\nthen they tip-toe forward -- react to echo --\n\n LION\n Wait a minute, fellahs. I was just\n thinkin'. I really don't want to see the\n Wizard this much. I better wait for you\n outside.\n\n SCARECROW\n What's the matter?\n\n TIN MAN\n Oh, he's just ascared again.\n\n\n\n\n (CONTINUED)\n\n 82.\nCONTINUED: (6)\n\n\n DOROTHY\n Don't you know the Wizard's going to give\n you some courage?\n\n LION\n I'd be too scared to ask him for it.\n\n DOROTHY\n Oh, well, then -- we'll ask him for you.\n\n LION\n I'd sooner wait outside.\n\n DOROTHY\n But why? Why?\n\n LION\n Because I'm still scared!\n\n DOROTHY\n Oh, come on.\n\n LION\n Ohh!\n\n SCARECROW\n What happened?\n\n LION\n Somebody pulled my tail.\n\n SCARECROW\n Oh, you did it yourself!\n\n LION\n I -- Oh --\n\n SCARECROW\n Here -- Come on.\n\n LION\n What was that?\n\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. CORRIDOR -- MS -- CAMERA TRUCKS BACK AHEAD OF THE\nACTION: Scarecrow, Lion, Dorothy and Tin Man as they come forward -- they stop at intervals, keep the Lion from running back --\nDIALOGUE — Lion: Wait a minute, fellahs. I was just thinkin'. I really don't want to see the Wizard this much. I better wait for you outside.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade Out:: Scarecrow, Lion, Dorothy and Tin Man as they come forward -- they stop at intervals, keep the Lion from running back --, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Scarecrow, Lion, Dorothy and Tin Man as they come forward -- they stop at intervals, keep the Lion from running back -- The camera pushes in with small amplitude at slow speed. Lion (S1) says: [English] Wait a minute, fellahs. I was just thinkin'. I really don't want to see the Wizard this much. I better wait for you outside. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: He Phone Message\n\nSCREENPLAY EXCERPT:\nKRAMER\n Hey, I just thought of a really funny\n thing for your act. Alright,\n\n\n you're up there, you're on the stage and you go \"Hey, you ever\n notice how cars\n\n\n here in New York, they never get out of the way of ambulances\n anymore.\n\n\n Someone's in a life-and-death situation, and we're thinking 'Well,\n sorry buddy,\n\n\n you should've thought of that when you were eating cheese omelettes\n and sauages\n\n\n for breakfast every morning for the last thirty years.'\" So you\n gonna use it?\n\n\n JERRY\n I don't think so.\n\n\n KRAMER\n It's funny.\n\n\n ELAINE\n It is funny.\n\n\n KRAMER\n That's as good as anything you do.\n\n\n\n GEORGE\n Alright, I gotta make a call. Everybody\n out, come on.\n\n\n JERRY\n Why do we have to leave?\n\n\n GEORGE\n Because I can't call a woman with other\n people in the room. Come\n\n\n on, let's go.\n\n\n ELAINE\n Oh, see, this is the problem.\n\n\n JERRY\n You're kicking me out of my house?\n\n\n\n GEORGE\n Yes.\n\n\n ELAINE\n Don't forget."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Hey, I just thought of a really funny thing for your act. Alright,\nDIALOGUE — Jerry: I don't think so.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from He Phone Message: Hey, I just thought of a really funny thing for your act. Alright,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hey, I just thought of a really funny thing for your act. Alright, The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] I don't think so. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DempseyAndMakepeace illustrating the trope 'WunzaPlot'. One's an American street smart loose cannon cowboy cop, one's a refined aristocratic proper young British police officer. Together They Fight Crime! and make sparks fly between them!\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying WunzaPlot.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DempseyAndMakepeace, WunzaPlot, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for WunzaPlot. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Precrime transports zoom through the city\nSetting: derived from \"Minority Report\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. THE SPRAWL\nACTION: Precrime transports zoom through the city\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Minority Report\": Precrime transports zoom through the city, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Precrime transports zoom through the city The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: No...no. You don't understand. A source...\nSetting: derived from Untitled\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: No...no. You don't understand. A source... a source has told me that Videodrome, _ i\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: No...no. You don't understand. A source... a source has told me that Videodrome, _ i, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. No...no. You don't understand. A source... a source has told me that Videodrome, _ i The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dawn Of\n\nSCREENPLAY EXCERPT:\nThere is a long silence.\n\n\tRoger:\tCOME ON, MARTINEZ!\n\n\tOne of the other S.W.A.T. team members is a big man, with a\n\trough and vicious looking face. He is WOOLEY, a hardened\n\tveteran, and a red neck of the first order.\n\n\tWooley:\t YEAH, COME ON, MARTINEZ...SHOW YOUR GREASY\n\t\t\tLITTLE PUERTO RICAN ASS...SO I CAN BLOW IT OFF...\n\n\tRoger looks over at the big man. He is distressed at the pent up\n\tviolence in Wooley.\n\n\tWooley:\t I'LL BLOW ALL THEIR ASSES OFF...LOW LIFE BASTARDS..\n\t\t\tBLOW ALL THEIR LITTLE LOW LIFE PUERTO RICAN AND\n\t\t\tNIGGER ASSES RIGHT OFF...\n\n\tRoger is greatly concerned. He looks at one of the other men, a\n\tyoung, smoothed faced rookie. The boy doesn't know now to react.\n\tHe is obviously nervous.\n\n\tRoger:\tKEEP COOL. JUST DON'T POP OFF IN THERE WHEN WE\n\t\t\tGO IN.\n\n\tThe boy nods, grateful for a more human contact.\n\n\tWooley:\t HOW THE HELL COME WE STICK THESE LOW LIFES\n\t\t\tIN THESE BIG ASS FANCY HOTELS ANYWAY? SHIT\n\t\t\tMAN. THIS' BETTER THAN I GOT. YOU AIN'T\n\t\t\tGONNA TALK 'EM OUTA HERE. YOU GOTTA BLOW\n\t\t\t'EM OUT. BLOW THEIR ASSES!\n\n\tRoger:\t(to the boy)\n\t\t\tYOU GONNA BE ALRIGHT?\n\n\tThe boy nods in the affirmative.\n\n\tWooley:\t LET'S GET ON WITH IT. THIS IS A WASTE OF\n\t\t\tMY TIME!\n\n44\t CRASH! Without warning, the metal door to the fire stair bursts\n\topen and several figures rush out of the darkness. Shots are\n\tfired from hand guns. A bullet smashes through the skull of the\n\tyoung boy next to Roger. He falls against Roger with a pleading\n\texpression on his face.\n\n\tFigures charge this way and that. More gunfire. The other\n\tS.W.A.T. men dodge and dive for cover. Wooley opens fire wi"}, {"role": "assistant", "content": "## SCENE 1 — INT. TO THE BOY\nACTION: There is a long silence. Roger:\tCOME ON, MARTINEZ!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dawn Of: There is a long silence. Roger:\tCOME ON, MARTINEZ!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. There is a long silence. Roger:\tCOME ON, MARTINEZ! The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dark City\n\nSCREENPLAY EXCERPT:\nINT. BEDROOM\n\n They sit on the edge of the bed - Sachiko holds Elizabeth's\n hand, smiles at her.\n\n ELIZABETH (calmer now)\n I'm sorry... This is embarrassing -\n my husband doesn't know who I am...\n\n With this, she starts to cry again.\n\n Sachiko leans forward and tenderly kisses Elizabeth's lips."}, {"role": "assistant", "content": "## SCENE 1 — INT. BEDROOM\nACTION: They sit on the edge of the bed - Sachiko holds Elizabeth's I'm sorry... This is embarrassing -\nDIALOGUE — Elizabeth: I'm sorry... This is embarrassing - my husband doesn't know who I am...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dark City: They sit on the edge of the bed - Sachiko holds Elizabeth's I'm sorry... This is embarrassing -, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They sit on the edge of the bed - Sachiko holds Elizabeth's I'm sorry... This is embarrassing - The camera pushes in with small amplitude at slow speed. Elizabeth (S1) says: [English] I'm sorry... This is embarrassing - my husband doesn't know who I am... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Are there terrorists on the plane? A\nSetting: derived from The Suicide\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Are there terrorists on the plane? A hotel fire. Is that it? Malaria? Yellow\nDIALOGUE — Newman: He did it right in this bed, Martin. Right in front of you.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Suicide: Are there terrorists on the plane? A hotel fire. Is that it? Malaria? Yellow, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Are there terrorists on the plane? A hotel fire. Is that it? Malaria? Yellow The camera pushes in with small amplitude at slow speed. Newman (S1) says: [English] He did it right in this bed, Martin. Right in front of you. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Big Lebowski\n\nSCREENPLAY EXCERPT:\nHello?\n\n\t\t\t\tDUDE\n\t\tYeah?\n\n\t\t\t\tVOICE\n\t\tOkay, listen--\n\nWalter looks over at the Dude and bellows:\n\n\t\t\t\tWALTER\n\t\tDude, are you fucking this up?\n\n\t\t\t\tVOICE\n\t\tWho is that?\n\n\t\t\t\tDUDE\n\t\tThe driver man, I told you--\n\nClick. Dial tone.\n\n\t\t\t\tDUDE\n\t\tOh shit. Walter.\n\n\t\t\t\tWALTER\n\t\tWhat the fuck is going on there?\n\n\t\t\t\tDUDE\n\t\tThey hung up, Walter! You fucked it\n\t\tup! You fucked it up! Her life was\n\t\tin our hands!\n\n\t\t\t\tWALTER\n\t\tEasy, Dude.\n\n\t\t\t\tDUDE\n\t\tWe're screwed now! We don't get\n\t\tshit and they're gonna kill her!\n\t\tWe're fucked, Walter!\n\n\t\t\t\tWALTER\n\t\tDude, nothing is fucked. Come on.\n\t\tYou're being very unDude. They'll\n\t\tcall back. Look, she kidnapped her--\n\nThe phone chirps.\n\n\t\t\t\tWALTER\n\t\tYa see? Nothing is fucked up here,\n\t\tDude. Nothing is fucked. These\n\t\tguys are fucking amateurs--\n\n\t\t\t\tDUDE\n\t\tShutup, Walter! Don't fucking say\n\t\tpeep when I'm doing business here.\n\n\t\t\t\tWALTER"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Walter looks over at the Dude and bellows: Dude, are you fucking this up?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Big Lebowski: Walter looks over at the Dude and bellows: Dude, are you fucking this up?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Walter looks over at the Dude and bellows: Dude, are you fucking this up? The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Basic Instinct\n\nSCREENPLAY EXCERPT:\nINT. BEDROOM - NIGHT\n\n They are seen rolling and turning around on each other.\n\n Their love making is sensual, sincere.\n\n Later in bed.\n\n NICK\n (calmly, but\n seriously)\n Do you think she killed Johnny\n Boz?\n\n CATHERINE\n (surprised, startled)\n For what... to set me up? She\n loved me she wouldn't frame me.\n\n NICK\n (a beat)\n Maybe she got jealous of Johnny\n Boz, too.\n\n CATHERINE\n (after a beat)\n No, she didn't... she never got\n jealous before... she got excited.\n (a beat)\n I don't have luck with women.\n There was this girl I met while I\n was in college. I slept with her\n once. She started following me\n around, taking my picture. She\n dyed her hair, copied my clothes.\n Lisa something... Oberman. It was\n awful.\n\n A long beat, he looks at her.\n\n NICK\n I thought you didn't do adolescent\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. BEDROOM - NIGHT\nACTION: They are seen rolling and turning around on each other. Their love making is sensual, sincere.\nDIALOGUE — Nick: (calmly, but seriously) Do you think she killed Johnny Boz?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Basic Instinct: They are seen rolling and turning around on each other. Their love making is sensual, sincere., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They are seen rolling and turning around on each other. Their love making is sensual, sincere. The camera pushes in with small amplitude at slow speed. Nick (S1) says: [English] (calmly, but seriously) Do you think she killed Johnny Boz? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Mrs. Teasdale\n\nSCREENPLAY EXCERPT:\nI shall report this indignity the my\n President.\n (To Mrs. Teasdale -\n polite but firm)\n Mrs. Teasdale, I feel this\n regrettable occurrence will plunge\n our countries into war.\n\n MRS. TEASDALE"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I shall report this indignity the my Mrs. Teasdale, I feel this\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Mrs. Teasdale: I shall report this indignity the my Mrs. Teasdale, I feel this, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I shall report this indignity the my Mrs. Teasdale, I feel this The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: 1\tWe see the face of a young woman. She is asleep. It is very\nSetting: derived from Dawn Of\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: 1\tWe see the face of a young woman. She is asleep. It is very quiet at first, as credits appear. The woman's face begins to\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dawn Of: 1\tWe see the face of a young woman. She is asleep. It is very quiet at first, as credits appear. The woman's face begins to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. 1\tWe see the face of a young woman. She is asleep. It is very quiet at first, as credits appear. The woman's face begins to The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BigBrother illustrating the trope 'TwoferTokenMinority'. \nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TwoferTokenMinority.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BigBrother, TwoferTokenMinority, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TwoferTokenMinority. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Puffy Shirt\n\nSCREENPLAY EXCERPT:\nKRAMER\n Alright. I'm ready. (To George) You\n know, I still don't understand - why\n do you want to move back in with your\n parents?\n\n\n GEORGE\n I don't want to! I'm outta money! I\n got 714 dollars left in the bank.\n\n\n\n KRAMER\n Well, move in here.\n\n\n JERRY\n (Stopping the notion) What's that?\n\n\n\n KRAMER\n Why doesn't he just move in here?\n\n\n GEORGE\n (Sarcastic) Yeah, yeah. I'm gonna move\n in with him. He doesn't even let you\n use the toilet!\n\n\n KRAMER\n You can move in with me, if you want.\n\n\n\n GEORGE\n (Sincerely) Thank you.. I, uh.. that\n might not work out."}, {"role": "assistant", "content": "## SCENE 1 — INT. TO GEORGE\nACTION: Alright. I'm ready. (To George) You know, I still don't understand - why\nDIALOGUE — George: I don't want to! I'm outta money! I got 714 dollars left in the bank.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Puffy Shirt: Alright. I'm ready. (To George) You know, I still don't understand - why, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Alright. I'm ready. (To George) You know, I still don't understand - why The camera pushes in with small amplitude at slow speed. George (S1) says: [English] I don't want to! I'm outta money! I got 714 dollars left in the bank. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: \"Laundry-day\" is the only exciting day\nSetting: derived from Good News, Bad News\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SOCK'S VOICE\nACTION: \"Laundry-day\" is the only exciting day in the live of clothes. It is...no,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Good News, Bad News: \"Laundry-day\" is the only exciting day in the live of clothes. It is...no,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. \"Laundry-day\" is the only exciting day in the live of clothes. It is...no, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: He's running a fever.\nSetting: derived from Alien\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: He's running a fever.\nDIALOGUE — Standard: Still unconscious?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Alien: He's running a fever., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He's running a fever. The camera pushes in with small amplitude at slow speed. Standard (S1) says: [English] Still unconscious? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Heavenly Creatures\n\nSCREENPLAY EXCERPT:\nINT. ILAM/JULIET'S BEDROOM - DAYPauline and Juliet kneading and shaping Plasticine with their fingers . . .\n\nPULL BACK to reveal sculpted horses, knights and dragons on the table top."}, {"role": "assistant", "content": "## SCENE 1 — INT. ILAM/JULIET'S BEDROOM - DAYPAULINE AND JULIET KNEADING AND SHAPING PLASTICINE WITH THEIR FINGERS . . .\nACTION: PULL BACK to reveal sculpted horses, knights and dragons on the table top.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Heavenly Creatures: PULL BACK to reveal sculpted horses, knights and dragons on the table top., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. PULL BACK to reveal sculpted horses, knights and dragons on the table top. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Alright. I'm ready. (To George) You\nSetting: derived from The Puffy Shirt\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TO GEORGE\nACTION: Alright. I'm ready. (To George) You know, I still don't understand - why\nDIALOGUE — George: I don't want to! I'm outta money! I got 714 dollars left in the bank.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Puffy Shirt: Alright. I'm ready. (To George) You know, I still don't understand - why, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Alright. I'm ready. (To George) You know, I still don't understand - why The camera pushes in with small amplitude at slow speed. George (S1) says: [English] I don't want to! I'm outta money! I got 714 dollars left in the bank. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Okay, this test is going to be about the\nSetting: derived from T W I N P E A K S\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Okay, this test is going to be about the theorems I told you about last week. You\nDIALOGUE — Laura: Don't tell me now. Tell me right before the test. I won't be able to remember long enough.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T W I N P E A K S: Okay, this test is going to be about the theorems I told you about last week. You, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Okay, this test is going to be about the theorems I told you about last week. You The camera pushes in with small amplitude at slow speed. Laura (S1) says: [English] Don't tell me now. Tell me right before the test. I won't be able to remember long enough. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Die Hard\"\n\nSCREENPLAY EXCERPT:\nObviously, Gail whatever his\n affiliation, it's safe to say\n that Gruber's terrorist actions\n in Los Angeles tonight are well,\n terroristic...\n\n As the bullshit continues, Ellis suddenly STANDS, head towards\n the terrorist \"office\". Immediately the chief guard here,\n Fritz, moves to intercept Ellis.\n\n HOLLY\n Where are you going?\n\n ELLIS\n I'm tired of sitting here waiting\n to see who gets us killed first...\n them...or your husband.\n (to the approaching\n Fritz)\n Hi there.\n\n HOLLY"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Obviously, Gail whatever his affiliation, it's safe to say\nDIALOGUE — Holly: Where are you going?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Die Hard\": Obviously, Gail whatever his affiliation, it's safe to say, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Obviously, Gail whatever his affiliation, it's safe to say The camera pushes in with small amplitude at slow speed. Holly (S1) says: [English] Where are you going? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: That's right! Show that turd who's\nSetting: derived from Austin Powers: International Man Of Mystery\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TEXAN'S STALL\nACTION: That's right! Show that turd who's\nDIALOGUE — Texan: That's right! Show that turd who's boss!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Austin Powers: International Man Of Mystery: That's right! Show that turd who's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. That's right! Show that turd who's The camera pushes in with small amplitude at slow speed. Texan (S1) says: [English] That's right! Show that turd who's boss! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: As the rain falls, we see the Vinyards sitting at the\nSetting: derived from American History X\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. VENICE BEACH HOUSE - A RAINY NIGHT - FLASHBACK\nACTION: As the rain falls, we see the Vinyards sitting at the table. Stacey and Murray Rosenberg, Doris' then\nDIALOGUE — Danny: And then came October 4th, 1993.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American History X: As the rain falls, we see the Vinyards sitting at the table. Stacey and Murray Rosenberg, Doris' then, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. As the rain falls, we see the Vinyards sitting at the table. Stacey and Murray Rosenberg, Doris' then The camera pushes in with small amplitude at slow speed. Danny (S1) says: [English] And then came October 4th, 1993. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: From the nozzle in Thomas' car. Thomas fills the tank, watching for\nSetting: derived from Final Destination 2\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. GAS STATION - DAY\nACTION: From the nozzle in Thomas' car. Thomas fills the tank, watching for signs of danger. Clear's hand is poised over the gas pumps' EMERGENCY\nDIALOGUE — Thomas: New life defeats death? Follow the signs? Where the hell did you find that guy?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Final Destination 2: From the nozzle in Thomas' car. Thomas fills the tank, watching for signs of danger. Clear's hand is poised over the gas pumps' EMERGENCY, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. From the nozzle in Thomas' car. Thomas fills the tank, watching for signs of danger. Clear's hand is poised over the gas pumps' EMERGENCY The camera pushes in with small amplitude at slow speed. Thomas (S1) says: [English] New life defeats death? Follow the signs? Where the hell did you find that guy? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Tombstone\n\nSCREENPLAY EXCERPT:\nSheriff, may I present a pair of\n Fellow sophisticates, Turkey\n Creek Jack Johnson and Texas Jack\n Vermillion? Watch your ear, Creek.\n\nDoc points to his bloody ear. Johnson touches it, sees the\nblood, gives a silent start. Just then White arrives, looking\nweary, facing Johnson and Vermillion.\n\n WHITE\n 'Fraid I'll have to have those guns.\n\n JOHNSON\n Fair fight. We were legal.\n\n WHITE\n Sorry, boys. Gotta take you\n Before Judge Spicer.\n\n VERMILLION\n Well law and order every time,\n That's us.\n\nThey hand over their guns while Virgil looks at the 2 dead men\nlying in the street, shaking his head:\n\n VIRGIL\n What kinda town is this?\n\n VERMILLION\n Nice scenery.\n\nThey look. A stagecoach stops in the street. JOSEPHINE\nMARCUS looks out the window, her little white dog under her\nchin. She and Wyatt spot each other instantly, both\nimpressed.\n\n BEHAN\n That must be the theatrical\n Troupe. There's a show tonight at\n Schieffelin Hall.\n\n JOHNSON\n Hey, Wyatt, you goin' to the\n Show? Maybe we'll see you there."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Sheriff, may I present a pair of Fellow sophisticates, Turkey\nDIALOGUE — White: 'Fraid I'll have to have those guns.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Tombstone: Sheriff, may I present a pair of Fellow sophisticates, Turkey, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sheriff, may I present a pair of Fellow sophisticates, Turkey The camera pushes in with small amplitude at slow speed. White (S1) says: [English] 'Fraid I'll have to have those guns. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in JAG illustrating the trope 'AirstrikeImpossible'. JAG: In the Pilot Movie, Captain Boone, flies a night-time reconnaissance mission in an F-14 Tomcat twice on very low altitude over Bosnia. On the second run hes hit by AA fire, but backseat rider Harm (who has night-blindness and hasn't flown a Tomcat in five years) manages to land it sound and safe on the carrier.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying AirstrikeImpossible.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for JAG, AirstrikeImpossible, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for AirstrikeImpossible. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Come on you son of a bitch, is that all you\nSetting: derived from Hackers\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. COMPUTER ROOM, ELLINGSON MINERAL.\nACTION: Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you\nDIALOGUE — Plague: Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you can do! You talking to me? Huh? Hahahaha. Are you nuts? Come at me!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hackers: Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you The camera pushes in with small amplitude at slow speed. Plague (S1) says: [English] Come on you son of a bitch, is that all you got, huh? Come on, let's see what else you can do! You talking to me? Huh? Hahahaha. Are you nuts? Come at me! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"The Thing\"\n\nSCREENPLAY EXCERPT:\nEXT. ANTARCTICA - ICESCAPE\n\n The two vehicles rip across the hard, flat ice, bolstered by\n the added horsepower. They follow the still visible dog tracks\n in the snow.\n\n CUT TO:\n\n THE SUN\n\n sliding across the horizon, signaling midday. The snowmobiles\n whoosh past. Bennings drives the one loaded with supplies.\n MacReady and Childs double up on the other.\n\n CUT TO:\n\n MACREADY\n\n steadying his binoculars, while Childs drives, spots something\n up ahead. The vehicles slow down and come to a halt. Something\n lies just ahead of them in the whiteness, in the middle of\n the dog tracks.\n\n THE MEN\n\n kneel down by the \"something.\" It is the half-eaten remains\n of a dog. Its hind legs and lower stomach picked clean. Its\n ripped hide, flapping in the wind. Its top half missing.\n\n CHILDS\n What is it?\n\n MacReady follows the line of continuing dog tracks.\n\n MACREADY\n Maybe dinner.\n\n BENNINGS\n Dogs don't eat each other.\n\n MACREADY\n (beat)\n I know.\n\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. ANTARCTICA - ICESCAPE\nACTION: The two vehicles rip across the hard, flat ice, bolstered by the added horsepower. They follow the still visible dog tracks\nDIALOGUE — The Sun: sliding across the horizon, signaling midday. The snowmobiles whoosh past. Bennings drives the one loaded with supplies. MacReady and Childs double up on the...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"The Thing\": The two vehicles rip across the hard, flat ice, bolstered by the added horsepower. They follow the still visible dog tracks, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The two vehicles rip across the hard, flat ice, bolstered by the added horsepower. They follow the still visible dog tracks The camera pushes in with small amplitude at slow speed. The Sun (S1) says: [English] sliding across the horizon, signaling midday. The snowmobiles whoosh past. Bennings drives the one loaded with supplies. MacReady and Childs double up on the... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Malignant\n\nSCREENPLAY EXCERPT:\nEXT. HOSPITAL - NIGHT\nGabrielle, face obscured by her hoodie and trench coat,\nenters through the automatic doors."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HOSPITAL - NIGHT\nACTION: Gabrielle, face obscured by her hoodie and trench coat, enters through the automatic doors.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Malignant: Gabrielle, face obscured by her hoodie and trench coat, enters through the automatic doors., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Gabrielle, face obscured by her hoodie and trench coat, enters through the automatic doors. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Avengers: Endgame\n\nSCREENPLAY EXCERPT:\nINT. CAMP LEHIGH, SHIELD FACILITY, HALLWAY - DAY (1970)\n\nA DOOR MARKED, \"DR. HENRY PYM,\" flies open. Hank exits,\nfuming, brushing past...STEVE.\n\nSteve watches Hank stalk off...THEN DUCKS INTO HANK'S LAB."}, {"role": "assistant", "content": "## SCENE 1 — INT. CAMP LEHIGH, SHIELD FACILITY, HALLWAY - DAY (1970)\nACTION: A DOOR MARKED, \"DR. HENRY PYM,\" flies open. Hank exits, fuming, brushing past...STEVE.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Avengers: Endgame: A DOOR MARKED, \"DR. HENRY PYM,\" flies open. Hank exits, fuming, brushing past...STEVE., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A DOOR MARKED, \"DR. HENRY PYM,\" flies open. Hank exits, fuming, brushing past...STEVE. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Weir enters the Second Containment. Freezes as he sees...\nSetting: derived from \"Event Horizon\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. EVENT HORIZON - SECOND CONTAINMENT\nACTION: Weir enters the Second Containment. Freezes as he sees... Peters body lying twisted and broken before the Core.\nDIALOGUE — Weir: Oh no. Peters...?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Event Horizon\": Weir enters the Second Containment. Freezes as he sees... Peters body lying twisted and broken before the Core., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Weir enters the Second Containment. Freezes as he sees... Peters body lying twisted and broken before the Core. The camera pushes in with small amplitude at slow speed. Weir (S1) says: [English] Oh no. Peters...? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: They are seen rolling and turning around on each other.\nSetting: derived from Basic Instinct\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BEDROOM - NIGHT\nACTION: They are seen rolling and turning around on each other. Their love making is sensual, sincere.\nDIALOGUE — Nick: (calmly, but seriously) Do you think she killed Johnny Boz?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Basic Instinct: They are seen rolling and turning around on each other. Their love making is sensual, sincere., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They are seen rolling and turning around on each other. Their love making is sensual, sincere. The camera pushes in with small amplitude at slow speed. Nick (S1) says: [English] (calmly, but seriously) Do you think she killed Johnny Boz? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: C'mon, Lenny - we'll take a look down\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: C'mon, Lenny - we'll take a look down there together. Then you'll know. You'll\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: C'mon, Lenny - we'll take a look down there together. Then you'll know. You'll, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. C'mon, Lenny - we'll take a look down there together. Then you'll know. You'll The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Natural Born Killers\"\n\nSCREENPLAY EXCERPT:\nINT. STAIRWELL - DAY\n\n The caravan bursts into the stairwell. Mickey turns to Duncan\n the remaining hostage deputy.\n\n MICKEY\n (to Duncan)\n Where does this lead?\n\n Duncan is hyper-ventilating. Mickey pushes him against to\n wall.\n\n MICKEY\n Where!?!\n\n DUNCAN\n Th-- the ground floor.\n\n MALLORY\n Is that the front door?\n\n Duncan nods frantically.\n\n MICKEY\n Let's go.\n\n Mickey grabs Duncan and the caravan starts running down the\n stairs. They go down a few flights.\n\n As they run down one last flight, they find Wurlitzer and a team\n of DEPUTIES waiting for them on the ground floor. The deputies\n raise their guns.\n\n Mallory grabs hold of Wayne, and gets behind him with one hand\n pressing the barrel of her gun against his temple and her other\n are wrapped around his neck, holding him close to her.\n\n MALLORY\n Back off or I'll blast him! Back off\n or I'll blast him! Back off or I'll\n blast him!\n\n None of the deputies lower their guns, but they appear less\n likely to start shooting.\n\n Wayne screams.\n\n WAYNE\n Don't shoot. I beg you, don't shoot!\n Please, please, please...\n\n Wayne continues begging.\n\n Wu"}, {"role": "assistant", "content": "## SCENE 1 — INT. STAIRWELL - DAY\nACTION: The caravan bursts into the stairwell. Mickey turns to Duncan the remaining hostage deputy.\nDIALOGUE — Mickey: Where does this lead?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Natural Born Killers\": The caravan bursts into the stairwell. Mickey turns to Duncan the remaining hostage deputy., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The caravan bursts into the stairwell. Mickey turns to Duncan the remaining hostage deputy. The camera pushes in with small amplitude at slow speed. Mickey (S1) says: [English] Where does this lead? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Asteroid City\n\nSCREENPLAY EXCERPT:\nEXT. HIGHWAY. DAY\n\nA free-standing telephone booth next to a fence-post in the\ncorner of a vacant plot between the gas station and the motel\nentrance. Bits of rubble and bottle-caps litter the dry, cracked\nearth. A single wire extends diagonally up from the top of the\nbooth to an overhead line. Augie waits with the receiver to his\near.\n 9.\n\n\n(Woodrow lingers nearby scribbling periodically in his notebook\nwhile his three sisters run in circles singing eerily and\nswirling soap bubbles into the air with plastic wands).\n\nSPLIT-SCREEN:\n\nOn one side: the interior of the telephone booth with chipped\npaint and pencil-graffitied numbers with abbreviated exchange-\nnames. On the other side: the vast, wood-and-stone, low-ceiling\nliving room of a luxurious ranch house overlooking a golf\ncourse. A houseboy in a yellow butler's jacket answers:\n\n HOUSEBOY\n Zak residence.\n\n AUGIE\n Romulus, this is Augie Steenbeck.\n\n HOUSEBOY\n Good morning, Mr. Augie. The gate is\n open.\n\n AUGIE\n We're not there.\n\n HOUSEBOY\n (mildly surprised)\n You're not here?\n\nThe houseboy checks his watch and frowns.\n\n AUGIE\n May I speak to Mr. Zak?\n\n HOUSEBOY\n Yes, Mr. Augie.\n\nThe houseboy sets down the telephone and hurries away from\ncamera, through the deep room, out via louvered double-doors.\nAugie mops his brow with a handkerchief.\n\nFaintly: t"}, {"role": "assistant", "content": "## SCENE 1 — EXT. HIGHWAY. DAY\nACTION: A free-standing telephone booth next to a fence-post in the corner of a vacant plot between the gas station and the motel\nDIALOGUE — Houseboy: Zak residence.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Asteroid City: A free-standing telephone booth next to a fence-post in the corner of a vacant plot between the gas station and the motel, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A free-standing telephone booth next to a fence-post in the corner of a vacant plot between the gas station and the motel The camera pushes in with small amplitude at slow speed. Houseboy (S1) says: [English] Zak residence. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Final Destination\"\n\nSCREENPLAY EXCERPT:\nEXT. LEWTON'S HOUSE - NIGHT\n\nAlex has moved closer to her house, remaining in the shadows. He checks for any passers-by. Being sure there are none, he creeps toward Ms. Lewton's car. He visually inspects the exterior. As he kicks the tires... an unmarked sedan SCREECHES up. Doors open...\n\nStartled, Alex turns to find...\n\nSpecial agents Schreck and Weine standing in the street, backlit in the strong headlights of their car.\n\n SCHRECK\n What are you doing?\n\nALEX\n\nis nervous but determined. He tells the truth...\n\n ALEX\n Checking the air in her tires to\n make sure they're safe.\n\nSPECIAL AGENTS SCHRECK AND WEINE\n\nafter a beat of incredulity...\n\n SCHRECK\n Get in the car."}, {"role": "assistant", "content": "## SCENE 1 — EXT. LEWTON'S HOUSE - NIGHT\nACTION: Alex has moved closer to her house, remaining in the shadows. He checks for any passers-by. Being sure there are none, he creeps toward Ms. Lewton's car. He visually inspects the exterior. As he kicks the tires... an unmarked sedan SCREECHES up. Doors open... Startled, Alex.\nDIALOGUE — Schreck: What are you doing?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Final Destination\": Alex has moved closer to her house, remaining in the shadows. He checks for any passers-by. Being sure there are none, he creeps toward Ms. Lewton's car. He visually inspects the exterior. As he kicks, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Alex has moved closer to her house, remaining in the shadows. He checks for any passers-by. Being sure there are none, he creeps toward Ms. Lewton's car. He visually inspects the exterior. As he kicks the tires... an unmarked sedan SCREECHES up. Doors open... Startled, Alex. The camera pushes in with small amplitude at slow speed. Schreck (S1) says: [English] What are you doing? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: DUKE pulls out the telegram from his top pocket.\nSetting: derived from Fear & Loathing In Las Vegas\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BAKER TRUCK STOP - DAY\nACTION: DUKE pulls out the telegram from his top pocket. You brainless scumbag! You're\nDIALOGUE — Gonzo'S Voice: You brainless scumbag! You're supposed to be covering the National District Attorney's conference! I made all the reservations... rented a white Cadillac...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fear & Loathing In Las Vegas: DUKE pulls out the telegram from his top pocket. You brainless scumbag! You're, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. DUKE pulls out the telegram from his top pocket. You brainless scumbag! You're The camera pushes in with small amplitude at slow speed. Gonzo'S Voice (S1) says: [English] You brainless scumbag! You're supposed to be covering the National District Attorney's conference! I made all the reservations... rented a white Cadillac... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: GOLDENROD REVISED 2nd DRAFT 6/4/08 43.\nSetting: derived from Ninja Assassin\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: GOLDENROD REVISED 2nd DRAFT 6/4/08 43. 47 CONTINUED: (2) 47\nDIALOGUE — Mika: No...not right now.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ninja Assassin: GOLDENROD REVISED 2nd DRAFT 6/4/08 43. 47 CONTINUED: (2) 47, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GOLDENROD REVISED 2nd DRAFT 6/4/08 43. 47 CONTINUED: (2) 47 The camera pushes in with small amplitude at slow speed. Mika (S1) says: [English] No...not right now. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Tony, I've got a feeling this is\nSetting: derived from The Bodyguard\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Tony, I've got a feeling this is the night. I think he's gonna go\nDIALOGUE — Frank: Tony, I've got a feeling this is the night. I think he's gonna go for her in front of all the cameras. The kind of thing only a lunatic would do. That's how he...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Bodyguard: Tony, I've got a feeling this is the night. I think he's gonna go, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Tony, I've got a feeling this is the night. I think he's gonna go The camera pushes in with small amplitude at slow speed. Frank (S1) says: [English] Tony, I've got a feeling this is the night. I think he's gonna go for her in front of all the cameras. The kind of thing only a lunatic would do. That's how he... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Scarecrow, Lion, Dorothy and Tin Man as they come forward --\nSetting: derived from Fade Out:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CORRIDOR -- MS -- CAMERA TRUCKS BACK AHEAD OF THE\nACTION: Scarecrow, Lion, Dorothy and Tin Man as they come forward -- they stop at intervals, keep the Lion from running back --\nDIALOGUE — Lion: Wait a minute, fellahs. I was just thinkin'. I really don't want to see the Wizard this much. I better wait for you outside.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade Out:: Scarecrow, Lion, Dorothy and Tin Man as they come forward -- they stop at intervals, keep the Lion from running back --, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Scarecrow, Lion, Dorothy and Tin Man as they come forward -- they stop at intervals, keep the Lion from running back -- The camera pushes in with small amplitude at slow speed. Lion (S1) says: [English] Wait a minute, fellahs. I was just thinkin'. I really don't want to see the Wizard this much. I better wait for you outside. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Yes... No... Not until I put someone on\nSetting: derived from Script Date 5/25/94\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Yes... No... Not until I put someone on him. Listen, I need you to send me\nDIALOGUE — Kovash: Keyser - Soze.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Script Date 5/25/94: Yes... No... Not until I put someone on him. Listen, I need you to send me, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Yes... No... Not until I put someone on him. Listen, I need you to send me The camera pushes in with small amplitude at slow speed. Kovash (S1) says: [English] Keyser - Soze. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Tombstone\n\nSCREENPLAY EXCERPT:\nYou win again. Well played, sir.\n You are on fire.\n\n HIGH ROLLER\n Told you. I'm red hot, I'm\n Blazin'! Now, red seven. Seven\n Stinkin' diamonds. Look out! Five\n Thousand! Let's go!\n\n WYATT\n Awful lot of money.\n\n HIGH ROLLER\n Can't take the heat, get outta\n The kitchen.\n\n WYATT\n You're the doctor."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You win again. Well played, sir. Told you. I'm red hot, I'm\nDIALOGUE — High Roller: Told you. I'm red hot, I'm Blazin'! Now, red seven. Seven Stinkin' diamonds. Look out! Five Thousand! Let's go!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Tombstone: You win again. Well played, sir. Told you. I'm red hot, I'm, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You win again. Well played, sir. Told you. I'm red hot, I'm The camera pushes in with small amplitude at slow speed. High Roller (S1) says: [English] Told you. I'm red hot, I'm Blazin'! Now, red seven. Seven Stinkin' diamonds. Look out! Five Thousand! Let's go! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: SPEEDING POV: through the forest from the back of the White\nSetting: derived from Black Screen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. TROLLSHAW FOREST -- NIGHT\nACTION: SPEEDING POV: through the forest from the back of the White ANGLE ON: FRODO, BOUNCING IN THE SADDLE,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: SPEEDING POV: through the forest from the back of the White ANGLE ON: FRODO, BOUNCING IN THE SADDLE,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. SPEEDING POV: through the forest from the back of the White ANGLE ON: FRODO, BOUNCING IN THE SADDLE, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Lord Of The Rings: The Return Of The King\n\nSCREENPLAY EXCERPT:\nINT. CRACK OF DOOM\n\nGOLLUM suddenly LIFTS HIS HANDS to his face ... and BITES\nHARD!\n\nANGLE ON: FRODO MATERIALIZES as he DROPS TO HIS KNEES,\nclutching his BLEEDING HAND . . . he SCREAMS.\n\n (CONTINUED)\n\n Final Revision - October, 2003 136,\nCONTINUED:\n\n\nThe FIRES below roar in anger, RED LIGHT blazes, and all the\ncavern is filled with a great glare and heat.\n\n INTERCUT WITH:"}, {"role": "assistant", "content": "## SCENE 1 — INT. CRACK OF DOOM\nACTION: GOLLUM suddenly LIFTS HIS HANDS to his face ... and BITES ANGLE ON: FRODO MATERIALIZES as he DROPS TO HIS KNEES,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Return Of The King: GOLLUM suddenly LIFTS HIS HANDS to his face ... and BITES ANGLE ON: FRODO MATERIALIZES as he DROPS TO HIS KNEES,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GOLLUM suddenly LIFTS HIS HANDS to his face ... and BITES ANGLE ON: FRODO MATERIALIZES as he DROPS TO HIS KNEES, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Script Date 5/25/94\n\nSCREENPLAY EXCERPT:\nI don't know what you came here for, but\n we won't have any part of it.\n\n KEATON\n Edie, please.\n\nKeaton takes Edie by the arm and tries to guide her toward\nthe other room. She pulls away, anger turning to rage.\n\n EDIE\n I've spent the last year of my life\n putting his back together again - I won't\n have you come in here and - What makes\n you think - GET OUT. GET OUT OF m HOME.\n HOW DARE YOU COME HERE?\n\nKeaton is pulling her now. She yanks her arm away and shoves\nhim.\n\n YELLOW 06/11/94\n\n\n 42.\n\n EDIE\n Don't touch me. Just don't,\n\nShe turns and walks out of the room. Somewhere in the back of\nthe apartment, a door slams.\n\nKeaton turns and glares at Verbal. Verbal cringes.\n\n KEATON\n Get out.\n\n VERBAL\n If you'll just let me -\n\nSUDDENLY, KEATON LUNGES. He grabs Verbal by the lapels and\nlifts him off the couch, moving him effortlessly across the\nroom and slamming him into the wall next to the front door.\nHe opens it.\n\n VERBAL (CONT'D)\n Don't hurt me.\n\n KEATON"}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: I don't know what you came here for, but we won't have any part of it.\nDIALOGUE — Keaton: Edie, please.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Script Date 5/25/94: I don't know what you came here for, but we won't have any part of it., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I don't know what you came here for, but we won't have any part of it. The camera pushes in with small amplitude at slow speed. Keaton (S1) says: [English] Edie, please. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dogma\n\nSCREENPLAY EXCERPT:\nINT. MOOBY CORP. BOARDROOM - DAY\n\nOh, this isn't your standard boardroom; this is Mooby Corp., home of Mooby,\nthe Golden Calf - which can only be described as a bovine variation on\nBarney: sickeningly simple and very non-threatening. A large table sits in\nthe middle, a media center behind the huge chair at the head. The walls are\nadorned with framed posters of Mooby, playing with kids, mouth agape in a\nstupid smile. At the center of the table is a large, gold plated statue of\nthe insipid creature.\n\nDoors open and the boardroom fills with suits - six men, one woman. They\nchatter and take their seats. After a beat, WHITLAND, the CEO, enters,\ntaking his place at the head of the conference table.\n\n WHITLAND\n\nGood morning, shoppers.\n\n(slaps a file on table)\n\nHas anyone seen the over-night's?\n\nAn anticipatory hush fills the room.\n\n WHITLAND\n\nWe creamed 'em.\n\nA cheer and applause goes up from the group. Whitland smiles.\n\n WHITLAND\n\n(reading from file)\n\nAnd last night was a rerun, which says to me that with the six months we\nhave to ready and promote the 'Very Nlooby Christmas' pay-per-view special,\nwe can produce history-making numbers. The record is held by that\nshock-jock's New Year's thing, but I see no reason why our little cash cow\ncan't supercede those numbers and...\n\n(stops and sniffs the air)\n\nDo I smell onions?\n\nBartleby and Loki sit behind the thrall on a black leather couch. Loki is\ncarving something out of an onion, while Bartleby looks on.\n\nWhitland and the rest of the boar"}, {"role": "assistant", "content": "## SCENE 1 — INT. MOOBY CORP. BOARDROOM - DAY\nACTION: Oh, this isn't your standard boardroom; this is Mooby Corp., home of Mooby, the Golden Calf - which can only be described as a bovine variation on\nDIALOGUE — Whitland: Good morning, shoppers.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dogma: Oh, this isn't your standard boardroom; this is Mooby Corp., home of Mooby, the Golden Calf - which can only be described as a bovine variation on, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Oh, this isn't your standard boardroom; this is Mooby Corp., home of Mooby, the Golden Calf - which can only be described as a bovine variation on The camera pushes in with small amplitude at slow speed. Whitland (S1) says: [English] Good morning, shoppers. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Joker\n\nSCREENPLAY EXCERPT:\nI'm sorry. Did I bring what?\n\n SOCIAL WORKER\n (impatient; she doesn't\n have time for this)\n Arthur, last time I asked you to\n bring your journal with you. For\n these appointments. Do you have it?\n\n JOKER\n Yes ma'am.\n\nBeat.\n\n SOCIAL WORKER\n Can I see it?\n\nHe reluctantly reaches into his bag. Pulls out a weathered\nnotebook. Slides it across to her--\n\n JOKER\n I've been using it as a journal,\n but also a joke diary. Funny\n thoughts or, or observations-- Did\n I tell you I'm pursuing a career in\n stand-up comedy?\n\nShe's half-listening as she flips through his journal.\n\n SOCIAL WORKER\n No. You didn't.\n\n JOKER\n I think I did.\n 3.\n\n\nShe doesn't respond, keeps flipping through his journal--\n\n SOCIAL WORKER\n Oh yeah. Because of what your\n mother said,-- about your purpose.\n \"To bring laughter and joy to the\n world,\" right?\n\n JOKER\n Right.\n\nANGLE ON JOURNAL, pages and pages of notes, all in neat,\nangry-looking handwriting. Also, cut out photos from hardcore\npornographic magazines and some crude handmade drawings.\n\nA flash of anger crosses Joker's face. We see him picking at\nhis right eyebrow, almost obsessively. Trying to stay calm.\nHis eyebrow is actually half-gone. Something he does a lot.\n\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I'm sorry. Did I bring what? (impatient; she doesn't\nDIALOGUE — Social Worker: (impatient; she doesn't have time for this) Arthur, last time I asked you to bring your journal with you. For these appointments. Do you have it?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Joker: I'm sorry. Did I bring what? (impatient; she doesn't, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm sorry. Did I bring what? (impatient; she doesn't The camera pushes in with small amplitude at slow speed. Social Worker (S1) says: [English] (impatient; she doesn't have time for this) Arthur, last time I asked you to bring your journal with you. For these appointments. Do you have it? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Int. Publishing House Reception Area - Day\n\nSCREENPLAY EXCERPT:\nINT. CHINESE RESTAURANT - NIGHT\n\nJoel finds himself eating Chinese food and sitting across\nfrom Clementine. He is ragged and jarred.\n\n\t\tJOEL\n\tI'm done, Clem. I'm just going to ride\n\tit out. Hiding is clearly not working.\n\n\t\tCLEMENTINE\n\tYeah.\n\n\t\tJOEL\n\tI want to enjoy my little time left with\n\tyou.\n\n\t\tCLEMENTINE\n\tThis is our first \"date\" date.\n\n\t\tJOEL\n\tDo you remember what we talked about?\n\n\t\tCLEMENTINE\n\tNaomi, I guess.\n\n\t\tJOEL\n\tYeah.\n\n\t\tCLEMENTINE\n\tWhat was I wearing?\n\n\t\tJOEL\n\tGod, I should know. Your hair was red.\n\tI remember it matched the wallpaper.\n\n\t\tCLEMENTINE\n\tEgad, were you horrified?\n\n\t\tJOEL\n\tNo! I think you were wearing that black\n\tdress, y'know, with the buttons.\n\nShe is wearing the black dress.\n\n\t\tCLEMENTINE\n\tNo, you were with me when I bought that.\n\tAt that place on East 6th. It was later."}, {"role": "assistant", "content": "## SCENE 1 — INT. CHINESE RESTAURANT - NIGHT\nACTION: Joel finds himself eating Chinese food and sitting across from Clementine. He is ragged and jarred.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int. Publishing House Reception Area - Day: Joel finds himself eating Chinese food and sitting across from Clementine. He is ragged and jarred., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Joel finds himself eating Chinese food and sitting across from Clementine. He is ragged and jarred. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The dog reaches Norris and Bennings, as they awkwardly wade\nSetting: derived from \"The Thing\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. COMPOUND\nACTION: The dog reaches Norris and Bennings, as they awkwardly wade through the snow, toward the downed copter.\nDIALOGUE — The Survivor: of the crash, his eyes crazed with determination, struggles to his feet. Heedless of his companion, he double-times his way to the men and the dog. He reloads...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"The Thing\": The dog reaches Norris and Bennings, as they awkwardly wade through the snow, toward the downed copter., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The dog reaches Norris and Bennings, as they awkwardly wade through the snow, toward the downed copter. The camera pushes in with small amplitude at slow speed. The Survivor (S1) says: [English] of the crash, his eyes crazed with determination, struggles to his feet. Heedless of his companion, he double-times his way to the men and the dog. He reloads... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: They stare at Justin's monitor.\nSetting: derived from \"Event Horizon\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LEWIS AND CLARK - BRIDGE\nACTION: They stare at Justin's monitor. What the hell is that?\nDIALOGUE — Smith: What the hell is that?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Event Horizon\": They stare at Justin's monitor. What the hell is that?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They stare at Justin's monitor. What the hell is that? The camera pushes in with small amplitude at slow speed. Smith (S1) says: [English] What the hell is that? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Outside the room. Dignan closes the door. There's three\nSetting: derived from Bottle Rocket\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. MOTEL HALLWAY. NIGHT\nACTION: Outside the room. Dignan closes the door. There's three housekeeping carts in the hall.\nDIALOGUE — Dignan: What the fuck is going on here?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bottle Rocket: Outside the room. Dignan closes the door. There's three housekeeping carts in the hall., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Outside the room. Dignan closes the door. There's three housekeeping carts in the hall. The camera pushes in with small amplitude at slow speed. Dignan (S1) says: [English] What the fuck is going on here? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Inglourious Basterds\n\nSCREENPLAY EXCERPT:\nSGT.DONOWITZ\n After-teri, you-a set-ta boom-a,\n five-o moment-o, you-a, fuck-o Pphisst.\n (After you set the bomb, wait five\n minutes and get the fuck out of here.)\n\n HIRSCHBERG\n Affirm-ato, affirm-ato."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: After-teri, you-a set-ta boom-a, five-o moment-o, you-a, fuck-o Pphisst.\nDIALOGUE — Hirschberg: Affirm-ato, affirm-ato.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Inglourious Basterds: After-teri, you-a set-ta boom-a, five-o moment-o, you-a, fuck-o Pphisst., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. After-teri, you-a set-ta boom-a, five-o moment-o, you-a, fuck-o Pphisst. The camera pushes in with small amplitude at slow speed. Hirschberg (S1) says: [English] Affirm-ato, affirm-ato. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Bunce sits on two stacked telephone books on a chair. He guts\nSetting: derived from Fantastic Mr. Fox\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BUNCE'S KITCHEN. DAY\nACTION: Bunce sits on two stacked telephone books on a chair. He guts a dead goose, cutting out its liver and mashing it with a\nDIALOGUE — Badger: He eats only doughnuts with smashed-up goose livers injected into them.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fantastic Mr. Fox: Bunce sits on two stacked telephone books on a chair. He guts a dead goose, cutting out its liver and mashing it with a, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Bunce sits on two stacked telephone books on a chair. He guts a dead goose, cutting out its liver and mashing it with a The camera pushes in with small amplitude at slow speed. Badger (S1) says: [English] He eats only doughnuts with smashed-up goose livers injected into them. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Dade hacks into an electronic personal ads system and\nSetting: derived from Hackers\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEW YORK CITY PUBLIC PHONE.\nACTION: Dade hacks into an electronic personal ads system and Alright, he's in the personal ads.\nDIALOGUE — Kate: Alright, he's in the personal ads. \"Disappointed white male, crossdresser, looking for discreet friend to bring dreams to reality. Leather, lace, and water...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hackers: Dade hacks into an electronic personal ads system and Alright, he's in the personal ads., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Dade hacks into an electronic personal ads system and Alright, he's in the personal ads. The camera pushes in with small amplitude at slow speed. Kate (S1) says: [English] Alright, he's in the personal ads. \"Disappointed white male, crossdresser, looking for discreet friend to bring dreams to reality. Leather, lace, and water... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Anytime you want to go home,\nSetting: derived from \"Die Hard\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Anytime you want to go home, Sergeant...consider yourself dismissed.\nDIALOGUE — Powell: No Sir. You couldn't drag me away.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Die Hard\": Anytime you want to go home, Sergeant...consider yourself dismissed., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Anytime you want to go home, Sergeant...consider yourself dismissed. The camera pushes in with small amplitude at slow speed. Powell (S1) says: [English] No Sir. You couldn't drag me away. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Basic Instinct\n\nSCREENPLAY EXCERPT:\nINT. THE DETECTIVE BUREAU - DAY\n\n GUS\n (seriously)\n I'm gonna make you believe in it,\n friend."}, {"role": "assistant", "content": "## SCENE 1 — INT. THE DETECTIVE BUREAU - DAY\nACTION: I'm gonna make you believe in it,\nDIALOGUE — Gus: I'm gonna make you believe in it, friend.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Basic Instinct: I'm gonna make you believe in it,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm gonna make you believe in it, The camera pushes in with small amplitude at slow speed. Gus (S1) says: [English] I'm gonna make you believe in it, friend. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: [The letters are in soft-purple against a black background. Purple is\nSetting: derived from Titles: \"E.T. - The Extra-Terrestial\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. POSSIBLY ALIEN PLANTS\nACTION: [The letters are in soft-purple against a black background. Purple is traditionally the color of that which is sacred.]\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Titles: \"E.T. - The Extra-Terrestial\": [The letters are in soft-purple against a black background. Purple is traditionally the color of that which is sacred.], detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. [The letters are in soft-purple against a black background. Purple is traditionally the color of that which is sacred.] The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: How could you do it? After\nSetting: derived from \"Total Recall\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: How could you do it? After I'm not your wife, Quail.\nDIALOGUE — Kirsten: I'm not your wife, Quail.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Total Recall\": How could you do it? After I'm not your wife, Quail., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. How could you do it? After I'm not your wife, Quail. The camera pushes in with small amplitude at slow speed. Kirsten (S1) says: [English] I'm not your wife, Quail. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Saw\n\nSCREENPLAY EXCERPT:\nWell there�s...there�s not much to tell,\n really. (Across the room, Adam gives\n up with the chain for the time being\n but remains sitting on the floor.) I\n was on my way home from work and uh,\n I don�t remember anything else.\n\n Adam just sits there, glancing over to the body and then focusing\n on it much more. The initial panic has worn off for now, replaced\n simply with fear and shock. His voice is a bit softer.\n\n ADAM\n First dead body I�ve ever seen. Look\n different in real life. They don�t move.\n\n\n Lawrence slowly lowers himself into a sitting position against\n the wall. He pokes slightly at the iron cuff around his foot.\n\n\n LAWRENCE\n From the looks of these chains, someone\n didn�t want us to go very far either.\n\n\n Suddenly, Adam leaps up, lifting up his shirt and exposing his\n abdomen. His panic is back, but not at full force like before.\n He studies himself then glances to Lawrence.\n\n ADAM\n Can you see any scars?\n\n LAWRENCE\n (looking at him strangely) What?\n\n ADAM\n Huh? This is what they do man, (he "}, {"role": "assistant", "content": "## SCENE 1 — INT. LOOKING AT HIM STRANGELY\nACTION: Well there�s...there�s not much to tell, really. (Across the room, Adam gives\nDIALOGUE — Adam: First dead body I�ve ever seen. Look different in real life. They don�t move.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Saw: Well there�s...there�s not much to tell, really. (Across the room, Adam gives, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Well there�s...there�s not much to tell, really. (Across the room, Adam gives The camera pushes in with small amplitude at slow speed. Adam (S1) says: [English] First dead body I�ve ever seen. Look different in real life. They don�t move. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"The Thing\"\n\nSCREENPLAY EXCERPT:\nINT. REC ROOM\n\n MacReady, cursing, rips the speakers off the wall.\n\n MACREADY\n What are they doing out there?!\n\n The music is now subdued within the room, but continues\n booming throughout the camp. Nauls' scream can be barely\n heard.\n\n GARRY\n What's he saying?"}, {"role": "assistant", "content": "## SCENE 1 — INT. REC ROOM\nACTION: MacReady, cursing, rips the speakers off the wall. What are they doing out there?!\nDIALOGUE — Macready: What are they doing out there?!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"The Thing\": MacReady, cursing, rips the speakers off the wall. What are they doing out there?!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. MacReady, cursing, rips the speakers off the wall. What are they doing out there?! The camera pushes in with small amplitude at slow speed. Macready (S1) says: [English] What are they doing out there?! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Coraline\n\nSCREENPLAY EXCERPT:\nINT. PASSAGEWAY � CONTINUOUS\n\n\n\n Hunched down, she moves as quickly as she can, tearing\n through the sticky cobwebs.\n\n\n\n TALL OTHER MOTHER (O.S.)\n (calling after her)\n Coraline!\n\n\n Strands of web brush her face, stick to her hair; she\n closes her eyes, hands out until she finally gets to the\n little door at the other end. She DIVES, TUCKS, --"}, {"role": "assistant", "content": "## SCENE 1 — INT. PASSAGEWAY � CONTINUOUS\nACTION: Hunched down, she moves as quickly as she can, tearing through the sticky cobwebs.\nDIALOGUE — Tall Other Mother: Coraline!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Coraline: Hunched down, she moves as quickly as she can, tearing through the sticky cobwebs., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hunched down, she moves as quickly as she can, tearing through the sticky cobwebs. The camera pushes in with small amplitude at slow speed. Tall Other Mother (S1) says: [English] Coraline! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DoctorWho illustrating the trope 'GoodAdulteryBadAdultery'. Blackpool features both. Ripley's affairs are casual sex, while Natalie's is about love. Ripley eventually tells her to go be with the man she loves. But who can compete with The Doctor?\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying GoodAdulteryBadAdultery.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DoctorWho, GoodAdulteryBadAdultery, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for GoodAdulteryBadAdultery. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: We see the team assembled around a table. They're drinking\nSetting: derived from Bottle Rocket\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT/EXT. PLANNING SEQUENCE. DAY\nACTION: We see the team assembled around a table. They're drinking coffee and looking at pictures, charts, maps, diagrams,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bottle Rocket: We see the team assembled around a table. They're drinking coffee and looking at pictures, charts, maps, diagrams,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We see the team assembled around a table. They're drinking coffee and looking at pictures, charts, maps, diagrams, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Puffy Shirt\n\nSCREENPLAY EXCERPT:\nMAN\n Ugh.. Ray.\n\n\n PHOTOGRAPHER\n He was it.\n\n\n GEORGE\n Who was he?\n\n\n PHOTOGRAPHER\n The most exquisite hands you've ever\n seen.. Oh, he had it all.\n\n\n GEORGE\n (Hands still out, even though they've\n stopped looking at them) What happened\n to him?\n\n\n (Obviously a touchy subject, the woman coyly walks over to the\n photographer, and they both occupy themselves. The man is left\n to tell George the answer to his\n\n\n question)\n\n\n MAN\n (Clears throat) Tragic story, I'm afraid.\n He could've had any woman in the world..\n but none could match the beauty of his\n own hand.. and that became his\n\n\n one true love.."}, {"role": "assistant", "content": "## SCENE 1 — INT. CLEARS THROAT\nACTION: The most exquisite hands you've ever seen.. Oh, he had it all.\nDIALOGUE — Photographer: He was it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Puffy Shirt: The most exquisite hands you've ever seen.. Oh, he had it all., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The most exquisite hands you've ever seen.. Oh, he had it all. The camera pushes in with small amplitude at slow speed. Photographer (S1) says: [English] He was it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You win again. Well played, sir.\nSetting: derived from Tombstone\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You win again. Well played, sir. Told you. I'm red hot, I'm\nDIALOGUE — High Roller: Told you. I'm red hot, I'm Blazin'! Now, red seven. Seven Stinkin' diamonds. Look out! Five Thousand! Let's go!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Tombstone: You win again. Well played, sir. Told you. I'm red hot, I'm, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You win again. Well played, sir. Told you. I'm red hot, I'm The camera pushes in with small amplitude at slow speed. High Roller (S1) says: [English] Told you. I'm red hot, I'm Blazin'! Now, red seven. Seven Stinkin' diamonds. Look out! Five Thousand! Let's go! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Lord Of The Rings: The Two Towers\n\nSCREENPLAY EXCERPT:\nTHE LORD OF THE RINGS: THE TWO TOWERS\n\n\n Screenplay by\n Peter Jackson, Fran Walsh and Philippa Boyens.\n\n Based on \"The Lord of The Rings\" trilogy by\n J.R.R Tolkien.\n\n [TITLE: THE LORD OF THE RINGS]\n\n\n\n [Camera pans over the Misty Mountains\n as voices drift in from the background.]\n\n\n GANDALF\n You cannot pass!\n\n FRODO\n Gandalf!\n\n GANDALF\n I am a servant of the Secret Fire, wielder\n of the Flame of Anor!\n\n [Camera pans closer to the mountain\n side.]\n\n GANDALF\n Argh! Go back to the shadow. The Dark\n Fire will not avail you, Flame of Ud�n!\n [Camera zooms in through the mountain\n and focuses on Gandalf and the Balrog\n on the bridge of Khazad-d�m. The Balrog\n strikes down on Gandalf with its flaming\n sword. Gandalf parries the blow with\n Glamdring, shattering the Balrog�s sword.]\n YOU� SHALL NOT... PASS!!!\n\n [Gandalf strikes his staff onto the\n b"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Peter Jackson, Fran Walsh and Philippa Boyens. Based on \"The Lord of The Rings\" trilogy by\nDIALOGUE — Gandalf: You cannot pass!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Two Towers: Peter Jackson, Fran Walsh and Philippa Boyens. Based on \"The Lord of The Rings\" trilogy by, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Peter Jackson, Fran Walsh and Philippa Boyens. Based on \"The Lord of The Rings\" trilogy by The camera pushes in with small amplitude at slow speed. Gandalf (S1) says: [English] You cannot pass! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Suicide\n\nSCREENPLAY EXCERPT:\nGEORGE\n Are there terrorists on the plane? A\n hotel fire. Is that it? Malaria? Yellow\n fever? Lupus? Is it Lupus?\n\n\n (Martin's room � Martin grabbing Jerry � Elaine eating the Drake's\n Coffee Cake)\n\n\n NEWMAN\n He did it right in this bed, Martin.\n Right in front of you.\n\n\n KRAMER\n I want my vacuum cleaner!\n\n\n JERRY\n Hey!\n\n\n NEWMAN\n It was disgusting."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Are there terrorists on the plane? A hotel fire. Is that it? Malaria? Yellow\nDIALOGUE — Newman: He did it right in this bed, Martin. Right in front of you.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Suicide: Are there terrorists on the plane? A hotel fire. Is that it? Malaria? Yellow, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Are there terrorists on the plane? A hotel fire. Is that it? Malaria? Yellow The camera pushes in with small amplitude at slow speed. Newman (S1) says: [English] He did it right in this bed, Martin. Right in front of you. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: WIDE ON: The fellowship run across a forest floor strewn with\nSetting: derived from Black Screen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. EDGE OF LOTHLORIEN--DUSK\nACTION: WIDE ON: The fellowship run across a forest floor strewn with yellow flowers..above is a roof of golden leaves, held up by\nDIALOGUE — Gimli: Stay close, young hobbits..they say a Sorceress lives in these woods. An elf- witch of terrible power. All who look upon her fall under her spell...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: WIDE ON: The fellowship run across a forest floor strewn with yellow flowers..above is a roof of golden leaves, held up by, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. WIDE ON: The fellowship run across a forest floor strewn with yellow flowers..above is a roof of golden leaves, held up by The camera pushes in with small amplitude at slow speed. Gimli (S1) says: [English] Stay close, young hobbits..they say a Sorceress lives in these woods. An elf- witch of terrible power. All who look upon her fall under her spell... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Mrs. Teasdale\n\nSCREENPLAY EXCERPT:\nThey're coming...\n (He recites the\n following command to\n Harpo)\n Ride through every village and town\n wake every citizen up hill and down\n tell them the enemy comes from afar\n with a hey-nonny-nonny and a ha-cha-\n char\n\n CAMERA TRUCKS back disclosing a real horse beside Harpo.\n\n GROUCHO\n There'll be two lamps in the steeple\n if they're coming by land and one\n lamp if they're coming by sea.\n\n He and Harpo go to the window and look out.\n\n CUT TO SHOT of a steeple with three lighted lamps -\n\n CUT BACK to Groucho and Harpo\n\n GROUCHO\n They double-crossed me... they're\n coming by land and sea -- Be off, my\n lad!\n\n With a grand gesture, Groucho exits out of scene.\n\n Harpo grips pommel of the saddle with both hands and in his\n attempt to make a flying mount, vaults clear over the horse\n landing on the other side. He lets down a rope ladder from\n under the saddle, climbs the ladder, mounts the horse and\n starts off.\n\n DISSOLVE to CLOSE UP of Harpo riding at breakneck speed. As\n he looks grimly ahead "}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Ride through every village and town wake every citizen up hill and down\nDIALOGUE — Groucho: There'll be two lamps in the steeple if they're coming by land and one lamp if they're coming by sea.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Mrs. Teasdale: Ride through every village and town wake every citizen up hill and down, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Ride through every village and town wake every citizen up hill and down The camera pushes in with small amplitude at slow speed. Groucho (S1) says: [English] There'll be two lamps in the steeple if they're coming by land and one lamp if they're coming by sea. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: He lowers a pair of binoculars and passes them to Whitman.\nSetting: derived from \"Hellboy\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEARBY - BROOM - NIGHT\nACTION: He lowers a pair of binoculars and passes them to Whitman. If he's here, this is worse than I\nDIALOGUE — Broom: If he's here, this is worse than I thought.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hellboy\": He lowers a pair of binoculars and passes them to Whitman. If he's here, this is worse than I, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He lowers a pair of binoculars and passes them to Whitman. If he's here, this is worse than I The camera pushes in with small amplitude at slow speed. Broom (S1) says: [English] If he's here, this is worse than I thought. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Jacob'S Ladder\"\n\nSCREENPLAY EXCERPT:\nEXT. GREENWICH VILLAGE - DAY\n\nJACOB and MICHAEL are walking through the STREETS OF GREENWICH VILLAGE.\nIt is early MORNING and the sidewalks are bustling with PEOPLE. JACOB\nstares into their faces and beams when they smile back. MICHAEL enjoys\nJACOB's happiness."}, {"role": "assistant", "content": "## SCENE 1 — EXT. GREENWICH VILLAGE - DAY\nACTION: JACOB and MICHAEL are walking through the STREETS OF GREENWICH VILLAGE. It is early MORNING and the sidewalks are bustling with PEOPLE. JACOB\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Jacob'S Ladder\": JACOB and MICHAEL are walking through the STREETS OF GREENWICH VILLAGE. It is early MORNING and the sidewalks are bustling with PEOPLE. JACOB, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. JACOB and MICHAEL are walking through the STREETS OF GREENWICH VILLAGE. It is early MORNING and the sidewalks are bustling with PEOPLE. JACOB The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Now we need a topcoat. Let's see\nSetting: derived from Sunset Boulevard\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Now we need a topcoat. Let's see what you have in camel's hair.\nDIALOGUE — Norma: How about some evening clothes?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Sunset Boulevard: Now we need a topcoat. Let's see what you have in camel's hair., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Now we need a topcoat. Let's see what you have in camel's hair. The camera pushes in with small amplitude at slow speed. Norma (S1) says: [English] How about some evening clothes? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Two cars parked outside. An old, quirky Volvo and a Land\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CATHERINE'S HOUSE - DAY\nACTION: Two cars parked outside. An old, quirky Volvo and a Land\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: Two cars parked outside. An old, quirky Volvo and a Land, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Two cars parked outside. An old, quirky Volvo and a Land The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Neo's face is knotted, teeth clenched, as he hurls\nSetting: derived from The Matrix\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. DOJO\nACTION: Neo's face is knotted, teeth clenched, as he hurls Good. Adaption. Improvisation.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Matrix: Neo's face is knotted, teeth clenched, as he hurls Good. Adaption. Improvisation., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Neo's face is knotted, teeth clenched, as he hurls Good. Adaption. Improvisation. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Legion\n\nSCREENPLAY EXCERPT:\nINT. PARADISE FALLS GAS 'N' GRUB - CONTINUOUS\n\n Michael reacts to the sound. An urgency comes over him.\n Audrey arrives with more steaming water.\n\n AUDREY\n What is that?!\n\n MICHAEL\n He's coming. We have to hurry.\n\n AUDREY\n Who's coming? Whaddya mean hurry?\n This isn't something you can\n exactly hurry, is it?\n\n MICHAEL\n Charlie, I need you to push. I\n need you to push as hard as you\n can.\n\n AUDREY\n Michael, who's coming?!\n Charlie pushes, tears streaming down her face, delirious with\n pain. It's a gruesome, bloody affair. It's getting close.\n\n MICHAEL\n We're almost there...\n\n\n\n 81."}, {"role": "assistant", "content": "## SCENE 1 — INT. PARADISE FALLS GAS 'N' GRUB - CONTINUOUS\nACTION: Michael reacts to the sound. An urgency comes over him. Audrey arrives with more steaming water.\nDIALOGUE — Audrey: What is that?!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Legion: Michael reacts to the sound. An urgency comes over him. Audrey arrives with more steaming water., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Michael reacts to the sound. An urgency comes over him. Audrey arrives with more steaming water. The camera pushes in with small amplitude at slow speed. Audrey (S1) says: [English] What is that?! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: So young lady, you are to close\nSetting: derived from Inglourious Basterds\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: So young lady, you are to close your cinema tonight, and have\nDIALOGUE — Goebbels: What German films do you have? Francesca asks..\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Inglourious Basterds: So young lady, you are to close your cinema tonight, and have, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. So young lady, you are to close your cinema tonight, and have The camera pushes in with small amplitude at slow speed. Goebbels (S1) says: [English] What German films do you have? Francesca asks.. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Script Date 5/25/94\n\nSCREENPLAY EXCERPT:\nHe won't go into the interrogation room.\n\n KUJAN\n Someplace else, then.\n\n RAB IN\n Where?\n\n Kujan looks around Rabin's messy office.\n\n RABIN (CONT'D)\n No, no, no, no, no.\n\n KUJAN\n If it was a dope deal, where's the dope,\n if it was a hit, who called it in?\n\n RABIN\n And I am sure you have a host of wild +\n theories to answer these questions.\n\n KUJAN\n You know damn well what I think.\n\n YELLOW 06/11/94\n\n\n 24A.\n\n \t\t\t\t RABIN\n That's crazy, Dave and it doesn't matter.\n He has total immunity and his story\n checks out. He doesn't know what you want\n to know.\n\n \t\t\t\tKUJAN\n I don't think he does. Not exactly, but\n there's a lot more to his story. I want\n to know why twenty-seven men died on that\n pier for what looks to be ninety-one\n million dollars worth of dope that wasn't\n there. Above all, I want to be sure that\n Dean Keaton is dead.\n\n \t \t \t\tRAB ZN\n He's dead.\n\n \t \t \t\tKUJAN\n Two hours. Just until he makes bail.\n\n \t\t"}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: He won't go into the interrogation room. Someplace else, then.\nDIALOGUE — Kujan: Someplace else, then.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Script Date 5/25/94: He won't go into the interrogation room. Someplace else, then., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He won't go into the interrogation room. Someplace else, then. The camera pushes in with small amplitude at slow speed. Kujan (S1) says: [English] Someplace else, then. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Spector considers her for a beat, then looks at the others.\nSetting: derived from The Bodyguard\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. O.S.\nACTION: Spector considers her for a beat, then looks at the others. The secretary interrupts from the back of the office.\nDIALOGUE — Secretary: Mr. Schiller calling on line one... Spector ignores her, still holding his look at Devaney, Rachel and Frank.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Bodyguard: Spector considers her for a beat, then looks at the others. The secretary interrupts from the back of the office., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Spector considers her for a beat, then looks at the others. The secretary interrupts from the back of the office. The camera pushes in with small amplitude at slow speed. Secretary (S1) says: [English] Mr. Schiller calling on line one... Spector ignores her, still holding his look at Devaney, Rachel and Frank. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Morty shrugs and puts the money into the file cabinet. The\nSetting: derived from Bride Of Leprechaun\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. O.S.\nACTION: Morty shrugs and puts the money into the file cabinet. The safe, which has wheels, is rolled easily towards the door by\nDIALOGUE — Morty: We'll get a double hernia if we try and lift this thing in the hearse. See if you can find something to make a ramp. I'll go get the car.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bride Of Leprechaun: Morty shrugs and puts the money into the file cabinet. The safe, which has wheels, is rolled easily towards the door by, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Morty shrugs and puts the money into the file cabinet. The safe, which has wheels, is rolled easily towards the door by The camera pushes in with small amplitude at slow speed. Morty (S1) says: [English] We'll get a double hernia if we try and lift this thing in the hearse. See if you can find something to make a ramp. I'll go get the car. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Behind the host, the lights come up on the theatre proscenium\nSetting: derived from Asteroid City\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. THEATRE PROSCENIUM\nACTION: Behind the host, the lights come up on the theatre proscenium set -- turned around to face the audience, with painted backdrop\nDIALOGUE — Host: Actor, immigrant, former student of the great theatrical guru Saltzburg Keitel.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Asteroid City: Behind the host, the lights come up on the theatre proscenium set -- turned around to face the audience, with painted backdrop, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Behind the host, the lights come up on the theatre proscenium set -- turned around to face the audience, with painted backdrop The camera pushes in with small amplitude at slow speed. Host (S1) says: [English] Actor, immigrant, former student of the great theatrical guru Saltzburg Keitel. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The boys walk out of the theatre with glazed eyes and wide\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. MOVIE THEATRE - DAY\nACTION: The boys walk out of the theatre with glazed eyes and wide Dude, that movie was fucking sweet!\nDIALOGUE — Kyle: Dude, that movie was fucking sweet!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: The boys walk out of the theatre with glazed eyes and wide Dude, that movie was fucking sweet!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The boys walk out of the theatre with glazed eyes and wide Dude, that movie was fucking sweet! The camera pushes in with small amplitude at slow speed. Kyle (S1) says: [English] Dude, that movie was fucking sweet! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: He won't go into the interrogation room.\nSetting: derived from Script Date 5/25/94\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: He won't go into the interrogation room. Someplace else, then.\nDIALOGUE — Kujan: Someplace else, then.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Script Date 5/25/94: He won't go into the interrogation room. Someplace else, then., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He won't go into the interrogation room. Someplace else, then. The camera pushes in with small amplitude at slow speed. Kujan (S1) says: [English] Someplace else, then. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Give it to me. (She rips it open.) Come\nSetting: derived from The Sponge\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SHE RIPS IT OPEN.\nACTION: Give it to me. (She rips it open.) Come Kramer surrounded by Cedric, Bob, and the other walkers.\nDIALOGUE — George: It's too late.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Sponge: Give it to me. (She rips it open.) Come Kramer surrounded by Cedric, Bob, and the other walkers., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Give it to me. (She rips it open.) Come Kramer surrounded by Cedric, Bob, and the other walkers. The camera pushes in with small amplitude at slow speed. George (S1) says: [English] It's too late. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in AgentsOfSHIELD illustrating the trope 'MauveShirt'. Maria Hill and Coulson from The Avengers. Coulson's popularity allowed him to graduate to Gold Shirt. He stars in Agents of S.H.I.E.L.D., but with Joss Whedon, you never know who is eligible for death.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying MauveShirt.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for AgentsOfSHIELD, MauveShirt, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for MauveShirt. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: We didn't just have sex in the\nSetting: derived from Int: Bedroom. Early-Morning Hours\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: We didn't just have sex in the Everyone is silent. Then...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int: Bedroom. Early-Morning Hours: We didn't just have sex in the Everyone is silent. Then..., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We didn't just have sex in the Everyone is silent. Then... The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Beavis And Butt-Head Do America\"\n\nSCREENPLAY EXCERPT:\nEXT. THE ANDERSON'S DRIVEWAY - DUSK\n\n\tB&B walk up. There's a camper in the driveway.\n\n\t\t\t\t\tBUTT-HEAD\n\t\t\tWhoa, check it out Beavis. I didn't\n\t\t\tknow Anderson had a Camper.\n\n\t\t\t\t\tBEAVIS\n\t\t\tYeah, heh heh. Maybe it has a TV.\n\t\t\tHeh heh. TV.\n\n\tB&B walk up to the camper and start to open the door just as MARCY\n\tANDERSON opens it. (As usual, she doesn't recognize them.)\n\n\t\t\t\t\tMARCY\n\t\t\tOh, hello. Are you guys here to look\n\t\t\tat the refrigerator?\n\n\t\t\t\t\tBUTT-HEAD\n\t\t\tUh, no.\n\n\t\t\t\t\tBEAVIS\n\t\t\tWe're here to look at the TV. Heh heh.\n\n\t\t\t\t\tMARCY\n\t\t\tOh, I didn't realize it was broken.\n\t\t\tCome on in.\n\n\t\t\t\t\tBUTT-HEAD\n\t\t\tCool. Huh huh huh.\n\n\tB&B walk into the camper. Marcy stays outside."}, {"role": "assistant", "content": "## SCENE 1 — EXT. THE ANDERSON'S DRIVEWAY - DUSK\nACTION: B&B walk up. There's a camper in the driveway. Whoa, check it out Beavis. I didn't\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Beavis And Butt-Head Do America\": B&B walk up. There's a camper in the driveway. Whoa, check it out Beavis. I didn't, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. B&B walk up. There's a camper in the driveway. Whoa, check it out Beavis. I didn't The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nMust be my imagination.\n\n In CU she turns back toward the sink area when suddenly the\n TRACK explodes with a MUSICAL STINGER. It happens in a flash.\n A shape lunges from the toilet booth across from the first\n shower. A hatchet glints. MARCIE screams. The hatchet\n strikes. The flashlight clatters to the floor.\n\n CUT TO:\n\n\n 75 EXT. DINER - NIGHT\n\n The sound of a cash register ringing up a sale hangs over an\n establishing shot of the diner. Through the window, which are\n being pelted with rain, we can see a WAITRESS and some CUSTOMERS\n seated on stools at the counter.\n\n CUT TO:\n\n\n 76 INT. DINER - NIGHT\n\n The WAITRESS is different from the morning woman, slightly\n older, but still attractive. She says good night to her\n CUSTOMER, who finishes paying his bill and crosses down the\n counter. Seated there is STEVE CHRISTY, finishing his cup of\n coffee. The waitress, SANDY, picks up his empty pie plate and\n tosses it in a pan of other dirty dishes.\n\n SANDY\n Anything else you want?\n\n STEVE\n No, thanks. I'm fine. Sandy.\n\n SANDY\n You can't go back there tonight. Not in\n that s"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Must be my imagination. In CU she turns back toward the sink area when suddenly the\nDIALOGUE — Sandy: Anything else you want?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: Must be my imagination. In CU she turns back toward the sink area when suddenly the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Must be my imagination. In CU she turns back toward the sink area when suddenly the The camera pushes in with small amplitude at slow speed. Sandy (S1) says: [English] Anything else you want? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Puffy Shirt\n\nSCREENPLAY EXCERPT:\nJERRY\n (Shrugs) I - I don't get it.\n\n\n GEORGE\n Me neither!\n\n\n JERRY\n What is it?\n\n\n GEORGE\n I don't know.\n\n\n JERRY\n They're hands!\n\n\n GEORGE\n This woman just set me up for a job!\n\n\n\n JERRY"}, {"role": "assistant", "content": "## SCENE 1 — INT. SHRUGS\nACTION: (Shrugs) I - I don't get it. This woman just set me up for a job!\nDIALOGUE — George: Me neither!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Puffy Shirt: (Shrugs) I - I don't get it. This woman just set me up for a job!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. (Shrugs) I - I don't get it. This woman just set me up for a job! The camera pushes in with small amplitude at slow speed. George (S1) says: [English] Me neither! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A Secret Service guard is talking on radio/phone.\nSetting: derived from \"Beavis And Butt-Head Do America\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HALLWAY OUTSIDE OF OVAL OFFICE - DAY\nACTION: A Secret Service guard is talking on radio/phone. Evacuation?... Probably just another bomb\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Beavis And Butt-Head Do America\": A Secret Service guard is talking on radio/phone. Evacuation?... Probably just another bomb, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A Secret Service guard is talking on radio/phone. Evacuation?... Probably just another bomb The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Inglourious Basterds\n\nSCREENPLAY EXCERPT:\nWe will be cruel to the Germans,\n and through our cruelty, they will\n know who we are. They will find the\n evidence of our cruelty, in the\n disembowed, dismembered, and\n disfigured bodies of their brothers\n we leave behind us. And the German\n will not be able to help themselves\n from imagining the cruelty their\n brothers endured at our hands, and\n our boot heels, and the edge of our\n knives.\n And the Germans, will be sickened by us.\n And the Germans, will talk about us.\n And the Germans, will fear us.\n And when the Germans close their eyes\n at night, and their sub conscious\n tortures them for the evil they've done,\n it will be with thoughts of us,\n that it tortures them with.\n He stops pacing, and looks at everybody.\n\n LT. ALDO\n Sound good?\n They all say;\n\n ALL\n Yes, sir!\n\n LT.ALDO\n That's what I like to hear. But I\n got a word of warning to all would-be\n warriors. When you join my command,\n you take on debit. A debit you owe\n me, personally. Every man under my\n command, owes me, one hundred nazi scalps.\n And I want my scalps.\n And all y'all will git me, one hundred\n Nazi scalps, taken from the heads of\n one hundred dead Nazi's...\n .or you will die trying.\n\n CUT TO\n\n EXT - MOUNTAIN TOP CHALET- DAY\n A huge Chalet on a mis"}, {"role": "assistant", "content": "## SCENE 1 — INT. HITLERS PRIVATE LAIR\nACTION: We will be cruel to the Germans, and through our cruelty, they will\nDIALOGUE — Lt. Aldo: Sound good? They all say;\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Inglourious Basterds: We will be cruel to the Germans, and through our cruelty, they will, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We will be cruel to the Germans, and through our cruelty, they will The camera pushes in with small amplitude at slow speed. Lt. Aldo (S1) says: [English] Sound good? They all say; The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Matrix\n\nSCREENPLAY EXCERPT:\nINT. DOJO\n\n\tNeo's face is knotted, teeth clenched, as he hurls\n\thimself at Morpheus.\n\n\t\t\t\t\tMORPHEUS\n\t\t\tGood. Adaption. Improvisation.\n\t\t\tBut your weakness isn't your\n\t\t\ttechnique.\n\n\tMorpheus attacks him and it is like nothing we have seen.\n\tHis feet and fists are everywhere taking Neo apart. For\n\tevery blow Neo blocks, five more hit their marks until --\n\n\tNeo falls.\n\n\tPanting, on his hands and knees, blood spits fror, his\n\tmouth speckling the white floor of the Dojo.\n\n\t\t\t\t\tMORPHEUS\n\t\t\tHow did I beat you?\n\n\t\t\t\t\tNEO\n\t\t\tYou -- You're too fast.\n\n\t\t\t\t\tMORPHEUS\n\t\t\tDo you think my being faster,\n\t\t\tstronger has anything to do with\n\t\t\tmy muscles in this place?\n\n\tNeo is frustrated, still unable to catch his breath.\n\n\t\t\t\t\tMORPHEUS\n\t\t\tDo you believe that's air you are\n\t\t\tbreathing now?\n\n\tNeo squints at him.\n\n\t\t\t\t\tMORPHEUS\n\t\t\tIf you can free your mind, the\n\t\t\tbody will follow.\n\n\tNeo stands, nodding.\n\n\t\t\t\t\tMORPHEUS\n\t\t\tAgain.\n\n\tTheir fists fly with pneumatic speed."}, {"role": "assistant", "content": "## SCENE 1 — INT. DOJO\nACTION: Neo's face is knotted, teeth clenched, as he hurls Good. Adaption. Improvisation.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Matrix: Neo's face is knotted, teeth clenched, as he hurls Good. Adaption. Improvisation., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Neo's face is knotted, teeth clenched, as he hurls Good. Adaption. Improvisation. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The CATWOMAN's on a landing near the stairwell, just below the\nSetting: derived from Batman 2\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TOP-FLOOR LANDING - THAT MOMENT\nACTION: The CATWOMAN's on a landing near the stairwell, just below the dormer that leads to the roof. She pries open a pair of\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman 2: The CATWOMAN's on a landing near the stairwell, just below the dormer that leads to the roof. She pries open a pair of, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The CATWOMAN's on a landing near the stairwell, just below the dormer that leads to the roof. She pries open a pair of The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Sunset Boulevard\n\nSCREENPLAY EXCERPT:\nI'm not worried. Everything will\n be fine. The old team together.\n Nothing can stop us.\n\n She turns and walks out of the shot. De Mille\n stands for a second watching her, then turns to\n his assistant.\n\n DE MILLE\n Get Gordon Cole. Tell him to forget\n about her car. He can find another\n old car. I'll buy him five old cars,\n if necessary.\n\n 1ST ASSISTANT\n Yes, Mr. De Mille.\n\n They turn back into Stage 18.\n\n D-23 THE ISOTTA\n\n Gillis seated in the rear. Max is helping Norma\n in and putting the robe over her.\n\n GILLIS"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I'm not worried. Everything will be fine. The old team together.\nDIALOGUE — De Mille: Get Gordon Cole. Tell him to forget about her car. He can find another old car. I'll buy him five old cars, if necessary.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Sunset Boulevard: I'm not worried. Everything will be fine. The old team together., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm not worried. Everything will be fine. The old team together. The camera pushes in with small amplitude at slow speed. De Mille (S1) says: [English] Get Gordon Cole. Tell him to forget about her car. He can find another old car. I'll buy him five old cars, if necessary. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Manray screws out the bare light bulb which is the sole\nSetting: derived from Black Screen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TENEMENT BUILDING - NIGHT\nACTION: Manray screws out the bare light bulb which is the sole source of illumination and lays down onto his mattress on\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: Manray screws out the bare light bulb which is the sole source of illumination and lays down onto his mattress on, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Manray screws out the bare light bulb which is the sole source of illumination and lays down onto his mattress on The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Nearly half a century later, the elderly Lizardo regards with\nSetting: derived from Buckaroo Banzai\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LIZARDO'S ROOM - NIGHT\t104\nACTION: Nearly half a century later, the elderly Lizardo regards with smoldering evil the turn of the key in his own door and the appearance\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Buckaroo Banzai: Nearly half a century later, the elderly Lizardo regards with smoldering evil the turn of the key in his own door and the appearance, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Nearly half a century later, the elderly Lizardo regards with smoldering evil the turn of the key in his own door and the appearance The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Hunter\"\n\nSCREENPLAY EXCERPT:\n------------------------------------------------------------------------\n 31\n\n\n\n 53-A CONTINUED: 53-A\n\n Ramirez charges through the DUST and thick SMOKE to\n cover the fallen men; Schaefer following close behind,\n jerking a new CLIP from his ammo belt.\n\n Suddenly, another Guerrilla erupts from behind some\n ammunition crates, drawing down on Ramirez with a\n machine pistol.\n\n SCHAEFER"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: ------------------------------------------------------------------------ 53-A CONTINUED: 53-A\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hunter\": ------------------------------------------------------------------------ 53-A CONTINUED: 53-A, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ------------------------------------------------------------------------ 53-A CONTINUED: 53-A The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The commander normally has first\nSetting: derived from Alien\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The commander normally has first Maybe we could write it together.\nDIALOGUE — Melkonis: Maybe we could write it together.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Alien: The commander normally has first Maybe we could write it together., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The commander normally has first Maybe we could write it together. The camera pushes in with small amplitude at slow speed. Melkonis (S1) says: [English] Maybe we could write it together. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Hackers\n\nSCREENPLAY EXCERPT:\nEXT. NEW YORK CITY PUBLIC PHONE.\n\n Dade hacks into an electronic personal ads system and\n changes an ad.\n\n KATE\n Alright, he's in the personal ads.\n (reading Dade's ad)\n \"Disappointed white male, crossdresser, looking\n for discreet friend to bring dreams to\n reality. Leather, lace, and water sports.\n Transvestites welcome.\""}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEW YORK CITY PUBLIC PHONE.\nACTION: Dade hacks into an electronic personal ads system and Alright, he's in the personal ads.\nDIALOGUE — Kate: Alright, he's in the personal ads. \"Disappointed white male, crossdresser, looking for discreet friend to bring dreams to reality. Leather, lace, and water...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hackers: Dade hacks into an electronic personal ads system and Alright, he's in the personal ads., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Dade hacks into an electronic personal ads system and Alright, he's in the personal ads. The camera pushes in with small amplitude at slow speed. Kate (S1) says: [English] Alright, he's in the personal ads. \"Disappointed white male, crossdresser, looking for discreet friend to bring dreams to reality. Leather, lace, and water... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in InspectorGeorgeGently illustrating the trope 'ByTheBookCop'. Played with. Gently knows when to play along with rules and when to show his hand. Bacchus is a Rabid Cop in the making.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying ByTheBookCop.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for InspectorGeorgeGently, ByTheBookCop, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for ByTheBookCop. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Basic Instinct\n\nSCREENPLAY EXCERPT:\nINT. LT. WALKER'S OFFICE\n\n He and Gus sit there with Lt. Walker, Harrigan, Andrews and\n Captain Talcott.\n\n HARRIGAN\n Sixteen stab wounds to the chest\n and neck. No usable prints, no\n forcible entry, nothing missing.\n No prints on the ice pick, either --\n it's available at any Safeway.\n The scarf is Hermes, expensive --\n they sell about 20,000 a year\n worldwide.\n\n ANDREWS\n The powder was cocaine, high-\n quality, high-content. He inhaled\n it; there were minute quantities\n on his lips and penis. Mr. Boz\n leaves five million dollars, no\n insurance, no direct survivors.\n He liked his coke, he liked his\n girls, and he liked rock and roll.\n\n NICK\n He liked the mayor, too, right?\n\n Talcott gives him a look.\n\n GUS\n What about his girlfriend?\n\n TALCOTT\n Is she relevant here? I didn't\n know she was a suspect.\n\n LT. WALKER\n She's a suspect.\n\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. LT. WALKER'S OFFICE\nACTION: He and Gus sit there with Lt. Walker, Harrigan, Andrews and Sixteen stab wounds to the chest\nDIALOGUE — Harrigan: Sixteen stab wounds to the chest and neck. No usable prints, no forcible entry, nothing missing. No prints on the ice pick, either -- it's available at any...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Basic Instinct: He and Gus sit there with Lt. Walker, Harrigan, Andrews and Sixteen stab wounds to the chest, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He and Gus sit there with Lt. Walker, Harrigan, Andrews and Sixteen stab wounds to the chest The camera pushes in with small amplitude at slow speed. Harrigan (S1) says: [English] Sixteen stab wounds to the chest and neck. No usable prints, no forcible entry, nothing missing. No prints on the ice pick, either -- it's available at any... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Big Lebowski\n\nSCREENPLAY EXCERPT:\nNo. No Mr. Lebowski, that had not\n\t\toccurred to me.\n\n\t\t\t\tBRANDT\n\t\tThat had not occurred to us, Dude.\n\n\t\t\t\tDUDE\n\t\tWell, okay, you're not privy to all\n\t\tthe new shit, so uh, you know, but\n\t\tthat's what you pay me for. Speaking\n\t\tof which, would it be possible for\n\t\tme to get my twenty grand in cash?\n\t\tI gotta check this with my accountant\n\t\tof course, but my concern is that,\n\t\tyou know, it could bump me into a\n\t\thigher tax--\n\n\t\t\t\tLEBOWSKI\n\t\tBrandt, give him the envelope.\n\n\t\t\t\tDUDE\n\t\tWell, okay, if you've already made\n\t\tout the check. Brandt is handing\n\t\thim a letter-sized envelope which is\n\t\tdistended by something inside.\n\n\t\t\t\tBRANDT\n\t\tWe received it this morning.\n\nThe Dude, frowning, untucks its flap, takes out some cotton\nwadding and unrolls it.\n\n\t\t\t\tLEBOWSKI\n\t\tSince you have failed to achieve,\n\t\teven in the modest task that was\n\t\tyour charge, since you have stolen\n\t\tmy money, and since you have\n\t\tunrepentantly betrayed my trust.\n\nThe wadding, undone, reveals a smaller wad of gauze taped up\ninside. The Dude undoes the tape with his fingernails and\nstarts to unroll the inner package.\n\n\t\t\t\tLEBOWSKI\n\t\tI have no choice but to tell these\n\t\tbums that they should do whatever is\n\t\tnecessary to recover their money\n\t\tfrom you, Jeffrey Lebowski. And\n\t\twith Brandt as my witness, tell you\n\t\tthis: Any further harm visited upon\n\t\tBunny, shall be visited tenfold upon\n\t\tyour head.\n\nBetween thumb and forefinger the Dude holds up the contents\nof the package--a little toe, with emerald green nail polish.\n\n\t\t\t\tLEBOWSKI\n\t\t...By God sir. I will not abide\n\t\tanother toe.\n\nCOFFEE SHOP\n\nThe Du"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: No. No Mr. Lebowski, that had not That had not occurred to us, Dude.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Big Lebowski: No. No Mr. Lebowski, that had not That had not occurred to us, Dude., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. No. No Mr. Lebowski, that had not That had not occurred to us, Dude. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Crank\n\nSCREENPLAY EXCERPT:\nEXT. CHOCOLATE'S APARTMENT - EARLIER\n\n EVE looks up into CHEV'S eyes, absolutely vulnerable,\n absolutely STUNNING. Pale sunlight punches through the half\n rolled blinds.\n\n EVE\n Do you promise?\n\n CHEV meets her gaze.\n\n CHEV\n I promise.\n\n CUT TO:\n\n EXT STANDARD HOTEL, DOWNTOWN LOS ANGELES - DAY\n 131 131\n CHEV bangs through the revolving glass doors to the hotel.\n\n\n (CONTINUED)\n\n R\n\n\n\n\n 1\n\n\n\n\n I\n\n\n\n\n 90.\n CONTINUED:\n\n\n The CAMERA shifts to a ROUGH HANDHELD POV, similar to the\n shot that starts the film. The MUSIC is replaced by the\n SOUND of CHEV'S labored BREATHING and HEARTBEAT. The POV\n doubles, goes in and out of focus.\n\n n one continuous shot he approaches the STANDARD HOTEL and\n walks past the valets into the RETRO-MOD LOBBY.\n\n INT LOBBY, STANDARD HOTEL - CONTINUOUS\n 132 132\n The shot continues as CHEV moves through the LOBBY and makes\n"}, {"role": "assistant", "content": "## SCENE 1 — EXT. CHOCOLATE'S APARTMENT - EARLIER\nACTION: EVE looks up into CHEV'S eyes, absolutely vulnerable, absolutely STUNNING. Pale sunlight punches through the half\nDIALOGUE — Eve: Do you promise?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Crank: EVE looks up into CHEV'S eyes, absolutely vulnerable, absolutely STUNNING. Pale sunlight punches through the half, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. EVE looks up into CHEV'S eyes, absolutely vulnerable, absolutely STUNNING. Pale sunlight punches through the half The camera pushes in with small amplitude at slow speed. Eve (S1) says: [English] Do you promise? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Untitled\n\nSCREENPLAY EXCERPT:\nCould you do something about the labels? This\nis supposed to be a clandestine operation.\n\n\n. 5. INF. NIGHT. TOWN HALL\n\n\nA public forum show is in progress in the stark, sleek poured-concrete\ndepths of the Town Hall. A panel of guests faces a ragged audience\ndrawn by the topic of debate announced on a backlit screen on stage:\n\n\nTV_AND SOCIAL RESPONSIBILITY.\n\n\nSome of the audience might even have come just to see the show's\nmoderator, [Rena King, a local plastic-earnest TV _interviewer,| whose\npresence is noted on a backlit panel all her own. 7\n\nRena has three guests on her panel: Max Renn, dressed respectably for\nthe occasion; a lady psychiatrist and pop personality in her om right\n\n\nnamed Nicki Brand;) anda TV monitor which sits on a chair identical\nto those supporting the other guests. ` Love iT\n\n\nOn the monitor is the face of| Professor Brian Q'Blivion) pop media\n\n\nprophet. He looks like and is a cross between Marshall McLuhan and\nAndy Warhol. oo oe É\n\nThree remote TV cameras are shooting the show live for one of the\nnetworks. Whenever we catch a glimpse. of O'Blivion's camera on one\nof the mobile studio's monitors, we notice that he is always shot\nloosely, so that we see the TV set, and that he is always neatly\nsubtitledProf. Brian O'Blivion, as though the name explained the\nstrangeness of his closeup. .\n\n\nThe forum is already hitting its stride.\n\n\nRENA"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Could you do something about the labels? This is supposed to be a clandestine operation.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: Could you do something about the labels? This is supposed to be a clandestine operation., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Could you do something about the labels? This is supposed to be a clandestine operation. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Django Unchained\n\nSCREENPLAY EXCERPT:\nNow we got it. .So let's go back in there\n and busts these motherfucker's chops.\n\n INT - DINING ROOM (BIG HOUSE) - NIGHT\n The Dinner Table Guests listen to Lara Lee'Candie-Fitzwilly hold court\n melodramatically ... . Till ... . Calvin Candie enters the room from the\n kitchen door.\n Lara Lee, oblivious to her brothers change in demeanor, blurts out upon\n\n SEEING HIM;\n\n LARA LEE\n There you are! I was beginning to think\n you and that of crow ran off together.\n Schultz and Django feel the change in their host's demeanor, as he\n stares down the table at them.\n\n CALVIN CANDLE\n That'd be a hellva note, wouldn't it\n Lara Lee?"}, {"role": "assistant", "content": "## SCENE 1 — INT. BIG HOUSE\nACTION: Now we got it. .So let's go back in there and busts these motherfucker's chops.\nDIALOGUE — Lara Lee: There you are! I was beginning to think you and that of crow ran off together. Schultz and Django feel the change in their host's demeanor, as he stares down...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Django Unchained: Now we got it. .So let's go back in there and busts these motherfucker's chops., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Now we got it. .So let's go back in there and busts these motherfucker's chops. The camera pushes in with small amplitude at slow speed. Lara Lee (S1) says: [English] There you are! I was beginning to think you and that of crow ran off together. Schultz and Django feel the change in their host's demeanor, as he stares down... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Tombstone\n\nSCREENPLAY EXCERPT:\nUh, that woman.\n\n WYATT\n I miss Doc. I miss that ol' rip.\n\n VIRGIL\n I don't\n\n ALLIE\n Neither do I.\n\n WYATT\n He makes me laugh.\n\nINT - PRESCOTT SALOON - NIGHT\n\nA handsomely appointed saloon. At a corner table, the pot is\nso rich 2 players have folded leaving ED BAILEY, a big,\nsullen, tough-looking gambler, facing gaunt, elegant DOC\nHOLLIDAY. Full of southern refinement and languid, almost\nfeline grace. Doc has such unerring style and aplomb that he\nmakes his constant tubercular coughing sound as if he's merely\nclearing his throat. Bailey leans forward, seething with\nimpatience:\n\n BAILEY\n I said that's 500 to you,\n Holliday. In or out?\n\n DOC\n 500? Sly boots, must be a peach\n of a hand.\n\nKATE HORONY, Doc's voluptuous Hungarian consort enters,\nrefilling his engraved silver stirrup cup. She has a faint\naccent:\n\n KATE\n Here, Doc.\n\n DOC\n Bless you, darling"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I miss Doc. I miss that ol' rip. A handsomely appointed saloon. At a corner table, the pot is\nDIALOGUE — Wyatt: I miss Doc. I miss that ol' rip.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Tombstone: I miss Doc. I miss that ol' rip. A handsomely appointed saloon. At a corner table, the pot is, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I miss Doc. I miss that ol' rip. A handsomely appointed saloon. At a corner table, the pot is The camera pushes in with small amplitude at slow speed. Wyatt (S1) says: [English] I miss Doc. I miss that ol' rip. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Tony and Steve round the corner, TONY IN A LAB COAT and STEVE\nSetting: derived from Avengers: Endgame\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CAMP LEHIGH - DAY (1970)\nACTION: Tony and Steve round the corner, TONY IN A LAB COAT and STEVE IN GREEN KHAKIS, LOW-PULLED CAP, AND MIRRORED SUNGLASSES.\nDIALOGUE — Steve: The idea of me was.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Avengers: Endgame: Tony and Steve round the corner, TONY IN A LAB COAT and STEVE IN GREEN KHAKIS, LOW-PULLED CAP, AND MIRRORED SUNGLASSES., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Tony and Steve round the corner, TONY IN A LAB COAT and STEVE IN GREEN KHAKIS, LOW-PULLED CAP, AND MIRRORED SUNGLASSES. The camera pushes in with small amplitude at slow speed. Steve (S1) says: [English] The idea of me was. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: COMMISSIONER GORDON and another cop, LT. EDDIE BULLOCK, are in\nSetting: derived from Batman 2\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. POLICE OBSERVATION ROOM - NIGHT\nACTION: COMMISSIONER GORDON and another cop, LT. EDDIE BULLOCK, are in darkened antechamber adjacent to an interrogation room.\nDIALOGUE — Bullock: It wasn't about the drugs. Whoever it was left thirty kilos sitting on the docks.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman 2: COMMISSIONER GORDON and another cop, LT. EDDIE BULLOCK, are in darkened antechamber adjacent to an interrogation room., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. COMMISSIONER GORDON and another cop, LT. EDDIE BULLOCK, are in darkened antechamber adjacent to an interrogation room. The camera pushes in with small amplitude at slow speed. Bullock (S1) says: [English] It wasn't about the drugs. Whoever it was left thirty kilos sitting on the docks. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Marine Biologist\n\nSCREENPLAY EXCERPT:\nGEORGE\n Did you see that whale thing on TV last\n night?\n\n\n JERRY\n No.\n\n\n GEORGE\n I'm such a Huge whale fan. These marine\n biologists were showing how\n\n\n they communicate with each other with these squeaks and squeals,\n what\n\n\n a fish!\n\n\n JERRY\n It's a mammal.\n\n\n GEORGE\n Whatever. (George looks to the table)\n Hey new tape recorder?\n\n\n JERRY\n Yeah, got it from the bank."}, {"role": "assistant", "content": "## SCENE 1 — INT. GEORGE LOOKS TO THE TABLE\nACTION: Did you see that whale thing on TV last I'm such a Huge whale fan. These marine\nDIALOGUE — Jerry: No.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Marine Biologist: Did you see that whale thing on TV last I'm such a Huge whale fan. These marine, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Did you see that whale thing on TV last I'm such a Huge whale fan. These marine The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] No. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Do you hear the people sing\nSetting: derived from Les Miserables\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Do you hear the people sing Singing the song of angry men\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Les Miserables: Do you hear the people sing Singing the song of angry men, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Do you hear the people sing Singing the song of angry men The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Sam Lowry. Hello. This sounds insane,\nSetting: derived from 1 Ext. Cityscape Sunset 1\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Sam Lowry. Hello. This sounds insane, I know, but I've been dreaming about\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from 1 Ext. Cityscape Sunset 1: Sam Lowry. Hello. This sounds insane, I know, but I've been dreaming about, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sam Lowry. Hello. This sounds insane, I know, but I've been dreaming about The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: BATEMAN is trying to hail a cab. Owen's body is at his\nSetting: derived from American Psycho\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. STREET - NIGHT\nACTION: BATEMAN is trying to hail a cab. Owen's body is at his feet. Luis Carruthers and a Japanese girl walk up to him.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American Psycho: BATEMAN is trying to hail a cab. Owen's body is at his feet. Luis Carruthers and a Japanese girl walk up to him., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. BATEMAN is trying to hail a cab. Owen's body is at his feet. Luis Carruthers and a Japanese girl walk up to him. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The hackers are now setting up laptops at a bank of pay\nSetting: derived from Hackers\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. GRAND CENTRAL PUBLIC PHONES.\nACTION: The hackers are now setting up laptops at a bank of pay phones. Dade is wearing a \"Pirate Eye\" eyepiece.\nDIALOGUE — Kate: Now listen up, use your best viruses to buy us time, we have to get into Plague's file and copy the worm.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hackers: The hackers are now setting up laptops at a bank of pay phones. Dade is wearing a \"Pirate Eye\" eyepiece., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The hackers are now setting up laptops at a bank of pay phones. Dade is wearing a \"Pirate Eye\" eyepiece. The camera pushes in with small amplitude at slow speed. Kate (S1) says: [English] Now listen up, use your best viruses to buy us time, we have to get into Plague's file and copy the worm. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Black Screen\n\nSCREENPLAY EXCERPT:\nEXT. EDGE OF LOTHLORIEN--DUSK\n\nWIDE ON: The fellowship run across a forest floor strewn with\nyellow flowers..above is a roof of golden leaves, held up by\nsilver pillars...the trunks of huge, grey trees. Gimli looks\nnervously around...\n\n GIMLI\n Stay close, young hobbits..they say a\n Sorceress lives in these woods. An elf-\n witch of terrible power. All who look\n upon her fall under her spell...\n\nFrodo hesitates...a STRANGE VOICE whispers in his head...\n\n GALADRIEL (V.O.)\n Frodo...\n\n GIMLI\n And are never seen again!\n\n GALADRIEL (V.O.)\n ...your coming to us is as the footsteps\n of doom. You bring great evil here,\n Ringbearer.\n\n SAM\n Mr. Frodo?\n\n GIMLI\n Well, here's one dwarf she won't ensnare\n so easily. I have the eyes of a hawk and\n the ears of a fox!\n\n\n\n\n (CONTINUED)\n\n 99.\nCONTINUED:\n\n\nThe FELLOWSHIP are suddenly surrounded by ARMED ELVES.\nDeadly arrows aimed at their heads. HALDIR, the Elvish\ncaptain steps forward... he looks at Gimli with disdain.\n\n HALDIR\n The dwarf breathes so loud we could have\n shot him in the dark.\n\n ARAGORN\n (Elvish: with subtitles)\n Haldir of Lorien, we come here for your\n help. We need your protection.\n\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. EDGE OF LOTHLORIEN--DUSK\nACTION: WIDE ON: The fellowship run across a forest floor strewn with yellow flowers..above is a roof of golden leaves, held up by\nDIALOGUE — Gimli: Stay close, young hobbits..they say a Sorceress lives in these woods. An elf- witch of terrible power. All who look upon her fall under her spell...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: WIDE ON: The fellowship run across a forest floor strewn with yellow flowers..above is a roof of golden leaves, held up by, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. WIDE ON: The fellowship run across a forest floor strewn with yellow flowers..above is a roof of golden leaves, held up by The camera pushes in with small amplitude at slow speed. Gimli (S1) says: [English] Stay close, young hobbits..they say a Sorceress lives in these woods. An elf- witch of terrible power. All who look upon her fall under her spell... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: CAMERA PULLS back to reveal Dorothy standing by the table as\nSetting: derived from Fade Out:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TOWER ROOM -- CLOSE ON THE HOUR GLASS ON THE TABLE --\nACTION: CAMERA PULLS back to reveal Dorothy standing by the table as she watches the sand run through the glass -- she looks about\nDIALOGUE — Lion: Oh -- Oh -- Oh -- I....\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade Out:: CAMERA PULLS back to reveal Dorothy standing by the table as she watches the sand run through the glass -- she looks about, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CAMERA PULLS back to reveal Dorothy standing by the table as she watches the sand run through the glass -- she looks about The camera pushes in with small amplitude at slow speed. Lion (S1) says: [English] Oh -- Oh -- Oh -- I.... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Inside are several marijuana plants. Sun lamps beam down on\nSetting: derived from \"The Thing\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. STORAGE ROOM\nACTION: Inside are several marijuana plants. Sun lamps beam down on them. Childs inspects them with a wide grin.\nDIALOGUE — Childs: How my brothers and sister doing today? Doin' fine.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"The Thing\": Inside are several marijuana plants. Sun lamps beam down on them. Childs inspects them with a wide grin., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Inside are several marijuana plants. Sun lamps beam down on them. Childs inspects them with a wide grin. The camera pushes in with small amplitude at slow speed. Childs (S1) says: [English] How my brothers and sister doing today? Doin' fine. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: HARRY, carrying the GIRL, comes out of the police station, as\nSetting: derived from The Heavy Metal Movie\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. POLICE STATION\nACTION: HARRY, carrying the GIRL, comes out of the police station, as a patrol car screeches to a halt in front of him. Two cops\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Heavy Metal Movie: HARRY, carrying the GIRL, comes out of the police station, as a patrol car screeches to a halt in front of him. Two cops, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. HARRY, carrying the GIRL, comes out of the police station, as a patrol car screeches to a halt in front of him. Two cops The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The bullpen is dark, empty save for Regina and Erik sitting\nSetting: derived from Malignant\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PRECINCT - BULLPEN - NIGHT\nACTION: The bullpen is dark, empty save for Regina and Erik sitting at their desks, pouring over the files they retrieved from\nDIALOGUE — Regina: That's it. I'm out. I can't keep any of this shit straight. My brain feels like it's about to push through my eyeballs.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Malignant: The bullpen is dark, empty save for Regina and Erik sitting at their desks, pouring over the files they retrieved from, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The bullpen is dark, empty save for Regina and Erik sitting at their desks, pouring over the files they retrieved from The camera pushes in with small amplitude at slow speed. Regina (S1) says: [English] That's it. I'm out. I can't keep any of this shit straight. My brain feels like it's about to push through my eyeballs. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Untitled\n\nSCREENPLAY EXCERPT:\nNo...no. You don't understand. A source...\na source has told me that Videodrome, _ i\nis for real. That it's smu > Nic\nBrand was murdered on Videodrome, and\n\n- we saw it. They wanted us to see it.\nThey were telling us to leave them alone.\n\n\ntapes back - the tapes that prove they.\nexist.\n. HARLAN -"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: No...no. You don't understand. A source... a source has told me that Videodrome, _ i\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: No...no. You don't understand. A source... a source has told me that Videodrome, _ i, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. No...no. You don't understand. A source... a source has told me that Videodrome, _ i The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Galactica1980 illustrating the trope 'BaseballEpisode'. * An episode of Galactica 1980 has a baseball game central to its plot.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying BaseballEpisode.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Galactica1980, BaseballEpisode, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for BaseballEpisode. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Anderson looks in the door Of Witwer's empty office. He\nSetting: derived from \"Minority Report\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. OUTSIDE WITWER'S OFFICE -- MOMENTS LATER\nACTION: Anderson looks in the door Of Witwer's empty office. He Witwer's booming voice sounds from behind him, startling\nDIALOGUE — Witwer: Breaking and entering. That'll get you five to ten, hard.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Minority Report\": Anderson looks in the door Of Witwer's empty office. He Witwer's booming voice sounds from behind him, startling, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Anderson looks in the door Of Witwer's empty office. He Witwer's booming voice sounds from behind him, startling The camera pushes in with small amplitude at slow speed. Witwer (S1) says: [English] Breaking and entering. That'll get you five to ten, hard. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Austin and Vanessa join the high-rollers table. Number Two\nSetting: derived from Austin Powers: International Man Of Mystery\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CASINO\nACTION: Austin and Vanessa join the high-rollers table. Number Two is there, complete with eyepatch. On one side of him is a\nDIALOGUE — Austin: Do you mind if I join you?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Austin Powers: International Man Of Mystery: Austin and Vanessa join the high-rollers table. Number Two is there, complete with eyepatch. On one side of him is a, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Austin and Vanessa join the high-rollers table. Number Two is there, complete with eyepatch. On one side of him is a The camera pushes in with small amplitude at slow speed. Austin (S1) says: [English] Do you mind if I join you? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I'm gonna make you believe in it,\nSetting: derived from Basic Instinct\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. THE DETECTIVE BUREAU - DAY\nACTION: I'm gonna make you believe in it,\nDIALOGUE — Gus: I'm gonna make you believe in it, friend.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Basic Instinct: I'm gonna make you believe in it,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm gonna make you believe in it, The camera pushes in with small amplitude at slow speed. Gus (S1) says: [English] I'm gonna make you believe in it, friend. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nINT. TUNNEL/CUBBYHOLE\n\n\tOn her hands and knees, she enters the TUNNEL, not waiting\n\tfor him to respond. She knows what he'll do... Catherine and\n\tNovak crawl through the dark, grimy claustrophobic space,\n\tplaying with his mirror locket.\n\n\t\t\t\tCATHERINE\n\t\tCarl...\n\n\tHe so desperately wants her affection, but fear prevents him\n\tfrom reaching out to her.\n\n\t\t\t\tCATHERINE (CONT'D)\n\t\tThis is Peter.\n\n\tScared, he shakes his head \"no.\"\n\n\t\t\t\tCATHERINE (CONT'D)\n\t\tDon't be scared. You're my friend and I\n\t\twould never hurt you.\n\n\tShe extends her hand. Oh-so-hesitantly he takes it.\n\n\t\t\t\tCATHERINE (CONT'D)\n\t\tWill you take us to her? To Julia?\n\n\tShyly, hesitantly, he nods. She smiles gratefully and\n\taffectionately touches the boy's face. Clasping her hand,\n\tYoung Stargher leads Catherine and Novak into utter darkness."}, {"role": "assistant", "content": "## SCENE 1 — INT. TUNNEL/CUBBYHOLE\nACTION: On her hands and knees, she enters the TUNNEL, not waiting for him to respond. She knows what he'll do... Catherine and\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: On her hands and knees, she enters the TUNNEL, not waiting for him to respond. She knows what he'll do... Catherine and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. On her hands and knees, she enters the TUNNEL, not waiting for him to respond. She knows what he'll do... Catherine and The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: American Psycho\n\nSCREENPLAY EXCERPT:\nINT. TAXI - EVENING\n\nEVELYN WILLIAMS, Patrick Bateman's\nfianc�e, is making notes with a gold Cross pen and sipping\na bottle of mineral water. Evelyn is blonde, classically\nbeautiful, expensively educated, and utterly pleased with\nherself. She usually addresses Patrick as if he were a\nsmall child.\n\nEVELYN\nI'd want a zydeco band, Patrick. That's what I'd\nwant, a zydeco band. Or mariachi. Or reggae. Something\nethnic to shock Daddy Oh, I can't decide...And lots\nof chocolate truffles. Godiva. And oysters on the halfshell.\n\nCLOSE-UP on Bateman, who is wearing a Walkman and staring\nout the window.\n\nBATEMAN (V.O.)\nI'm trying to listen to the new George\nMichael tape but Evelyn-my supposed fianc�e-keeps buzzing\nin my ear.\n\nEvelyn continues to make notes.\n\nEVELYN\nMarzipan. Pink tents. Hundreds, thousands of roses.\nPhotographers. Annie Leibovitz. We'll get Annie Leibovitz.\nAnd we'll hire someone to videotape. Patrick, we should do it.\n\nBATEMAN\n(Removing his Walkman)\nDo...what.\n\nEVELYN\nGet married. Have a wedding.\n\nBATEMAN\nEvelyn?\n\nEVELYN\nYes, darling?\n\nBATEMAN\nIs your Evian spiked?\n\nEVELYN\nWe should do it.\n\nBATEMAN\nNo-I can't take the time off work.\n\nEVELYN\nYour father practically owns the company. You can do\nanything you like, silly.\n\nBATEMAN\nI don't want to talk about it.\n\nEVELYN\nWell, you hate that job anyway. Why don't you just\nquit? You don't have to work.\n\nBATEMAN\nBecause I...want...to...fit...in.\n\nThe taxi bumps to a halt."}, {"role": "assistant", "content": "## SCENE 1 — INT. TAXI - EVENING\nACTION: EVELYN WILLIAMS, Patrick Bateman's fianc�e, is making notes with a gold Cross pen and sipping\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American Psycho: EVELYN WILLIAMS, Patrick Bateman's fianc�e, is making notes with a gold Cross pen and sipping, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. EVELYN WILLIAMS, Patrick Bateman's fianc�e, is making notes with a gold Cross pen and sipping The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: These gentlemen are with the New York\nSetting: derived from Script Date 5/25/94\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: These gentlemen are with the New York police department. You look great,\nDIALOGUE — Renault: Is there a problem, Mr. Keaton?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Script Date 5/25/94: These gentlemen are with the New York police department. You look great,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. These gentlemen are with the New York police department. You look great, The camera pushes in with small amplitude at slow speed. Renault (S1) says: [English] Is there a problem, Mr. Keaton? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-\nSetting: derived from L A B Y R I N T H\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- A WHITE BIRD soaring. The sky is a glorious explosion of blue and\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from L A B Y R I N T H: -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- A WHITE BIRD soaring. The sky is a glorious explosion of blue and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- A WHITE BIRD soaring. The sky is a glorious explosion of blue and The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: We're gonna go with it.\nSetting: derived from Joker\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: We're gonna go with it. Gene rubs his temples, he doesn't like this, but Murray is\nDIALOGUE — Joker: Thank you Murray.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Joker: We're gonna go with it. Gene rubs his temples, he doesn't like this, but Murray is, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We're gonna go with it. Gene rubs his temples, he doesn't like this, but Murray is The camera pushes in with small amplitude at slow speed. Joker (S1) says: [English] Thank you Murray. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fantastic Mr. Fox\n\nSCREENPLAY EXCERPT:\nINT. ATTIC. DAY\n\n The top floor of Bean Annex. The room is filled with boxes of\n Christmas ornaments, old sports equipment, two stained\n mattresses, and a broken birdcage. Cobwebs hang from the\n rafters below the sloped roof.\n Kristofferson stands with his hands in his pockets looking\n out between the slats from inside a padlocked apple crate on\n top of a gun locker in the corner. He clears his throat. He\n calls out politely:\n\n KRISTOFFERSON\n Could I have a cup of water, please?\n Kristofferson waits for a reply, but no one answers. He\n whistles to himself for a minute. He clears his throat again.\n He calls out:\n\n KRISTOFFERSON\n Excuse me! Excuse me?\n\n\n\n\n\n\n\n\n\n 71.\n Silence."}, {"role": "assistant", "content": "## SCENE 1 — INT. ATTIC. DAY\nACTION: The top floor of Bean Annex. The room is filled with boxes of Christmas ornaments, old sports equipment, two stained\nDIALOGUE — Kristofferson: Could I have a cup of water, please? Kristofferson waits for a reply, but no one answers. He whistles to himself for a minute. He clears his throat again. He...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fantastic Mr. Fox: The top floor of Bean Annex. The room is filled with boxes of Christmas ornaments, old sports equipment, two stained, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The top floor of Bean Annex. The room is filled with boxes of Christmas ornaments, old sports equipment, two stained The camera pushes in with small amplitude at slow speed. Kristofferson (S1) says: [English] Could I have a cup of water, please? Kristofferson waits for a reply, but no one answers. He whistles to himself for a minute. He clears his throat again. He... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BabylonFive illustrating the trope 'ThePuppetCutsHisStrings'. A Downplayed Example on Babylon 5. Captain Sheridan was chosen by the Clark administration on the basis of his loyalty to Earth (including a highly public incident during the Earth-Minbari war where he destroyed one of their war cruisers). Unfortunately for them, after he took command, he wasted no time in establishing that he was loyal to Earth not necessarily the man leading it.\n A very straigh...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying ThePuppetCutsHisStrings.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BabylonFive, ThePuppetCutsHisStrings, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for ThePuppetCutsHisStrings. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in House illustrating the trope 'JitterCam'. House, from about the third season on.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying JitterCam.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for House, JitterCam, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for JitterCam. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in House illustrating the trope 'LonelyRichKid'. Dr. Robert Chase from House.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying LonelyRichKid.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for House, LonelyRichKid, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for LonelyRichKid. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Air Poland, you are cleared for takeoff.\nSetting: derived from Airplane\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. TOWER - NIGHT\nACTION: Air Poland, you are cleared for takeoff.\nDIALOGUE — Ground Control: Air Poland, you are cleared for takeoff.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Airplane: Air Poland, you are cleared for takeoff., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Air Poland, you are cleared for takeoff. The camera pushes in with small amplitude at slow speed. Ground Control (S1) says: [English] Air Poland, you are cleared for takeoff. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Foot2Rue illustrating the trope 'YoungerThanTheyLook'. Shark is said to be 15 years old as it was hois birthday in Season 2.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying YoungerThanTheyLook.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Foot2Rue, YoungerThanTheyLook, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for YoungerThanTheyLook. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: He Phone Message\n\nSCREENPLAY EXCERPT:\nDONNA\n So, er, thanks for dinner. It was great.\n\n\n\n GEORGE\n Yeah. We should do this again.\n\n\n DONNA\n Would you like to come upstairs for\n some coffee?\n\n\n GEORGE\n Oh, no, thanks. I can't drink coffee\n late at night, it keeps me up.\n\n\n DONNA\n (Looks disappointed) So, um, OK.\n\n\n GEORGE\n OK.\n\n\n DONNA\n Goodnight.\n\n\n GEORGE\n Yeah, take it easy.\n\n\n (Donna leaves car. George realizes what he has done and bashes\n his\n\n\n forehead in disgust)\n\n\n\n\n\n DONNA\n Thanks again for the movie.\n\n\n JERRY\n You're welcome.\n\n\n DONNA\n I'd invite you up, but the place is\n being painted.\n\n\n JERRY\n Oh, that's OK.\n\n\n DONNA\n Unless you want to go to your place.\n\n\n\n JERRY\n OK, but there's no cake or anything,\n if that's what you're looking\n\n\n for.\n\n\n\n\n\n GEORGE\n Take it easy, huh, take it easy.\n\n\n\n\n\n JERRY\n I think if one's going to kill oneself,\n the l"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOOKS DISAPPOINTED\nACTION: So, er, thanks for dinner. It was great. Yeah. We should do this again.\nDIALOGUE — George: Yeah. We should do this again.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from He Phone Message: So, er, thanks for dinner. It was great. Yeah. We should do this again., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. So, er, thanks for dinner. It was great. Yeah. We should do this again. The camera pushes in with small amplitude at slow speed. George (S1) says: [English] Yeah. We should do this again. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Event Horizon\"\n\nSCREENPLAY EXCERPT:\nINT. EVENT HORIZON - SECOND CONTAINMENT\n\n Weir enters the Second Containment. Freezes as he sees...\n\n Peters body lying twisted and broken before the Core.\n\n WEIR\n Oh no. Peters...?\n\n He rushes to Peters. Reaches out to touch her but pulls his\n hand back. Her eyes are black, eight-ball hemorrhage darkening\n the irises. She is dead.\n\n WEIR\n Why did you do that? You didn't have\n to do that...\n\n CLAIRE (O.S.)\n Billy.\n\n Weir looks up from Peters' corpse.\n\n CLAIRE stands before the Core. She is naked. Her skin is\n pale and beautiful and cold and wet. Her hair hangs in her\n face, covering her milk-white eyes..."}, {"role": "assistant", "content": "## SCENE 1 — INT. EVENT HORIZON - SECOND CONTAINMENT\nACTION: Weir enters the Second Containment. Freezes as he sees... Peters body lying twisted and broken before the Core.\nDIALOGUE — Weir: Oh no. Peters...?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Event Horizon\": Weir enters the Second Containment. Freezes as he sees... Peters body lying twisted and broken before the Core., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Weir enters the Second Containment. Freezes as he sees... Peters body lying twisted and broken before the Core. The camera pushes in with small amplitude at slow speed. Weir (S1) says: [English] Oh no. Peters...? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: HENRY Your mother and I have talked things over and we've made some decisions . . . I am resigning from the university and will be taking up a position back in England.\nSetting: derived from Heavenly Creatures\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ILAM/JULIET'S BEDROOM - NIGHTJULIET IS PROPPED UP IN BED, LOOKING SICK AND PALE. HILDA IS SITTING BY THE SIDE OF THE BED. HENRY STANDS AT THE FOOT OF THE BED, HANDS BEHIND HIS BACK.\nACTION: HENRY Your mother and I have talked things over and we've made some decisions . . . I am resigning from the university and will be taking up a position back in England. Juliet looks horrified.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Heavenly Creatures: HENRY Your mother and I have talked things over and we've made some decisions . . . I am resigning from the university and will be taking up a position back in England. Juliet looks horrified., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. HENRY Your mother and I have talked things over and we've made some decisions . . . I am resigning from the university and will be taking up a position back in England. Juliet looks horrified. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: American Psycho\n\nSCREENPLAY EXCERPT:\nEXT. ALLEY - NIGHT\n\nA COP rushes toward him, seemingly from\nout of nowhere, and tackles him, trying to get the gun away\nfrom him.\n\nBateman manages to shoot the cop in the face while both of\nthem have their hands on the gun, then shoots him again. He\nreloads the gun. The sound of more COP CARS arriving.\n\nHe runs out of the alley."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ALLEY - NIGHT\nACTION: A COP rushes toward him, seemingly from out of nowhere, and tackles him, trying to get the gun away\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American Psycho: A COP rushes toward him, seemingly from out of nowhere, and tackles him, trying to get the gun away, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A COP rushes toward him, seemingly from out of nowhere, and tackles him, trying to get the gun away The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fear & Loathing In Las Vegas\n\nSCREENPLAY EXCERPT:\nINT. HOTEL FLAMINGO SUITE - BATHROOM - NEXT MORNING\n\nGONZO throws up in the toilet bowl.\n\nIn the background, DUKE opens curtains. Daylight blinds him.\n\n DUKE\n Come on, we're going to be late.\n\nGONZO looks up at his sick reflection -- wipes his mouth\nwith a towel.\n\n GONZO\n This goddamn mescaline. Why the\n fuck can't they make it a little\n less pure? Maybe mix it up with\n Rolaids or something.\n\n 72."}, {"role": "assistant", "content": "## SCENE 1 — INT. HOTEL FLAMINGO SUITE - BATHROOM - NEXT MORNING\nACTION: GONZO throws up in the toilet bowl. In the background, DUKE opens curtains. Daylight blinds him.\nDIALOGUE — Duke: Come on, we're going to be late.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fear & Loathing In Las Vegas: GONZO throws up in the toilet bowl. In the background, DUKE opens curtains. Daylight blinds him., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GONZO throws up in the toilet bowl. In the background, DUKE opens curtains. Daylight blinds him. The camera pushes in with small amplitude at slow speed. Duke (S1) says: [English] Come on, we're going to be late. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: This is gonna hurt. But every pok�mon is fighting each other,\nSetting: derived from Untitled\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: This is gonna hurt. But every pok�mon is fighting each other, What? I'm more unfair? It's because I'm not fighting with you?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: This is gonna hurt. But every pok�mon is fighting each other, What? I'm more unfair? It's because I'm not fighting with you?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. This is gonna hurt. But every pok�mon is fighting each other, What? I'm more unfair? It's because I'm not fighting with you? The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Boy she's sexy isn't she?\nSetting: derived from The Suicide\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Boy she's sexy isn't she? (Jerry leaves garbage bag by Kramer's apartment and knocks on\nDIALOGUE — Jerry: Do you believe that guy?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Suicide: Boy she's sexy isn't she? (Jerry leaves garbage bag by Kramer's apartment and knocks on, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Boy she's sexy isn't she? (Jerry leaves garbage bag by Kramer's apartment and knocks on The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] Do you believe that guy? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: F I G H T C L U B\n\nSCREENPLAY EXCERPT:\nINT. KITCHEN - NIGHT\n\n Jack enters. Space Monkeys render fat and make soap. They\n pinch HERBS, adding them to the mix. They add VODKA. Off\n to the side, a couple Monkeys stir a vat of RICE. On the\n wall is a big bulletin board with HUNDREDS of DRIVER's\n LICENSES; a sign above it: \"HUMAN SACRIFICES.\"\n\n FRECKLED SPACE MONKEY\n \"You are not a beautiful and unique\n snowflake. You are the same decaying\n organic matter as everything else.\n We are all part of the same compost\n heap.\"\n\n JACK (V.O.)\n Planet Tyler.\n\n Jack dips a spoon into the rice, chomps on it irritatingly.\n\n FRECKLED SPACE MONKEY\n \"We are the all-singing, all-dancing\n crap of the world.\"\n\n Jack picks up a BOTTLE of VODKA.\n\n JACK (V.O.)\n I had to hug the walls, trapped\n inside this clockwork of Space\n Monkeys, cooking and working and\n sleeping in teams."}, {"role": "assistant", "content": "## SCENE 1 — INT. KITCHEN - NIGHT\nACTION: Jack enters. Space Monkeys render fat and make soap. They pinch HERBS, adding them to the mix. They add VODKA. Off\nDIALOGUE — Freckled Space Monkey: \"You are not a beautiful and unique snowflake. You are the same decaying organic matter as everything else. We are all part of the same compost heap.\"\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from F I G H T C L U B: Jack enters. Space Monkeys render fat and make soap. They pinch HERBS, adding them to the mix. They add VODKA. Off, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack enters. Space Monkeys render fat and make soap. They pinch HERBS, adding them to the mix. They add VODKA. Off The camera pushes in with small amplitude at slow speed. Freckled Space Monkey (S1) says: [English] \"You are not a beautiful and unique snowflake. You are the same decaying organic matter as everything else. We are all part of the same compost heap.\" The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Late afternoon. They're driving on the interstate. Anthony's\nSetting: derived from Bottle Rocket\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CAR. DAY\nACTION: Late afternoon. They're driving on the interstate. Anthony's got a copy of Runner's World magazine.\nDIALOGUE — Anthony: Here's mine right here. See?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bottle Rocket: Late afternoon. They're driving on the interstate. Anthony's got a copy of Runner's World magazine., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Late afternoon. They're driving on the interstate. Anthony's got a copy of Runner's World magazine. The camera pushes in with small amplitude at slow speed. Anthony (S1) says: [English] Here's mine right here. See? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Ninja Assassin\n\nSCREENPLAY EXCERPT:\n-- because I don't want anything\n to happen to you.\n\n He glances up at her, and it's clear: he has affection\n for her."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: -- because I don't want anything He glances up at her, and it's clear: he has affection\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ninja Assassin: -- because I don't want anything He glances up at her, and it's clear: he has affection, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. -- because I don't want anything He glances up at her, and it's clear: he has affection The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Bodyguard\n\nSCREENPLAY EXCERPT:\nYeah, Frank.\n FRANK\n Tony, I've got a feeling this is\n the night. I think he's gonna go\n for her in front of all the\n cameras. The kind of thing only a\n lunatic would do. That's how he\n wants it to look. Except he isn't\n a lunatic. He's very clever. I\n need you to help me.\n TONY"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Tony, I've got a feeling this is the night. I think he's gonna go\nDIALOGUE — Frank: Tony, I've got a feeling this is the night. I think he's gonna go for her in front of all the cameras. The kind of thing only a lunatic would do. That's how he...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Bodyguard: Tony, I've got a feeling this is the night. I think he's gonna go, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Tony, I've got a feeling this is the night. I think he's gonna go The camera pushes in with small amplitude at slow speed. Frank (S1) says: [English] Tony, I've got a feeling this is the night. I think he's gonna go for her in front of all the cameras. The kind of thing only a lunatic would do. That's how he... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \" T E R M I N A T O R \"\n\nSCREENPLAY EXCERPT:\nYou know your weapons, buddy.\n\n Terminator examines each in turn, working the actions with\n curt, precise movements.\n\n CLERK"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You know your weapons, buddy. Terminator examines each in turn, working the actions with\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \" T E R M I N A T O R \": You know your weapons, buddy. Terminator examines each in turn, working the actions with, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You know your weapons, buddy. Terminator examines each in turn, working the actions with The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The caravan bursts into the stairwell. Mickey turns to Duncan\nSetting: derived from \"Natural Born Killers\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. STAIRWELL - DAY\nACTION: The caravan bursts into the stairwell. Mickey turns to Duncan the remaining hostage deputy.\nDIALOGUE — Mickey: Where does this lead?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Natural Born Killers\": The caravan bursts into the stairwell. Mickey turns to Duncan the remaining hostage deputy., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The caravan bursts into the stairwell. Mickey turns to Duncan the remaining hostage deputy. The camera pushes in with small amplitude at slow speed. Mickey (S1) says: [English] Where does this lead? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Avengers: Endgame\n\nSCREENPLAY EXCERPT:\nINT. STARK TOWER, LOBBY - DAY (AVENGERS 1)\n\nA1 TONY carries the TESSERACT CASE through THE LOBBY,\nfollowed by A1 THOR and LOKI.\n\nSWAT TROOPERS clear a path. ONE TURNS ASIDE, RAISING HIS\nMASK, REVEALING...OUR TONY.\n\n TONY (INTO COM)\n Okay, Thumbelina, you're a go.\n\nMOVE IN CLOSE on A1 TONY to see...SCOTT, HIDING IN HIS HAIR.\n\n SCOTT LANG (INTO COM)\n Bombs away.\n\nSCOTT DROPS BENEATH TONY'S BLACK SABBATH SHIRT."}, {"role": "assistant", "content": "## SCENE 1 — INT. STARK TOWER, LOBBY - DAY (AVENGERS 1)\nACTION: A1 TONY carries the TESSERACT CASE through THE LOBBY, followed by A1 THOR and LOKI.\nDIALOGUE — Tony: Okay, Thumbelina, you're a go.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Avengers: Endgame: A1 TONY carries the TESSERACT CASE through THE LOBBY, followed by A1 THOR and LOKI., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A1 TONY carries the TESSERACT CASE through THE LOBBY, followed by A1 THOR and LOKI. The camera pushes in with small amplitude at slow speed. Tony (S1) says: [English] Okay, Thumbelina, you're a go. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Frodo is walking beneath the trees...lost in thought. His\nSetting: derived from Black Screen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. SLOPES OF AMON HEN -- DAY\nACTION: Frodo is walking beneath the trees...lost in thought. His feet hit the rough edge of an ancient stone slab...his eyes\nDIALOGUE — Boromir: None of us should wander alone; you least of all. So much depends on you...Frodo?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: Frodo is walking beneath the trees...lost in thought. His feet hit the rough edge of an ancient stone slab...his eyes, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Frodo is walking beneath the trees...lost in thought. His feet hit the rough edge of an ancient stone slab...his eyes The camera pushes in with small amplitude at slow speed. Boromir (S1) says: [English] None of us should wander alone; you least of all. So much depends on you...Frodo? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The old, dilapidated Ghostbusters logo comes crashing to the ground,\nSetting: derived from Ext. Manhattan Island - Day\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. FIREHOUSE - DAY\nACTION: The old, dilapidated Ghostbusters logo comes crashing to the ground, dropped by a pair of workmen on a ladder, and the new logo is hoisted\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ext. Manhattan Island - Day: The old, dilapidated Ghostbusters logo comes crashing to the ground, dropped by a pair of workmen on a ladder, and the new logo is hoisted, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The old, dilapidated Ghostbusters logo comes crashing to the ground, dropped by a pair of workmen on a ladder, and the new logo is hoisted The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A FLYING PERSPECTIVE past buildings. An INSECT-LIKE BUZZING\nSetting: derived from Dark City\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. FLYING P.O.V. - NIGHT\nACTION: A FLYING PERSPECTIVE past buildings. An INSECT-LIKE BUZZING O.S. Way down BELOW, in a canyon of silent buildings, a LONE\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dark City: A FLYING PERSPECTIVE past buildings. An INSECT-LIKE BUZZING O.S. Way down BELOW, in a canyon of silent buildings, a LONE, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A FLYING PERSPECTIVE past buildings. An INSECT-LIKE BUZZING O.S. Way down BELOW, in a canyon of silent buildings, a LONE The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in GilmoreGirls illustrating the trope 'SitcomArchNemesis'. Town selectman Taylor Doose from Gilmore Girls generally exists only to annoy the main characters, and generally the whole town of Stars Hollow with his obsessive need to control every aspect of the town. Luke in particular sees him as an arch nemesis.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying SitcomArchNemesis.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for GilmoreGirls, SitcomArchNemesis, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for SitcomArchNemesis. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I know about your fucking condition,\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I know about your fucking condition, Leonard! I probably know more about it\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: I know about your fucking condition, Leonard! I probably know more about it, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I know about your fucking condition, Leonard! I probably know more about it The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Seven\n\nSCREENPLAY EXCERPT:\nINT. SLUM APARTMENT, HALLWAY -- MORNING\n\n The siege cops are in the hall, holding neighbors at bay."}, {"role": "assistant", "content": "## SCENE 1 — INT. SLUM APARTMENT, HALLWAY -- MORNING\nACTION: The siege cops are in the hall, holding neighbors at bay.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Seven: The siege cops are in the hall, holding neighbors at bay., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The siege cops are in the hall, holding neighbors at bay. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Hunter\"\n\nSCREENPLAY EXCERPT:\n------------------------------------------------------------------------\n REVISED - \"HUNTER\" - 4/17/86 81\n\n *\n\n 140 CONTINUED: 140\n\n In frustration, Schaefer hikes Ramirez higher onto his\n back, digs in and sprints to the top of the hill, Anna\n waiting at the top.\n\n\n\n 141 OUT 141\n\n\n\n 142 EXT. BILLY - DAY 142\n\n Crouches low, knife extended in a fighting position.\n\n\n\n 143- OUT 143-\n\n\n\n 144 144\n\n\n\n 145 EXT. RIDGETOP - DAY 145\n\n Over the top, in a low depression, Schaefer props\n Ramirez against some rocks, reaching for his weapon.\n They HEAR Billy's echoing SCREAM. Instantly their\n weapons are raised, cocked and ready.\n\n SCHAEFER\n Get back, into the rocks!\n\n Schaefer sweeps Anna behind him as Ramirez struggles\n to his feet. Together they back up, covering the\n ridgeline over which they've just come.\n\n Suddenly, from one side, near Ramirez, the Hunter\n bursts from the jungle, Ramirez, seeing the Hunter,\n spins raising his weapons to fire...\n\n Ramirez SEES the Hunter's onrushing face, still in\n camouflage, a montage of organic tex"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/17/86 81\nDIALOGUE — Schaefer: Get back, into the rocks!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hunter\": ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/17/86 81, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/17/86 81 The camera pushes in with small amplitude at slow speed. Schaefer (S1) says: [English] Get back, into the rocks! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Malignant\n\nSCREENPLAY EXCERPT:\nEXT. MITCHELL HOME - DAY\nSydney's car pulls into the driveway of Madison's house."}, {"role": "assistant", "content": "## SCENE 1 — EXT. MITCHELL HOME - DAY\nACTION: Sydney's car pulls into the driveway of Madison's house.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Malignant: Sydney's car pulls into the driveway of Madison's house., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sydney's car pulls into the driveway of Madison's house. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Joel gets off on the sixth floor. He searches for a room\nSetting: derived from Int. Publishing House Reception Area - Day\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ELEVATOR - DAY\nACTION: Joel gets off on the sixth floor. He searches for a room number. As he turns the corner, he sees that the hallway is\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int. Publishing House Reception Area - Day: Joel gets off on the sixth floor. He searches for a room number. As he turns the corner, he sees that the hallway is, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Joel gets off on the sixth floor. He searches for a room number. As he turns the corner, he sees that the hallway is The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Suicide\n\nSCREENPLAY EXCERPT:\nGEORGE\n It's Martine. I think he's dying. He\n tried to kill himself with pills.\n\n\n\n JERRY\n What?\n\n\n GINA\n Come on.\n\n\n JERRY\n In my pajamas? I better get my robe.\n\n\n\n GINA\n We don't have enough time.\n\n\n JERRY\n It'll take two seconds.\n\n\n GINA\n There is no time.\n\n\n JERRY\n We don't have two seconds?\n\n\n GINA\n All right. Go ahead.\n\n\n JERRY\n Nah, forget it.\n\n\n GINA\n No, go ahead.\n\n\n JERRY\n Nah. I'll just wear the pajamas.\n\n\n GINA\n Will you just get it.\n\n\n JERRY\n Are you sure?\n\n\n GINA\n Forget it. Come on.\n\n\n JERRY\n Nah, I'll go get the robe."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: It's Martine. I think he's dying. He tried to kill himself with pills.\nDIALOGUE — Jerry: What?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Suicide: It's Martine. I think he's dying. He tried to kill himself with pills., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. It's Martine. I think he's dying. He tried to kill himself with pills. The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] What? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Inglourious Basterds\n\nSCREENPLAY EXCERPT:\nYou were saying?\n\n LT.HICOX\n I was saying that makes two of us. I've\n had a gun pointed at your balls since\n you sat down.\n\n SGT.STIGLITZ\n That makes three of us.\n\n UNDERTABLE\n We See all three guns pointed at appropriate crotches. As well\n as Bridget's legs, right besides the Nazi Major's. Her pretty\n gams are sure to be chewed up in the possible crossfire.\n\n SGT.STIGLITZ\n And at this range, I'm a real\n Fredrick Zoller.\n Hugo alsobrings out his dagger, and sitcks it in the table top.\n\n MAJOR HELLSTROM\n Hummmmm ... Looks like we have a bit\n of a sticky situation here.\n\n LT. H ICOX\n What's going to happen, Major, is your\n going to stand up, and walk out that\n door with us.\n\n MAJOR HELLSTROM\n No no no no no no, I don't think so.\n I'm afraid you and I both know, no matter\n what happens to anybody else in this room,\n the two of us aren't going anywhere.\n\n\n\n\n\n\n\n\n\n \"7 .\n\n MAJOR HELLSTROM\n (pointing behind\n him at the table)\n Too bad about them though. They seem\n like a likeable\n (referring to\n Stiglitz and Wicki)\n You two will have to shoot them.\n\n BRIDGET\n Then Major, i implore you. For the\n sake of those German troops, will\n you please leave with us?\n\n MAJOR HELLSTROM\n Oh Bridget, your concern for German\n t"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I was saying that makes two of us. I've had a gun pointed at your balls since\nDIALOGUE — Lt.Hicox: I was saying that makes two of us. I've had a gun pointed at your balls since you sat down.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Inglourious Basterds: I was saying that makes two of us. I've had a gun pointed at your balls since, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I was saying that makes two of us. I've had a gun pointed at your balls since The camera pushes in with small amplitude at slow speed. Lt.Hicox (S1) says: [English] I was saying that makes two of us. I've had a gun pointed at your balls since you sat down. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Alien\n\nSCREENPLAY EXCERPT:\nHe's running a fever.\n\n STANDARD\n Still unconscious?\n\n HUNTER"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: He's running a fever.\nDIALOGUE — Standard: Still unconscious?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Alien: He's running a fever., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He's running a fever. The camera pushes in with small amplitude at slow speed. Standard (S1) says: [English] Still unconscious? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Heavy Metal Movie\n\nSCREENPLAY EXCERPT:\nINT. - THE \"PACIFIC PEARL\"\n\n\n NAVIGATOR\n\n We're there Stepek, right on\n target ... Rabual dead ahead!\n\n\n Suddenly a cacophony of explosions split the air as anti-aircraft\n shells explode ahead of the aircraft, lighting the inside with\n electric blue flashes.\n\n\n BOMBARDIER\n\n Gee! Thanks for telling me,\n Holden. I thought it was Cleveland.\n\n\n CUT TO\n\n\n THE GROUND\n\n Ackack guns blast shells skyward. The gun blasts light up\n the palm tree background, causing momentary silhouettes.\n\n\n CUT TO\n\n\n THE BOMBARDIER\n\n\n BOMBARDIER\n\n For Christ sake, hold her steady!\n\n\n CUT TO\n\n\n THE BELLYGUNNER\n\n He can be seen from outside the plane as the shell flashes\n light up his turret.\n\n\n BELLYGUNNER\n\n Quit yer bitching Stepek, just drop\n 'em and let's get outta here!\n\n\n CUT BACK TO\n\n\n THE BOMBARDIER\n\n\n BOMBARDIER\n\n Keep your shirts on guys, tonight I'm\n gonna hit something.\n (he settles over the bomb sight)\n Left two degrees ... steady ...\n steady ...\n\n\n CUT TO\n\n\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. - THE \"PACIFIC PEARL\"\nACTION: We're there Stepek, right on target ... Rabual dead ahead!\nDIALOGUE — Navigator: We're there Stepek, right on target ... Rabual dead ahead!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Heavy Metal Movie: We're there Stepek, right on target ... Rabual dead ahead!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We're there Stepek, right on target ... Rabual dead ahead! The camera pushes in with small amplitude at slow speed. Navigator (S1) says: [English] We're there Stepek, right on target ... Rabual dead ahead! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I'm not worried. Everything will\nSetting: derived from Sunset Boulevard\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I'm not worried. Everything will be fine. The old team together.\nDIALOGUE — De Mille: Get Gordon Cole. Tell him to forget about her car. He can find another old car. I'll buy him five old cars, if necessary.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Sunset Boulevard: I'm not worried. Everything will be fine. The old team together., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm not worried. Everything will be fine. The old team together. The camera pushes in with small amplitude at slow speed. De Mille (S1) says: [English] Get Gordon Cole. Tell him to forget about her car. He can find another old car. I'll buy him five old cars, if necessary. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: 1 Ext. Cityscape Sunset 1\n\nSCREENPLAY EXCERPT:\nYou're a stupid, fat arsed,\n obstructive, fascist moron aren't\n you?\n\n PORTER\n If you say so.\n\n JILL\n You think these are tits don't you?\n\n PORTER\n Ah.\n\n JILL\n I bet you'd like to touch them?\n\n PORTER\n Oh.\n\n JILL\n Well don't. You're looking at twenty\n pounds of high explosive! And if you\n don't stamp this form I'm going to\n blow the place up!\n\n JILL thumps the desk with her fist.\n\n\n79 INT. BASEMENT DAY 79\n\n The GUARD throws the alarm switch.\n\n\n80 INT. LOBBY DAY 80\n\n Alarm bells start ringing and from secret doors heavily\n armed GUARDS leap out, their guns trained on JILL as she\n appears to be the only person around.\n\n GUARD"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You're a stupid, fat arsed, obstructive, fascist moron aren't\nDIALOGUE — Porter: If you say so.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from 1 Ext. Cityscape Sunset 1: You're a stupid, fat arsed, obstructive, fascist moron aren't, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You're a stupid, fat arsed, obstructive, fascist moron aren't The camera pushes in with small amplitude at slow speed. Porter (S1) says: [English] If you say so. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Bumstead enters darkness - pneumatic MACHINE NOISE fills the\nSetting: derived from Dark City\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HOSPITAL ROOM\nACTION: Bumstead enters darkness - pneumatic MACHINE NOISE fills the gloom. He walks quietly to his mother, encased in a metallic\nDIALOGUE — Bumstead: Brought you these.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dark City: Bumstead enters darkness - pneumatic MACHINE NOISE fills the gloom. He walks quietly to his mother, encased in a metallic, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Bumstead enters darkness - pneumatic MACHINE NOISE fills the gloom. He walks quietly to his mother, encased in a metallic The camera pushes in with small amplitude at slow speed. Bumstead (S1) says: [English] Brought you these. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: They enter the large, dark industrial section of the\nSetting: derived from Watchmen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PYRAMID INDUSTRIES - PROCESSING PLANT - NIGHT\nACTION: They enter the large, dark industrial section of the building. They pass huge PROCESSING VATS, CONVEYOR BELTS.\nDIALOGUE — Dan: I tapped into all the databases--tax records, business registries, real estate records--trying to get a bead on who's behind Pyramid but it's a never ending...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Watchmen: They enter the large, dark industrial section of the building. They pass huge PROCESSING VATS, CONVEYOR BELTS., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They enter the large, dark industrial section of the building. They pass huge PROCESSING VATS, CONVEYOR BELTS. The camera pushes in with small amplitude at slow speed. Dan (S1) says: [English] I tapped into all the databases--tax records, business registries, real estate records--trying to get a bead on who's behind Pyramid but it's a never ending... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in ChildrenOfTheStones illustrating the trope 'Telepathy'. Matthew finds that he has a telepathic ability to see what they are experiencing if he is holding one of their possessions. He uses this to experience the brain-washing of Margaret and her daughter Sandra.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying Telepathy.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for ChildrenOfTheStones, Telepathy, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for Telepathy. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DropTheDeadDonkey illustrating the trope 'PetTheDog'. Sally's frequent staged attempts generally end in disaster.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying PetTheDog.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DropTheDeadDonkey, PetTheDog, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for PetTheDog. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fantastic Mr. Fox\n\nSCREENPLAY EXCERPT:\nEXT. ROAD. DAY\n\n Fox, Kylie, Ash, and Kristofferson ride down a country road.\n Kylie sees something across the meadow. He says warily:\n\n KYLIE\n Don't turn around!\n\n FOX\n What?\n Fox turns around. A huge, wild, grey wolf with ice-blue eyes\n stands on a rock fifty feet away from them. Fox slams on the\n brakes. The motorcycle slides to a halt.\n\n FOX\n Where'd he come from?\n\n (LOUDLY)\n Where'd you come from? What are you doing\n here?\n Pause. Fox points toward the wolf:\n\n FOX\n Canis lupus!\n Fox points to himself:\n\n FOX\n Vulpes Vulpes!\n The wolf does not answer. Fox, Kylie, Ash, and Kristofferson\n watch idling from the motorcycle.\n\n FOX\n I don't think he speaks English or Latin.\n\n (LOUDLY)\n Pensez-vous que 1'hiver sera rude?\n\n (ASIDE)\n I'm asking if he thinks we're in for a\n hard winter.\n\n\n\n\n\n\n\n\n\n 88.\n The wolf shakes his head. Fox nods.\n\n FOX\n He doesn't seem to know.\n Silence. Fox shouts to the wolf with a strange hitch in his\n\n VOICE:\n\n FOX\n I have a phobia of wolves!\n The wolf does not answer. It breathes heavily with its mouth\n open. Its teeth are long, sh"}, {"role": "assistant", "content": "## SCENE 1 — EXT. ROAD. DAY\nACTION: Fox, Kylie, Ash, and Kristofferson ride down a country road. Kylie sees something across the meadow. He says warily:\nDIALOGUE — Kylie: Don't turn around!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fantastic Mr. Fox: Fox, Kylie, Ash, and Kristofferson ride down a country road. Kylie sees something across the meadow. He says warily:, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Fox, Kylie, Ash, and Kristofferson ride down a country road. Kylie sees something across the meadow. He says warily: The camera pushes in with small amplitude at slow speed. Kylie (S1) says: [English] Don't turn around! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The siege cops are in the hall, holding neighbors at bay.\nSetting: derived from Seven\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SLUM APARTMENT, HALLWAY -- MORNING\nACTION: The siege cops are in the hall, holding neighbors at bay.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Seven: The siege cops are in the hall, holding neighbors at bay., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The siege cops are in the hall, holding neighbors at bay. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ZARKOV, BARIN and TIGER MAN lie in chains, heads cocked at\nSetting: derived from Flash Gordon\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. DUNGEON\nACTION: ZARKOV, BARIN and TIGER MAN lie in chains, heads cocked at SOUNDS of the facas outside. THUD of a body falling, brief\nDIALOGUE — Flash'S Voice: It's me! 74.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Flash Gordon: ZARKOV, BARIN and TIGER MAN lie in chains, heads cocked at SOUNDS of the facas outside. THUD of a body falling, brief, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ZARKOV, BARIN and TIGER MAN lie in chains, heads cocked at SOUNDS of the facas outside. THUD of a body falling, brief The camera pushes in with small amplitude at slow speed. Flash'S Voice (S1) says: [English] It's me! 74. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: No. No Mr. Lebowski, that had not\nSetting: derived from The Big Lebowski\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: No. No Mr. Lebowski, that had not That had not occurred to us, Dude.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Big Lebowski: No. No Mr. Lebowski, that had not That had not occurred to us, Dude., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. No. No Mr. Lebowski, that had not That had not occurred to us, Dude. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Script Date 5/25/94\n\nSCREENPLAY EXCERPT:\nThese gentlemen are with the New York\n police department. You look great,\n Keaton. Better than I would have thought.\n\n RENAULT\n Is there a problem, Mr. Keaton?\n\n KUJAN\n The small matter of a stolen truck-load\n of guns that wound up on a boat to\n Ireland last night.\n\nRenault and Fortier's confusion is giving way to suspicion.\n\n FORTIER\n Mr. Keaton?\n\n KEATON\n If you will excuse us for a moment,\n gentlemen.\n\n KUJAN\n We need to ask you some questions\n downtown. You'll be quite awhile.\n\nRenault starts to get up.\n\n RENAULT\n We should leave you to discuss whatever\n this is.\n\n KEATON\n Please. Sit.\n\nKeaton stands up and throws a wad of money on the table to\ncover the check. He looks at Edie. She moves to stand, but he\nsits her back down with a hand on her shoulder.\n\n KEATON\n Enjoy the meal."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: These gentlemen are with the New York police department. You look great,\nDIALOGUE — Renault: Is there a problem, Mr. Keaton?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Script Date 5/25/94: These gentlemen are with the New York police department. You look great,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. These gentlemen are with the New York police department. You look great, The camera pushes in with small amplitude at slow speed. Renault (S1) says: [English] Is there a problem, Mr. Keaton? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: It's raining like hell.\nSetting: derived from \"Hellboy\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BPRD BUILDING - DAY\nACTION: It's raining like hell. The PALLBEARERS, Manning and Myers among them, load Broom's\nDIALOGUE — Liz'S Voice: He hasn't spoken to anyone in three days. Not a word. He doesn't eat, he doesn't sleep...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hellboy\": It's raining like hell. The PALLBEARERS, Manning and Myers among them, load Broom's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. It's raining like hell. The PALLBEARERS, Manning and Myers among them, load Broom's The camera pushes in with small amplitude at slow speed. Liz'S Voice (S1) says: [English] He hasn't spoken to anyone in three days. Not a word. He doesn't eat, he doesn't sleep... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in CoronationStreet illustrating the trope 'SmugSmiler'. Mike Baldwin from Coronation Street has the greatest, self-satisfied, shit-eating grin to end all self-satisfied, shit-eating grins.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying SmugSmiler.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for CoronationStreet, SmugSmiler, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for SmugSmiler. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Chip is cutting through the park on his way to the dance.\nSetting: derived from Jennifer'S Body\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. MC CULLUM PARK - CONTINUOUS\nACTION: Chip is cutting through the park on his way to the dance. He whistles to himself.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jennifer'S Body: Chip is cutting through the park on his way to the dance. He whistles to himself., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Chip is cutting through the park on his way to the dance. He whistles to himself. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Seven\n\nSCREENPLAY EXCERPT:\nINT. OLD HOUSE -- DAY\n\n Sunlight comes through the soot on the windows, more brown than\n bright. SOMERSET, 45, in a suit and tie, stands in this empty\n second-story room. He looks around, at the ceiling, at the worn\n wooden floor, at the peeling wallpaper on the walls.\n\n Somerset walks to one wall where the current wallpaper is peeled\n away to reveal flowery wallpaper underneath. He runs his finger\n across one of the pale red roses that decorates the older paper.\n He pushes the grime away, brings the rose out more clearly.\n\n He pulls at the edge of the paper, carefully ripping off a\n roughly squared section with the rose at its center.\n\n He studies it in his hand."}, {"role": "assistant", "content": "## SCENE 1 — INT. OLD HOUSE -- DAY\nACTION: Sunlight comes through the soot on the windows, more brown than bright. SOMERSET, 45, in a suit and tie, stands in this empty\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Seven: Sunlight comes through the soot on the windows, more brown than bright. SOMERSET, 45, in a suit and tie, stands in this empty, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sunlight comes through the soot on the windows, more brown than bright. SOMERSET, 45, in a suit and tie, stands in this empty The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Crank\n\nSCREENPLAY EXCERPT:\nEXT. REAR EXIT, EVE'S BUILDING - CONTINUOUS\n\n From overhead, the CAMERA CORKSCREWS CLOCKWISE as CHEV\n wrenches the HOOD'S neck COUNTERCLOCKWISE, snapping it.\n\n EVE\n You know, I could use a little help here.\n\n CHEV shoves the HOOD'S body into a dumpster just as EVE\n turns... CHEV snatches up the first thing he sees - a grime\n encrusted plastic SHOWER CAP - and holds it up lamely.\n\n CHEV\n Is this yours?\n\n EVE rolls her eyes, looks around.\n\n EVE\n Where's your car?\n\n\n\n (CONTINUED)\n\n W\n\n\n\n\n 8\n\n\n F\n\n\n\n\n 59.\n CONTINUED:\n\n\n CHEV\n My car. Actually... I took a cab."}, {"role": "assistant", "content": "## SCENE 1 — EXT. REAR EXIT, EVE'S BUILDING - CONTINUOUS\nACTION: From overhead, the CAMERA CORKSCREWS CLOCKWISE as CHEV wrenches the HOOD'S neck COUNTERCLOCKWISE, snapping it.\nDIALOGUE — Eve: You know, I could use a little help here.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Crank: From overhead, the CAMERA CORKSCREWS CLOCKWISE as CHEV wrenches the HOOD'S neck COUNTERCLOCKWISE, snapping it., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. From overhead, the CAMERA CORKSCREWS CLOCKWISE as CHEV wrenches the HOOD'S neck COUNTERCLOCKWISE, snapping it. The camera pushes in with small amplitude at slow speed. Eve (S1) says: [English] You know, I could use a little help here. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Int: Bedroom. Early-Morning Hours\n\nSCREENPLAY EXCERPT:\nWe didn't just have sex in the\n\t\tbathroom?\n\n\t\t\t\tDANTE\n\t\tNo.\n\n\tEveryone is silent. Then...\n\n\t\t\t\tCAITLIN\n\t\tStop this. This isn't funny.\n\n\t\t\t\tDANTE\n\t\tI'm not kidding. I just got back\n\t\tfrom outside.\n\n\t\t\t\tCAITLIN"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: We didn't just have sex in the Everyone is silent. Then...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int: Bedroom. Early-Morning Hours: We didn't just have sex in the Everyone is silent. Then..., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We didn't just have sex in the Everyone is silent. Then... The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nINT. CAMPBELL CENTER - LABORATORY\n\n\tShaken, but oddly transcendent, Catherine watches as\n\tStargher's body is wheeled out of the lab on a gurney. She\n\tcomforts a whimpering Valentine and tells him...\n\n\t\t\t\tCATHERINE\n\t\tEverything's alright now, Valentine. I\n\t\tpromise."}, {"role": "assistant", "content": "## SCENE 1 — INT. CAMPBELL CENTER - LABORATORY\nACTION: Shaken, but oddly transcendent, Catherine watches as Stargher's body is wheeled out of the lab on a gurney. She\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: Shaken, but oddly transcendent, Catherine watches as Stargher's body is wheeled out of the lab on a gurney. She, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Shaken, but oddly transcendent, Catherine watches as Stargher's body is wheeled out of the lab on a gurney. She The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\n78.\n\n123 CONTINUED:\t\t\t\t\t\t\t 123\n\n\t\t\t\tNATALIE\n\t\tI know about your fucking condition,\n\t\tLeonard! I probably know more about it\n\t\tthan you do! You don't have a fucking\n\t\tclue about anything else!\n\n\t\t\t\tLEONARD\n\t\tWhat happened?\n\n\t\t\t\tNATALIE\n\t\tWhat happened is that Jimmy went to meet\n\t\ta guy called Teddy. He took a lot of\n\t\tmoney with him and he didn't come back.\n\t\tJimmy's partners think I set him up. I\n\t\tdon't know whether you know this Teddy or\n\t\thow well -\n\nLeonard is getting frustrated.\n\n\t\t\t\tLEONARD\n\t\tNeither do I.\n\n\t\t\t\tNATALIE\n\t\tDon't protect him.\n\n\t\t\t\tLEONARD\n\t\tI'm not.\n\n\t\t\t\tNATALIE\n\t\tHelp me.\n\n\t\t\t\tLEONARD\n\t\tHow?\n\n\t\t\t\tNATALIE\n\t\tGet rid of Dodd for me.\n\n\t\t\t\tLEONARD\n\t\tWhat?\n\n\t\t\t\tNATALIE\n\t\tKill him. I'll pay you.\n\n\t\t\t\tLEONARD\n\t\tWhat do you think I am?! I'm not gonna\n\t\tkill someone for money.\n\n\t\t\t\tNATALIE\n\t\tWhat then? Love? What would you kill for?\n\t\tFor your wife, right?\n\n\t\t\t\tLEONARD\n\t\tThat's different."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I know about your fucking condition, Leonard! I probably know more about it\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: I know about your fucking condition, Leonard! I probably know more about it, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I know about your fucking condition, Leonard! I probably know more about it The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Frodo, Sam, Merry, and Pippin hurry through the\nSetting: derived from Black Screen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. FERRY LANE -- NIGHT\nACTION: Frodo, Sam, Merry, and Pippin hurry through the trees...slipping and sliding on the muddy ground.\nDIALOGUE — Sam: Anything?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: Frodo, Sam, Merry, and Pippin hurry through the trees...slipping and sliding on the muddy ground., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Frodo, Sam, Merry, and Pippin hurry through the trees...slipping and sliding on the muddy ground. The camera pushes in with small amplitude at slow speed. Sam (S1) says: [English] Anything? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Joker\n\nSCREENPLAY EXCERPT:\n52.\n\n JOKER (CONT'D)\n Apparently I'm Thomas Wayne's son!\n How could you keep that from me?\n\nHis mother slowly getting up out of bed.\n\n MOM\n Stop yelling at me, you're gonna\n kill me, give me a heart attack!\n\nShe goes into the bathroom.\n\n JOKER"}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: Apparently I'm Thomas Wayne's son! How could you keep that from me?\nDIALOGUE — Joker: Apparently I'm Thomas Wayne's son! How could you keep that from me?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Joker: Apparently I'm Thomas Wayne's son! How could you keep that from me?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Apparently I'm Thomas Wayne's son! How could you keep that from me? The camera pushes in with small amplitude at slow speed. Joker (S1) says: [English] Apparently I'm Thomas Wayne's son! How could you keep that from me? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Ninja Assassin\n\nSCREENPLAY EXCERPT:\nGOLDENROD REVISED 2nd DRAFT 6/4/08 48A.\n54 CONTINUED: (2) 54\n\n He waits. *\n\n MIKA (CONT'D) *\n Uhh-- Well, there are these *\n clans...we're not sure how many. *\n\n RAIZO *\n Nine. *\n\n MIKA *\n Okay. Nine. Nine clans. And *\n they apparently have been *\n supplying assassins to governments *\n or anyone who happens to have one *\n hundred pounds of gold lying *\n around for the last thousand *\n years. Give or take. *\n\n He nods faintly. Then turns to his dead \"Brother\". *\n\n RAIZO *\n His name was Jin. *\n\n MIKA *\n You knew him? *\n\n RAIZO *\n He was not born an \"Assassin\". "}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: GOLDENROD REVISED 2nd DRAFT 6/4/08 48A. 54 CONTINUED: (2) 54\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ninja Assassin: GOLDENROD REVISED 2nd DRAFT 6/4/08 48A. 54 CONTINUED: (2) 54, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GOLDENROD REVISED 2nd DRAFT 6/4/08 48A. 54 CONTINUED: (2) 54 The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Wait a minute Yori! YORI!! LOOK!\nSetting: derived from T R O N\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Wait a minute Yori! YORI!! LOOK! 575 FULL SHOT 575\nDIALOGUE — Yori: TRON!!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T R O N: Wait a minute Yori! YORI!! LOOK! 575 FULL SHOT 575, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Wait a minute Yori! YORI!! LOOK! 575 FULL SHOT 575 The camera pushes in with small amplitude at slow speed. Yori (S1) says: [English] TRON!! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Bodyguard\n\nSCREENPLAY EXCERPT:\nFrank stays.\nSpector considers her for a beat, then looks at the others.\nThe secretary interrupts from the back of the office.\n SECRETARY (O.S.)\n Mr. Schiller calling on line one...\nSpector ignores her, still holding his look at Devaney,\nRachel and Frank.\n SPECTOR\n Well, I guess this is democracy in\n action.\n\n 75.\n\n\nRachel nods. Spector, turning to take the call, flashes\nher a \"little boy\" look.\n SPECTOR\n I gave in. That's not easy for\n me. Don't I get a hug?\n\nRachel is stone-faced. Spector covers by picking up the\nphone.\n SPECTOR"}, {"role": "assistant", "content": "## SCENE 1 — INT. O.S.\nACTION: Spector considers her for a beat, then looks at the others. The secretary interrupts from the back of the office.\nDIALOGUE — Secretary: Mr. Schiller calling on line one... Spector ignores her, still holding his look at Devaney, Rachel and Frank.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Bodyguard: Spector considers her for a beat, then looks at the others. The secretary interrupts from the back of the office., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Spector considers her for a beat, then looks at the others. The secretary interrupts from the back of the office. The camera pushes in with small amplitude at slow speed. Secretary (S1) says: [English] Mr. Schiller calling on line one... Spector ignores her, still holding his look at Devaney, Rachel and Frank. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The powerful pirate starship blasts through space as it heads toward\nSetting: derived from The Empire Strikes Back\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BESPIN SYSTEM - MILLENNIUM FALCON - DAWN\nACTION: The powerful pirate starship blasts through space as it heads toward the soft pink planet of Bespin.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Empire Strikes Back: The powerful pirate starship blasts through space as it heads toward the soft pink planet of Bespin., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The powerful pirate starship blasts through space as it heads toward the soft pink planet of Bespin. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Black Screen\n\nSCREENPLAY EXCERPT:\nEXT. FERRY LANE -- NIGHT\n\nFrodo, Sam, Merry, and Pippin hurry through the\ntrees...slipping and sliding on the muddy ground.\n\n SAM\n Anything?\n\n FRODO\n Nothing.\n\n PIPPIN\n What is going on?\n\nMerry moves past pippin, toward Frodo, watching\nintently...Sam keeps looking around nervously.\n\n MERRY\n The Black rider was looking for\n something...or someone...Frodo?\n\n SAM\n Get down!\n\nThe SILHOUETTE OF A BLACK RIDER looms against the skyline.\nThe Four Hobbits... sprawled on the ground, holding their\nbreath. The BLACK RIDER turns and departs.\n\n FRODO\n I have to leave the Shire...Sam and I\n must get to Bree.\n\nMerry looks at his friend...realizing Frodo is in deep\ntrouble.\n\n MERRY\n Right...Buckleberry Ferry...follow me !\n\nThe Hobbits break cover. SUDDENLY, A RINGWRAITH bursts out of\nthe forest TOWARD THEM!\n\n MERRY (CONT'D)\n There's another one!! Frodo, this way! !\n\nThe Hobbits run THE RINGWRAITH SHRIEKS!\n\n QUICK CUTS:\n\n (CONTINUED)\n\n 43.\nCONTINUED:\n\n\nBlack horse hooves... snarling horse mouths...a fleeting\nblack cowl.\n\n MERRY (CONT'D)\n Frodo, follow me!\n\n CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. FERRY LANE -- NIGHT\nACTION: Frodo, Sam, Merry, and Pippin hurry through the trees...slipping and sliding on the muddy ground.\nDIALOGUE — Sam: Anything?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: Frodo, Sam, Merry, and Pippin hurry through the trees...slipping and sliding on the muddy ground., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Frodo, Sam, Merry, and Pippin hurry through the trees...slipping and sliding on the muddy ground. The camera pushes in with small amplitude at slow speed. Sam (S1) says: [English] Anything? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages":[{"role":"system","content":"You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment + integrated_multimodal_description + overall_soundscape + non_diegetic_music), DURATION 5."},{"role":"user","content":"Premise: A courier in a yellow raincoat delivers a package through a neon alley in the rain.\nSetting: Night, rain-slicked street, pink and cyan neon\nTone: moody, cinematic\nWrite SCENE 1 now."},{"role":"assistant","content":"## SCENE 1 — EXT. NEON ALLEY - NIGHT\nACTION: A courier in a yellow raincoat walks toward camera through steam and puddles; the camera pushes in until their hooded face fills the frame.\nDIALOGUE — COURIER: Almost there.\nSHOT: medium-wide shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic night alley, rain-slicked pavement, pink and cyan neon signs reflecting in puddles, courier in yellow raincoat walking toward camera, steam from street vent, moody live-action, sharp detail\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium-wide shot opens on the rain-slicked neon alley composition of Picture 1: a courier in a yellow raincoat walks toward the camera through rising steam, pink and cyan signs reflecting in puddles. The camera pushes in with small amplitude at slow speed as rain streaks past the lens. The courier (S1) says in a quiet, breathy voice: [English] Almost there. Their pace steadies until the framing settles into the extreme close-up of the hooded face established by Picture 2, neon catching in their eyes.\n\noverall_soundscape: Steady rain on asphalt and metal awnings, soft steam hiss from a street vent, distant traffic wet-road rumble.\n\nnon_diegetic_music: Sparse synth pads at a slow tempo with a low sub-bass pulse that gently rises then holds.\nLORA: none\nAUDIO: rain bed, optional neon buzz\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Buckaroo Banzai\n\nSCREENPLAY EXCERPT:\nINT. JET CAR - DAY\t51\n\nBuckaroo STILL ACCELERATING, struggling with his mighty\nmachine...fierce vibrations, OIL AND SMOKE filling the cockpit...as he\nwipes hot oil off his goggles..."}, {"role": "assistant", "content": "## SCENE 1 — INT. JET CAR - DAY\t51\nACTION: Buckaroo STILL ACCELERATING, struggling with his mighty machine...fierce vibrations, OIL AND SMOKE filling the cockpit...as he\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Buckaroo Banzai: Buckaroo STILL ACCELERATING, struggling with his mighty machine...fierce vibrations, OIL AND SMOKE filling the cockpit...as he, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Buckaroo STILL ACCELERATING, struggling with his mighty machine...fierce vibrations, OIL AND SMOKE filling the cockpit...as he The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jay And Silent Bob Strike Back\n\nSCREENPLAY EXCERPT:\nINT. SEWER TUNNEL--SAME\n\n Jay and Suzanne fall backwards, as Silent Bob pops through.\n\n JAY\n INCOMING!!!\n\n SILENT BOB\n AAAIIIGGGGGHHHHHHHH!!!\n\n SUZANNE\n OOOOOOOOO!!!"}, {"role": "assistant", "content": "## SCENE 1 — INT. SEWER TUNNEL--SAME\nACTION: Jay and Suzanne fall backwards, as Silent Bob pops through.\nDIALOGUE — Jay: INCOMING!!!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jay And Silent Bob Strike Back: Jay and Suzanne fall backwards, as Silent Bob pops through., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jay and Suzanne fall backwards, as Silent Bob pops through. The camera pushes in with small amplitude at slow speed. Jay (S1) says: [English] INCOMING!!! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I was saying that makes two of us. I've\nSetting: derived from Inglourious Basterds\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I was saying that makes two of us. I've had a gun pointed at your balls since\nDIALOGUE — Lt.Hicox: I was saying that makes two of us. I've had a gun pointed at your balls since you sat down.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Inglourious Basterds: I was saying that makes two of us. I've had a gun pointed at your balls since, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I was saying that makes two of us. I've had a gun pointed at your balls since The camera pushes in with small amplitude at slow speed. Lt.Hicox (S1) says: [English] I was saying that makes two of us. I've had a gun pointed at your balls since you sat down. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: GOLDENROD REVISED 2nd DRAFT 6/4/08 8A.\nSetting: derived from Ninja Assassin\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: GOLDENROD REVISED 2nd DRAFT 6/4/08 8A. 5A CONTINUED: 5A\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ninja Assassin: GOLDENROD REVISED 2nd DRAFT 6/4/08 8A. 5A CONTINUED: 5A, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. GOLDENROD REVISED 2nd DRAFT 6/4/08 8A. 5A CONTINUED: 5A The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Gloria and Sasha are totally impressed by the beauty of\nSetting: derived from Barbie\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BARBIE LAND. DAY.\nACTION: Gloria and Sasha are totally impressed by the beauty of BARBIE LAND as they drive by the ocean in Barbie's\nDIALOGUE — Barbie Margot: I can feel my heels lifting already. Yes, this is what I was supposed to do. Bring you back here!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Barbie: Gloria and Sasha are totally impressed by the beauty of BARBIE LAND as they drive by the ocean in Barbie's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Gloria and Sasha are totally impressed by the beauty of BARBIE LAND as they drive by the ocean in Barbie's The camera pushes in with small amplitude at slow speed. Barbie Margot (S1) says: [English] I can feel my heels lifting already. Yes, this is what I was supposed to do. Bring you back here! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nEXT. USO SHOW - NIGHT\n\n The MAC Mothers are on a stage, preparing two electric\n chairs.\n\n KYLE'S MOTHER\n Something must be done!\n Tomorrow it will be!\n We're going to execute this threat\n to our democracy!\n\n All the mothers join in as they decorate the chairs with\n American flags. Kenny's mother is counting money she has made\n from selling t-shirts.\n\n M.A.C. MOTHERS\n And after that our kids will be\n safe from all the Canadian scum!\n The time is now the time is here\n Tomorrow something will be done!\n\n The song still CONTINUES as we\n\n CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. USO SHOW - NIGHT\nACTION: The MAC Mothers are on a stage, preparing two electric Something must be done!\nDIALOGUE — Kyle'S Mother: Something must be done! Tomorrow it will be! We're going to execute this threat to our democracy!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: The MAC Mothers are on a stage, preparing two electric Something must be done!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The MAC Mothers are on a stage, preparing two electric Something must be done! The camera pushes in with small amplitude at slow speed. Kyle'S Mother (S1) says: [English] Something must be done! Tomorrow it will be! We're going to execute this threat to our democracy! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Sunlight comes through the soot on the windows, more brown than\nSetting: derived from Seven\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. OLD HOUSE -- DAY\nACTION: Sunlight comes through the soot on the windows, more brown than bright. SOMERSET, 45, in a suit and tie, stands in this empty\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Seven: Sunlight comes through the soot on the windows, more brown than bright. SOMERSET, 45, in a suit and tie, stands in this empty, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sunlight comes through the soot on the windows, more brown than bright. SOMERSET, 45, in a suit and tie, stands in this empty The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Lord Of The Rings: The Return Of The King\n\nSCREENPLAY EXCERPT:\nEXT. PELENNOR FIELDS - DAY\n\nFARAMIR and a few survivors flee WEST OSGILIATH, desperately\nriding toward MINAS TIRITH . . . in the SKY above DARK SHAPES\npursue them ...\n\nANGLE ON: SUDDENLY, from on HIGH, the NAZGUL strike!\n\nTILT DOWN ... following the NAZGUL as it SWOOPS out of LOW-\nLYING, MURKY CLOUDS tossing MEN in the AIR like LIMP RAG-DOLLS!\n\nAT THAT MOMENT: Across the PLAIN, a FIGURE clothed in WHITE\nrides towards them!\n\nFrom the BATTLEMENTS of MINAS TIRITH, a call . . . \"The White\nRider!\""}, {"role": "assistant", "content": "## SCENE 1 — EXT. PELENNOR FIELDS - DAY\nACTION: FARAMIR and a few survivors flee WEST OSGILIATH, desperately riding toward MINAS TIRITH . . . in the SKY above DARK SHAPES\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Return Of The King: FARAMIR and a few survivors flee WEST OSGILIATH, desperately riding toward MINAS TIRITH . . . in the SKY above DARK SHAPES, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. FARAMIR and a few survivors flee WEST OSGILIATH, desperately riding toward MINAS TIRITH . . . in the SKY above DARK SHAPES The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Heavenly Creatures\n\nSCREENPLAY EXCERPT:\nINT. ILAM/JULIET'S BEDROOM - NIGHTJuliet is propped up in bed, looking sick and pale. Hilda is sitting by the side of the bed. Henry stands at the foot of the bed, hands behind his back.\n\nHENRY Your mother and I have talked things over and we've made some decisions . . . I am resigning from the university and will be taking up a position back in England.\n\nJuliet looks horrified.\n\nJULIET\n(Vehemently) But Daddy, you can't just leave me with Mother!\n\nHilda looks hurt.\n\nHILDA We thought it best that you accompany your father. . .\n\nJULIET\n(Worried) Is Gina coming, too?\n\nHENRY (Annoyed) Of course not!\n\nJULIET\n(Angry) I'm not going to England without Gina!\n\nHilda and Henry exchange an awkward glance.\n\nHILDA You're not going to England, darling.\n\nJuliet looks confused.\n\nHENRY I'm leaving you in South Africa with Auntie Ina.\n\nJuliet starts to tremble.\n\nHILDA That chest of yours isn't getting any better. The warmer climate is just what you need . . .\n\nHENRY For the good of your health.\n\nJuliet opens her mouth and screams.\n\nCRASH CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — INT. ILAM/JULIET'S BEDROOM - NIGHTJULIET IS PROPPED UP IN BED, LOOKING SICK AND PALE. HILDA IS SITTING BY THE SIDE OF THE BED. HENRY STANDS AT THE FOOT OF THE BED, HANDS BEHIND HIS BACK.\nACTION: HENRY Your mother and I have talked things over and we've made some decisions . . . I am resigning from the university and will be taking up a position back in England. Juliet looks horrified.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Heavenly Creatures: HENRY Your mother and I have talked things over and we've made some decisions . . . I am resigning from the university and will be taking up a position back in England. Juliet looks horrified., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. HENRY Your mother and I have talked things over and we've made some decisions . . . I am resigning from the university and will be taking up a position back in England. Juliet looks horrified. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Legion\n\nSCREENPLAY EXCERPT:\nINT. DINER BATHROOM - CONTINUOUS\n\n Sandra splashes water on her face. Stares at herself in the\n mirror. A living wreck."}, {"role": "assistant", "content": "## SCENE 1 — INT. DINER BATHROOM - CONTINUOUS\nACTION: Sandra splashes water on her face. Stares at herself in the mirror. A living wreck.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Legion: Sandra splashes water on her face. Stares at herself in the mirror. A living wreck., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sandra splashes water on her face. Stares at herself in the mirror. A living wreck. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \" T E R M I N A T O R \"\n\nSCREENPLAY EXCERPT:\n...Suspect vehicle sighted on\n Motor at Pico, southbound.\n Units Two-Zero-Six and Five-\n Seven, attempt intercept.\n Unit One-Four-Three, come in.\n\n Terminator picks up the mike. He speaks in a\n simulation of the young cop's southern twang.\n\n TERMINATOR\n This is One-Four-Three. West-\n bound on Olympic, approaching\n Overland.\n\n CUT TO:\n\n\n118 EXT. SANTA MONICA FREEWAY - NIGHT 118\n\n The grey sedan moves through traffic like a hell-bent\n wraith. Reese has the hammer down. He handles the\n car with nerves of steel.\n\n CUT TO:\n\n\n119 EXT. POLICE HELICOPTER - NIGHT 119\n\n Below, Reese's sedan snakes along at 110 plus. The\n chopper, F.G., drops toward it.\n\n PILOT (V.O.)"}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: ...Suspect vehicle sighted on Motor at Pico, southbound.\nDIALOGUE — Terminator: This is One-Four-Three. West- bound on Olympic, approaching Overland.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \" T E R M I N A T O R \": ...Suspect vehicle sighted on Motor at Pico, southbound., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ...Suspect vehicle sighted on Motor at Pico, southbound. The camera pushes in with small amplitude at slow speed. Terminator (S1) says: [English] This is One-Four-Three. West- bound on Olympic, approaching Overland. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Watchmen\n\nSCREENPLAY EXCERPT:\nINT. OWL-SHIP (MOVING) - NIGHT\n\n Dan and Laurie are back in costume.\n\n LAURIE\n Look, Dan, I'm glad you were able to\n rediscover your calling in life, but I\n think you're getting carried away. You're\n talking about springing a homicidal\n psychopath from prison.\n\n DAN\n Rorschach may be sick, but I think he's\n on to something--four heroes attacked in\n a week is not a coincidence. I mean, that\n whole cancer thing with ion, does that\n make sense to you? You didn't contract\n cancer from him.\n He says as Laurie's about to light a cigarette. She changes\n her mind.\n\n LAURIE\n Why would someone want to drive ion away?\n Look what's happening now. With Jon gone\n it looks like we're headed for another\n World War. Who would want that?\n\n (CONTINUED)\n\n\n\n\n\n\n\n\n\n 81.\n\n CONTINUED:\n OUTSIDE THE WINDOW: The Owl-Ship approaches a ROOFTOP with a\n HELIPAD. On the helipad is a HELICOPTER with the Veidt\n Enterprises LOGO.\n\n DAN\n I don't know. But I know someone who\n might be able to help.\n Dan lands the Owl-Ship on the roof."}, {"role": "assistant", "content": "## SCENE 1 — INT. OWL-SHIP (MOVING) - NIGHT\nACTION: Dan and Laurie are back in costume. Look, Dan, I'm glad you were able to\nDIALOGUE — Laurie: Look, Dan, I'm glad you were able to rediscover your calling in life, but I think you're getting carried away. You're talking about springing a homicidal...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Watchmen: Dan and Laurie are back in costume. Look, Dan, I'm glad you were able to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Dan and Laurie are back in costume. Look, Dan, I'm glad you were able to The camera pushes in with small amplitude at slow speed. Laurie (S1) says: [English] Look, Dan, I'm glad you were able to rediscover your calling in life, but I think you're getting carried away. You're talking about springing a homicidal... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Jacob'S Ladder\"\n\nSCREENPLAY EXCERPT:\nEXT. BAR - DAY\n\nThe TWO MEN leave the bar on a dingy side street. It is cold outside.\nChristmas lights seem ludicrous dangling in the bar's front window.\nPAUL looks at them and smiles.\n\n\t\t\t\tPAUL\n\t\tMerry Christmas.\n\nPAUL steps into the street and walks to the driver's side of his car.\nHe pulls out his keys and opens the door. JACOB looks down on the\nsidewalk and notices a dime.\n\n\t\t\t\tJACOB\n\t\tGoddamn, this is my lucky day.\n\nHe bends down to pick it up. PAUL inserts the key into the ignition and\nsteps on the gas. He turns the key.\n\nTHE CAR EXPLODES. Pieces of metal and flesh fly into the air. JACOB\nsprawls out flat on the ground as the debris hurls above him. He covers\nhis head."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BAR - DAY\nACTION: The TWO MEN leave the bar on a dingy side street. It is cold outside. Christmas lights seem ludicrous dangling in the bar's front window.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Jacob'S Ladder\": The TWO MEN leave the bar on a dingy side street. It is cold outside. Christmas lights seem ludicrous dangling in the bar's front window., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The TWO MEN leave the bar on a dingy side street. It is cold outside. Christmas lights seem ludicrous dangling in the bar's front window. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Avengers: Endgame\n\nSCREENPLAY EXCERPT:\nEXT. CAMP LEHIGH - DAY (1970)\n\nSOLDIERS MARCH PAST AS...\n\nTony and Steve round the corner, TONY IN A LAB COAT and STEVE\nIN GREEN KHAKIS, LOW-PULLED CAP, AND MIRRORED SUNGLASSES.\n\n STEVE\n The idea of me was.\n\n TONY\n (looking around)\n All right. If I was SHIELD and I\n wanted to hide my quasi-fascistic\n black site, where would I hide it?\n\n STEVE\n In plain sight.\n\nSteve nods at A MAN IN A GRAY SUIT approaching a building,\n\"MUNITIONS.\" He opens the door with a retro-tech PUNCH CARD.\n\nTONY TOUCHES HIS GLASSES: HE X-RAYS INTO THE GROUND,\nILLUMINATING...TWENTY FLOORS OF SECRET, UNDERGROUND BASE."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CAMP LEHIGH - DAY (1970)\nACTION: Tony and Steve round the corner, TONY IN A LAB COAT and STEVE IN GREEN KHAKIS, LOW-PULLED CAP, AND MIRRORED SUNGLASSES.\nDIALOGUE — Steve: The idea of me was.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Avengers: Endgame: Tony and Steve round the corner, TONY IN A LAB COAT and STEVE IN GREEN KHAKIS, LOW-PULLED CAP, AND MIRRORED SUNGLASSES., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Tony and Steve round the corner, TONY IN A LAB COAT and STEVE IN GREEN KHAKIS, LOW-PULLED CAP, AND MIRRORED SUNGLASSES. The camera pushes in with small amplitude at slow speed. Steve (S1) says: [English] The idea of me was. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: PULL BACK to reveal sculpted horses, knights and dragons on the table top.\nSetting: derived from Heavenly Creatures\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ILAM/JULIET'S BEDROOM - DAYPAULINE AND JULIET KNEADING AND SHAPING PLASTICINE WITH THEIR FINGERS . . .\nACTION: PULL BACK to reveal sculpted horses, knights and dragons on the table top.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Heavenly Creatures: PULL BACK to reveal sculpted horses, knights and dragons on the table top., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. PULL BACK to reveal sculpted horses, knights and dragons on the table top. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Hellboy\"\n\nSCREENPLAY EXCERPT:\nEXT. BPRD BUILDING - DAY\n\n It's raining like hell.\n\n The PALLBEARERS, Manning and Myers among them, load Broom's\n casket into a HEARSE. Flanking it are two rows of BPRD agents.\n\n The hearse doors close and the vehicle pulls away.\n\n Watching like a gargoyle from a distant roof: Hellboy. Rain\n bounces off his wet overcoat.\n\n Liz observes him, worried.\n\n LIZ'S VOICE\n He hasn't spoken to anyone in three\n days. Not a word. He doesn't eat, he\n doesn't sleep..."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BPRD BUILDING - DAY\nACTION: It's raining like hell. The PALLBEARERS, Manning and Myers among them, load Broom's\nDIALOGUE — Liz'S Voice: He hasn't spoken to anyone in three days. Not a word. He doesn't eat, he doesn't sleep...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hellboy\": It's raining like hell. The PALLBEARERS, Manning and Myers among them, load Broom's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. It's raining like hell. The PALLBEARERS, Manning and Myers among them, load Broom's The camera pushes in with small amplitude at slow speed. Liz'S Voice (S1) says: [English] He hasn't spoken to anyone in three days. Not a word. He doesn't eat, he doesn't sleep... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Hunter\"\n\nSCREENPLAY EXCERPT:\n------------------------------------------------------------------------\n REVISED - \"HUNTER\" - 4/21/86 15\n\n *\n\n SCHAEFER\n Billy's on it."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/21/86 15\nDIALOGUE — Schaefer: Billy's on it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hunter\": ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/21/86 15, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/21/86 15 The camera pushes in with small amplitude at slow speed. Schaefer (S1) says: [English] Billy's on it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: She sits on the edge of DANNY'S bed. Danny hands her his\nSetting: derived from American History X\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BEDROOM - TWO SINGLE BEDS\nACTION: She sits on the edge of DANNY'S bed. Danny hands her his assignment instructions.\nDIALOGUE — Davina: I got shit to do, Dan.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American History X: She sits on the edge of DANNY'S bed. Danny hands her his assignment instructions., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. She sits on the edge of DANNY'S bed. Danny hands her his assignment instructions. The camera pushes in with small amplitude at slow speed. Davina (S1) says: [English] I got shit to do, Dan. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Come and get it. Fucking nihilist.\nSetting: derived from The Big Lebowski\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Come and get it. Fucking nihilist. Show me what you got. Nihilist.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Big Lebowski: Come and get it. Fucking nihilist. Show me what you got. Nihilist., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Come and get it. Fucking nihilist. Show me what you got. Nihilist. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Barbie Sharon argues a case passionately in front of the\nSetting: derived from Barbie\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SUPREME COURT. DAY\nACTION: Barbie Sharon argues a case passionately in front of the Supreme Court, all Barbies.\nDIALOGUE — Barbie Sharon: Only Barbies are Barbies, and we would argue that corporations have no \"free speech\" rights to begin with, so any claim on their part to be exercising a right...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Barbie: Barbie Sharon argues a case passionately in front of the Supreme Court, all Barbies., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Barbie Sharon argues a case passionately in front of the Supreme Court, all Barbies. The camera pushes in with small amplitude at slow speed. Barbie Sharon (S1) says: [English] Only Barbies are Barbies, and we would argue that corporations have no \"free speech\" rights to begin with, so any claim on their part to be exercising a right... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The TWO MEN leave the bar on a dingy side street. It is cold outside.\nSetting: derived from \"Jacob'S Ladder\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BAR - DAY\nACTION: The TWO MEN leave the bar on a dingy side street. It is cold outside. Christmas lights seem ludicrous dangling in the bar's front window.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Jacob'S Ladder\": The TWO MEN leave the bar on a dingy side street. It is cold outside. Christmas lights seem ludicrous dangling in the bar's front window., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The TWO MEN leave the bar on a dingy side street. It is cold outside. Christmas lights seem ludicrous dangling in the bar's front window. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ...Suspect vehicle sighted on\nSetting: derived from \" T E R M I N A T O R \"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: ...Suspect vehicle sighted on Motor at Pico, southbound.\nDIALOGUE — Terminator: This is One-Four-Three. West- bound on Olympic, approaching Overland.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \" T E R M I N A T O R \": ...Suspect vehicle sighted on Motor at Pico, southbound., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ...Suspect vehicle sighted on Motor at Pico, southbound. The camera pushes in with small amplitude at slow speed. Terminator (S1) says: [English] This is One-Four-Three. West- bound on Olympic, approaching Overland. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages":[{"role":"system","content":"You write ONE MiniMax-H3 FL2VA scene beat for Backlot."},{"role":"user","content":"Premise: Close-up of a glass tipping off a table and shattering (end on broken glass).\nWrite SCENE 1 as FL2VA single shot."},{"role":"assistant","content":"## SCENE 1 — INT. TABLE EDGE - DAY\nACTION: A hand knocks a drinking glass; it falls and shatters; fragments settle into a final arrangement.\nSHOT: close shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: intact drinking glass near edge of dark wooden table, hand approaching from right, soft daylight, cinematic close-up\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a close shot begins with an intact drinking glass near the edge of a dark wooden table as in Picture 1, while a hand and sleeve approach from the right. The camera pushes in with small amplitude at slow speed as fingertips strike the rim. The glass tips, falls, and hits the floor; cracks spread and fragments slide outward, settling into the broken arrangement and final composition of Picture 2.\n\noverall_soundscape: Fingertips tap glass; it scrapes the tabletop, falls, and breaks with a sharp crash; small fragments scatter and stop.\n\nnon_diegetic_music: A low electronic pulse at a slow tempo, ending immediately after the glass breaks.\nLORA: none\nAUDIO: glass break\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: F I G H T C L U B\n\nSCREENPLAY EXCERPT:\nINT. HALLWAY - SAME\n\n Jack's leaving, looks like a war casualty, passing COWORKERS\n who coldly stare at him. His face is totally passive.\n\n JACK (V.O.)\n I got right in everyone's hostile\n little face. Yes, these are bruises\n from fighting. I'm comfortable with\n them. I am enlightened."}, {"role": "assistant", "content": "## SCENE 1 — INT. HALLWAY - SAME\nACTION: Jack's leaving, looks like a war casualty, passing COWORKERS who coldly stare at him. His face is totally passive.\nDIALOGUE — Jack: I got right in everyone's hostile little face. Yes, these are bruises from fighting. I'm comfortable with them. I am enlightened.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from F I G H T C L U B: Jack's leaving, looks like a war casualty, passing COWORKERS who coldly stare at him. His face is totally passive., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack's leaving, looks like a war casualty, passing COWORKERS who coldly stare at him. His face is totally passive. The camera pushes in with small amplitude at slow speed. Jack (S1) says: [English] I got right in everyone's hostile little face. Yes, these are bruises from fighting. I'm comfortable with them. I am enlightened. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: As Credits Roll\n\nSCREENPLAY EXCERPT:\nINT. POLICE STATION - INTERROGATION ROOM - NIGHT\nFred's POV: A BIG FIST slams into Fred's face, knocking him\nback in a chair. Ed and Al are standing in front of Fred.\n\n\t\t\t\tED\n\t\tStay put, killer!\n\nThe Detectives' POV: Fred's face is a bloody mask. He is\ncrying, whimpering.\n\n\t\t\t\tFRED\n\t\tI didn't kill her! Tell me I didn't kill\n\t\ther!\n\n\t\t\t\t\t\t\t\tCUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — INT. POLICE STATION - INTERROGATION ROOM - NIGHT\nACTION: Fred's POV: A BIG FIST slams into Fred's face, knocking him back in a chair. Ed and Al are standing in front of Fred.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from As Credits Roll: Fred's POV: A BIG FIST slams into Fred's face, knocking him back in a chair. Ed and Al are standing in front of Fred., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Fred's POV: A BIG FIST slams into Fred's face, knocking him back in a chair. Ed and Al are standing in front of Fred. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Now we got it. .So let's go back in there\nSetting: derived from Django Unchained\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BIG HOUSE\nACTION: Now we got it. .So let's go back in there and busts these motherfucker's chops.\nDIALOGUE — Lara Lee: There you are! I was beginning to think you and that of crow ran off together. Schultz and Django feel the change in their host's demeanor, as he stares down...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Django Unchained: Now we got it. .So let's go back in there and busts these motherfucker's chops., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Now we got it. .So let's go back in there and busts these motherfucker's chops. The camera pushes in with small amplitude at slow speed. Lara Lee (S1) says: [English] There you are! I was beginning to think you and that of crow ran off together. Schultz and Django feel the change in their host's demeanor, as he stares down... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: HAWKMEN and LIZARD MEN SLAVES engage MING'S PEOPLE.\nSetting: derived from Flash Gordon\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PLASMA CAVERN\nACTION: HAWKMEN and LIZARD MEN SLAVES engage MING'S PEOPLE. Aided by VULTAN and TIGER MAN, ZARKOV battles his way up to\nDIALOGUE — Zarkov: Attach the electrodes to the shovels! Pitch them into the plasma!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Flash Gordon: HAWKMEN and LIZARD MEN SLAVES engage MING'S PEOPLE. Aided by VULTAN and TIGER MAN, ZARKOV battles his way up to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. HAWKMEN and LIZARD MEN SLAVES engage MING'S PEOPLE. Aided by VULTAN and TIGER MAN, ZARKOV battles his way up to The camera pushes in with small amplitude at slow speed. Zarkov (S1) says: [English] Attach the electrodes to the shovels! Pitch them into the plasma! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Danny looks at the two.\nSetting: derived from American History X\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. BEDROOM - THE THREESOME\nACTION: Danny looks at the two. If I don't finish this thing I'm dead.\nDIALOGUE — Danny: If I don't finish this thing I'm dead.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American History X: Danny looks at the two. If I don't finish this thing I'm dead., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Danny looks at the two. If I don't finish this thing I'm dead. The camera pushes in with small amplitude at slow speed. Danny (S1) says: [English] If I don't finish this thing I'm dead. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Minority Report\"\n\nSCREENPLAY EXCERPT:\nEXT. ANDERSON'S HOUSE\n\n A Precrime HOVERCRAFT floats high above her house, a dark\n moon in the low clouds."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ANDERSON'S HOUSE\nACTION: A Precrime HOVERCRAFT floats high above her house, a dark moon in the low clouds.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Minority Report\": A Precrime HOVERCRAFT floats high above her house, a dark moon in the low clouds., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A Precrime HOVERCRAFT floats high above her house, a dark moon in the low clouds. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade In:\n\nSCREENPLAY EXCERPT:\nI can't let them kill any more children.\n Come out now.\n\n She wades through table after table, pushing them out of her\n way with the strength she has demonstrated throughout as the\n PROWLER.\n\n When MRS. VOORHEES is too quick for her, shoving a table across\n her path like a barred door. Before ALICE can think twice,\n MRS. VOORHEES has shoved a second table in behind her so that\n now ALICE is sandwiched between them.\n\n She can duck down or climb over or come out the only open end--\n the end where MRS. VOORHEES is standing.\n\n ALICE looks behind her and sees a shelf. On the shelf is a box.\n\n MRS. VOORHEES is coming down the aisle between the two tables.\n Her machete is coming up in an attack.\n\n ALICE reaches into the box in CU and pulls out a handful of\n leather-working tools. She hurls them at MRS. VOORHEES as hard\n as she can.\n\n MRS. VOORHEES has to put her arms up to ward off the spinning\n knives. One cuts her under the eye and she lets out a low\n animal bellow.\n\n ALICE leaps up, climbs over the table and runs for the door.\n\n MRS. VOORHEES shoves the tables over and hurries after her.\n\n ALICE gets to the door, shuts off the light switch and runs out.\n\n We can hear JASON'S VOICE.\n\n JASON (V.O.)\n Kill her, mommy! Kill"}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: I can't let them kill any more children. She wades through table after table, pushing them out of her\nDIALOGUE — Prowler.: When MRS. VOORHEES is too quick for her, shoving a table across her path like a barred door. Before ALICE can think twice, MRS. VOORHEES has shoved a second...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: I can't let them kill any more children. She wades through table after table, pushing them out of her, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I can't let them kill any more children. She wades through table after table, pushing them out of her The camera pushes in with small amplitude at slow speed. Prowler. (S1) says: [English] When MRS. VOORHEES is too quick for her, shoving a table across her path like a barred door. Before ALICE can think twice, MRS. VOORHEES has shoved a second... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: But... won't you be de-rezzed?\nSetting: derived from T R O N\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: But... won't you be de-rezzed?\nDIALOGUE — Yori: But... won't you be de-rezzed?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T R O N: But... won't you be de-rezzed?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. But... won't you be de-rezzed? The camera pushes in with small amplitude at slow speed. Yori (S1) says: [English] But... won't you be de-rezzed? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: F I G H T C L U B\n\nSCREENPLAY EXCERPT:\nINT. KITCHEN - MOMENTS LATER\n\n Jack sits at the table, sips coffee. He's pale, dazed,\n seems broken. He hears the faint SOUND of SAWING and\n HAMMERING, unsure of where it's coming from.\n\n Marla walks into the kitchen and goes straight to the\n counter. Her back is to Jack as he looks at her. She pours\n ccffee and lights a cigarette. A beat of silence, then:\n\n MARLA\n I'll be out of your way in a sec.\n\n She seems to be as weak as Jack.\n\n JACK\n You... don't have to... leave.\n\n MARLA\n Whatever.\n\n JACK\n Really... I mean it.\n (pause)\n Have you been going to your groups?\n\n MARLA\n Chloe's dead.\n\n JACK\n When?\n\n MARLA\n Do you care?\n\n JACK\n I don't know.\n\n MARLA\n It was the smart move on her part.\n\n Marla turns to face Jack, a grim expression on her face.\n There's a BRUISE on her ARM. Jack gets up, moves closer.\n\n JACK\n Why are we both... caught up like\n this... with... ?\n\n JACK (V.O.)\n I came so close to saying Tyler's\n name, I could feel it vibrate inside\n my mouth.\n\n Marla looks"}, {"role": "assistant", "content": "## SCENE 1 — INT. KITCHEN - MOMENTS LATER\nACTION: Jack sits at the table, sips coffee. He's pale, dazed, seems broken. He hears the faint SOUND of SAWING and\nDIALOGUE — Marla: I'll be out of your way in a sec.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from F I G H T C L U B: Jack sits at the table, sips coffee. He's pale, dazed, seems broken. He hears the faint SOUND of SAWING and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack sits at the table, sips coffee. He's pale, dazed, seems broken. He hears the faint SOUND of SAWING and The camera pushes in with small amplitude at slow speed. Marla (S1) says: [English] I'll be out of your way in a sec. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Lord Of The Rings: The Return Of The King\n\nSCREENPLAY EXCERPT:\nEXT. THEODEN'S TENT, DUNHARROW PLATEAU - DAWN\n\nANGLE ON: THEODEN looks at EOWYN who stands as if turned to\nstone ... she is looking up at a LIGHTENING SKY ... The QUIET\nbefore the MEN depart for WAR . . .\n\n THEODEN\n (gentle)\n I have left instruction: the people are to\n follow your rule in my stead ... take up my\n seat in the Golden Hall ... long may you\n defend Edoras if the battle goes ill.\n\nANGLE ON: EOWYN turns to the KING ... her FACE a BLANK ...\nShe speaks without emotion.\n\n EOWYN\n What other duty would you have me do, my\n Lord?\n\nTHEODEN moves towards her, taking up her HAND . . .\n\n THEODEN\n (quiet)\n Duty? ... No, I would have you smile again -\n not grieve for those whose time has come.\n\nEOWYN slowly lifts her head . . .\n\n THEODEN (cont'd)\n You shall live to see these days renewed.\n\nCLOSE ON: THEODEN lifts EOWYN faces to look into her EYES . . .\n\n THEODEN (cont'd)\n (whisper)\n No more despair."}, {"role": "assistant", "content": "## SCENE 1 — EXT. THEODEN'S TENT, DUNHARROW PLATEAU - DAWN\nACTION: ANGLE ON: THEODEN looks at EOWYN who stands as if turned to stone ... she is looking up at a LIGHTENING SKY ... The QUIET\nDIALOGUE — Theoden: I have left instruction: the people are to follow your rule in my stead ... take up my seat in the Golden Hall ... long may you defend Edoras if the battle...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Lord Of The Rings: The Return Of The King: ANGLE ON: THEODEN looks at EOWYN who stands as if turned to stone ... she is looking up at a LIGHTENING SKY ... The QUIET, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ANGLE ON: THEODEN looks at EOWYN who stands as if turned to stone ... she is looking up at a LIGHTENING SKY ... The QUIET The camera pushes in with small amplitude at slow speed. Theoden (S1) says: [English] I have left instruction: the people are to follow your rule in my stead ... take up my seat in the Golden Hall ... long may you defend Edoras if the battle... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BeingErica illustrating the trope 'CanadianSeries'. Being Erica\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying CanadianSeries.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BeingErica, CanadianSeries, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for CanadianSeries. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Pleasantville\n\nSCREENPLAY EXCERPT:\nEXT. PARKER HOUSE. NIGHT.\n\n\tA single light is burning in the kitchen window.\n\n\t\t\t\tBETTY (OS)\n\t\tDo you want some more cookies?"}, {"role": "assistant", "content": "## SCENE 1 — EXT. PARKER HOUSE. NIGHT.\nACTION: A single light is burning in the kitchen window. Do you want some more cookies?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Pleasantville: A single light is burning in the kitchen window. Do you want some more cookies?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A single light is burning in the kitchen window. Do you want some more cookies? The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Asteroid City\n\nSCREENPLAY EXCERPT:\nINT. THEATRE PROSCENIUM\n\nBehind the host, the lights come up on the theatre proscenium\nset -- turned around to face the audience, with painted backdrop\ndepicting the empty house: seats, boxes, balconies. A wiry,\nintense, shirtless man works onstage repainting a backdrop. He\nis the director.\n 59.\n\n\n HOST\n Actor, immigrant, former student of the\n great theatrical guru Saltzburg Keitel.\n\nThe director wanders into the wings: adjusting a lamp; resewing\nthe trim of a costume; briefly touching-up a young actor's old-\nage make-up.\n\n HOST\n Known for his limitless energy, his\n voracious enthusiasms (a well-known\n actress described him, sexually, as: \"an\n animal -- specifically a rabbit\"), and\n his long, deep, and intimate relationship\n with success.\n\nThe director arrives at a makeshift bedroom installation -- fake\nwalls, folding bunk, sink basin, toaster and hot plate -- in a\ndeep backstage corner behind crates, carts, and flats (all\nlabeled: Property of \"Asteroid City Scenic Department\" and\nvaguely recognizable as the luncheonette). In the background, a\nhulking stagehand finishes stapling lace curtains along the\nsides of a fake window. They both survey the space.\n\n DIRECTOR\n What do you think, Lunky?\n\n STAGEHAND\n (deepest basso)\n Good.\n\nThe director begins to shadow-box in the corner. Off-screen: an\nannoyed/perplexed woman's voice says:\n\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. THEATRE PROSCENIUM\nACTION: Behind the host, the lights come up on the theatre proscenium set -- turned around to face the audience, with painted backdrop\nDIALOGUE — Host: Actor, immigrant, former student of the great theatrical guru Saltzburg Keitel.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Asteroid City: Behind the host, the lights come up on the theatre proscenium set -- turned around to face the audience, with painted backdrop, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Behind the host, the lights come up on the theatre proscenium set -- turned around to face the audience, with painted backdrop The camera pushes in with small amplitude at slow speed. Host (S1) says: [English] Actor, immigrant, former student of the great theatrical guru Saltzburg Keitel. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: This sequence is filmed in 16MM COLOR, cinema verite a'la 'COPS.'\nSetting: derived from \"Natural Born Killers\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. CIRCLE K - DAY\nACTION: This sequence is filmed in 16MM COLOR, cinema verite a'la 'COPS.' The CAMERA runs behind three blue windbreaker clad COPS, as they\nDIALOGUE — Wayne: They started off robbing 7/11 type stores and gas stations and later graduated to banks and the big time.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Natural Born Killers\": This sequence is filmed in 16MM COLOR, cinema verite a'la 'COPS.' The CAMERA runs behind three blue windbreaker clad COPS, as they, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. This sequence is filmed in 16MM COLOR, cinema verite a'la 'COPS.' The CAMERA runs behind three blue windbreaker clad COPS, as they The camera pushes in with small amplitude at slow speed. Wayne (S1) says: [English] They started off robbing 7/11 type stores and gas stations and later graduated to banks and the big time. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Final Destination 2\n\nSCREENPLAY EXCERPT:\nINT. KIMBERLY'S ROOM - NIGHT\n\nThe room is dark except for one burning candle. Kimberly tosses and\nturns. The wind sends a single branch scraping against her window. A\nporch swing SQUEAKS back and forth. Adding to her anxiety.\n\nA BREEZE ENTERS, sputtering out the candle. Kimberly sits upright in\nbed and sees the WINDOW IS SHUT. Huh? She lights a match and sees the\ncandle wax dripped onto her table, forming the gooey shape of\na...180? The match goes out in her hand. The room is dark again.\n\nHeadlights fr om passing cars cast eerie shadows on her ceiling.\nShadows resembling PLANES seem to transform into a SKULL.\n\nKimberly, freaked out, looks around to see the blinking light from\nthe VCR: it flashes 18:0 a few times, then 12:00.\n\nShe can't take it anymore. She yanks the VCR cord from the wall and\nflicks on the lights, ridding the room of shadows. And anxiously\nturns on her computer."}, {"role": "assistant", "content": "## SCENE 1 — INT. KIMBERLY'S ROOM - NIGHT\nACTION: The room is dark except for one burning candle. Kimberly tosses and turns. The wind sends a single branch scraping against her window. A\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Final Destination 2: The room is dark except for one burning candle. Kimberly tosses and turns. The wind sends a single branch scraping against her window. A, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The room is dark except for one burning candle. Kimberly tosses and turns. The wind sends a single branch scraping against her window. A The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You believe it's possible to dream\nSetting: derived from A Nightmare On Elm Street\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You believe it's possible to dream about what's going to happen?\nDIALOGUE — Glen: No.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from A Nightmare On Elm Street: You believe it's possible to dream about what's going to happen?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You believe it's possible to dream about what's going to happen? The camera pushes in with small amplitude at slow speed. Glen (S1) says: [English] No. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Final Destination\"\n\nSCREENPLAY EXCERPT:\nINT. GARAGE - CLEAR RIVER'S HOUSE - NIGHT\n\na chunk of the ceiling collapses upon Clear's artworks, including the canvas incorporating a jagged piece of debris from Flight 180. The metal SLAMS to the floor, on top of...\n\nTHE CAN OF TURPENTINE\n\nwhich causes the flammable liquid to stream out, toward the driveway."}, {"role": "assistant", "content": "## SCENE 1 — INT. GARAGE - CLEAR RIVER'S HOUSE - NIGHT\nACTION: a chunk of the ceiling collapses upon Clear's artworks, including the canvas incorporating a jagged piece of debris from Flight 180. The metal SLAMS to the floor, on top of... which causes the flammable liquid to stream out, toward the driveway.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Final Destination\": a chunk of the ceiling collapses upon Clear's artworks, including the canvas incorporating a jagged piece of debris from Flight 180. The metal SLAMS to the floor, on top of... which causes the flammab, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. a chunk of the ceiling collapses upon Clear's artworks, including the canvas incorporating a jagged piece of debris from Flight 180. The metal SLAMS to the floor, on top of... which causes the flammable liquid to stream out, toward the driveway. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Hyperdrive illustrating the trope 'BolivianArmyEnding'. Both seasons. First they are about to get gunned down by the Space Force after going rogue (turns out that Chloe's father stepped in). Second Henderson gives everyone (including those keeping the air circulating and power on) the day off, leaving them with a few minutes' worth of air...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying BolivianArmyEnding.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Hyperdrive, BolivianArmyEnding, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for BolivianArmyEnding. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DoctorWho illustrating the trope 'AvengersAssemble'. Doctor Who\n \"A Good Man Goes to War\" has a sequence of the Doctor rounding up his allies for a mission. There are quick scenes of them all going about their business (nursing soldiers on a battlefield, KILLING JACK THE RIPPER, etc.), only to come home and discover a big blue box on their doorstep.\n \"Day of the Moon\" begins with 3 of the main characters being killed one-by-one by another character....\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying AvengersAssemble.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DoctorWho, AvengersAssemble, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for AvengersAssemble. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Weir gazes up at her, transfixed. He takes her hand and raises\nSetting: derived from \"Event Horizon\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. EVENT HORIZON - SECOND CONTAINMENT\nACTION: Weir gazes up at her, transfixed. He takes her hand and raises it to his face. She caresses his cheek. And reaches for his\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Event Horizon\": Weir gazes up at her, transfixed. He takes her hand and raises it to his face. She caresses his cheek. And reaches for his, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Weir gazes up at her, transfixed. He takes her hand and raises it to his face. She caresses his cheek. And reaches for his The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I'm Sam Stanley. If you ever need me.\nSetting: derived from T W I N P E A K S\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I'm Sam Stanley. If you ever need me. Thanks, Sam, for the good work. You\nDIALOGUE — Desmond: Thanks, Sam, for the good work. You have a good eye for detail.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T W I N P E A K S: I'm Sam Stanley. If you ever need me. Thanks, Sam, for the good work. You, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm Sam Stanley. If you ever need me. Thanks, Sam, for the good work. You The camera pushes in with small amplitude at slow speed. Desmond (S1) says: [English] Thanks, Sam, for the good work. You have a good eye for detail. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Flash Gordon\n\nSCREENPLAY EXCERPT:\nINT. DUNGEON\n (FLASH'S VOICE)\n\n ZARKOV, BARIN and TIGER MAN lie in chains, heads cocked at\n SOUNDS of the facas outside. THUD of a body falling, brief\n silence, then sound of a KEY IN THE LOCK.\n\n The door opens itself.\n\n FLASH'S VOICE\n It's me!\n 74."}, {"role": "assistant", "content": "## SCENE 1 — INT. DUNGEON\nACTION: ZARKOV, BARIN and TIGER MAN lie in chains, heads cocked at SOUNDS of the facas outside. THUD of a body falling, brief\nDIALOGUE — Flash'S Voice: It's me! 74.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Flash Gordon: ZARKOV, BARIN and TIGER MAN lie in chains, heads cocked at SOUNDS of the facas outside. THUD of a body falling, brief, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ZARKOV, BARIN and TIGER MAN lie in chains, heads cocked at SOUNDS of the facas outside. THUD of a body falling, brief The camera pushes in with small amplitude at slow speed. Flash'S Voice (S1) says: [English] It's me! 74. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Malignant\n\nSCREENPLAY EXCERPT:\nINT. PRECINCT - BULLPEN - NIGHT\nThe bullpen is dark, empty save for Regina and Erik sitting\nat their desks, pouring over the files they retrieved from\nDr. Weaver's house. There are several cups of coffee and\nfast food wrappers between them. Erik is deep in pile of\nfiles. Regina rubs the bridge of her nose, tired.\n REGINA\n (taps the desk)\n That's it. I'm out. I can't keep\n any of this shit straight. My brain\n feels like it's about to push\n through my eyeballs.\n ERIK\n Tell Marcus I said hello.\n REGINA\n He's gonna be pissed at you, you\n know. Tonight was our date night.\n ERIK\n I'll make it up to him.\n REGINA\n I doubt that unless you're gonna\n wear the silk panties I bought.\nErik laughs. Not gonna happen. Regina grabs her stuff.\n 38.\n\n REGINA (CONT'D)\n We both want to catch this asshole\n but you look how I feel. Don't stay\n too late.\nErik nods. She leaves.\nErik flips through the medical files. He's tired, starting\nto fade. He stands, stretching. He reaches for the coffee\npot but it's EMPTY. He makes a fresh pot. As he waits, his\nattention falls on the photo of EMILY. It's upside down.\nThe image distorted in the jaundiced light of the bullpen.\nErik turns it around, facing him. The photo looks normal.\nEmily's young face stares back at him. Erik gets an "}, {"role": "assistant", "content": "## SCENE 1 — INT. PRECINCT - BULLPEN - NIGHT\nACTION: The bullpen is dark, empty save for Regina and Erik sitting at their desks, pouring over the files they retrieved from\nDIALOGUE — Regina: That's it. I'm out. I can't keep any of this shit straight. My brain feels like it's about to push through my eyeballs.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Malignant: The bullpen is dark, empty save for Regina and Erik sitting at their desks, pouring over the files they retrieved from, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The bullpen is dark, empty save for Regina and Erik sitting at their desks, pouring over the files they retrieved from The camera pushes in with small amplitude at slow speed. Regina (S1) says: [English] That's it. I'm out. I can't keep any of this shit straight. My brain feels like it's about to push through my eyeballs. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The two vehicles rip across the hard, flat ice, bolstered by\nSetting: derived from \"The Thing\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ANTARCTICA - ICESCAPE\nACTION: The two vehicles rip across the hard, flat ice, bolstered by the added horsepower. They follow the still visible dog tracks\nDIALOGUE — The Sun: sliding across the horizon, signaling midday. The snowmobiles whoosh past. Bennings drives the one loaded with supplies. MacReady and Childs double up on the...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"The Thing\": The two vehicles rip across the hard, flat ice, bolstered by the added horsepower. They follow the still visible dog tracks, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The two vehicles rip across the hard, flat ice, bolstered by the added horsepower. They follow the still visible dog tracks The camera pushes in with small amplitude at slow speed. The Sun (S1) says: [English] sliding across the horizon, signaling midday. The snowmobiles whoosh past. Bennings drives the one loaded with supplies. MacReady and Childs double up on the... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Between the burned ruin and the canvas tent: the five teenagers\nSetting: derived from Asteroid City\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. MOTEL CABIN #7. EVENING\nACTION: Between the burned ruin and the canvas tent: the five teenagers sit together in a circle on the ground (under a pink oleander,\nDIALOGUE — Dinah: After that, the second person says the name the first one said and adds another; then the third person says both plus a new name; then the next person keeps...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Asteroid City: Between the burned ruin and the canvas tent: the five teenagers sit together in a circle on the ground (under a pink oleander,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Between the burned ruin and the canvas tent: the five teenagers sit together in a circle on the ground (under a pink oleander, The camera pushes in with small amplitude at slow speed. Dinah (S1) says: [English] After that, the second person says the name the first one said and adds another; then the third person says both plus a new name; then the next person keeps... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in JAG illustrating the trope 'FriendsRentControl'. JAG: Mac's Georgetown apartment probably costs as much for one month's rent as a JAG lawyer would earn in three months. It can't even be handwaved in any meaningful way, since it's well-established that Mac came from a poor family. (To their credit, they never really even tried.) The same arguably applies to Harm's apartment, but it has the mitigating circumstance of being located \"north of Union ...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying FriendsRentControl.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for JAG, FriendsRentControl, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for FriendsRentControl. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BabylonFive illustrating the trope 'LightbulbJoke'. Babylon 5:\n One told by Sheridan, with a somewhat black undercurrent to it given Earth's initial interactions with the Minbari:\nJohn Sheridan: How many Minbari does it take to screw in a light bulb? None. They surrender right before they finish the task and they never tell you why.\n Another one, in which Londo Mollari pokes semi-bitter fun at his own people's nostalgia for their past:\nMollari: How...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying LightbulbJoke.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BabylonFive, LightbulbJoke, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for LightbulbJoke. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Heavenly Creatures\n\nSCREENPLAY EXCERPT:\nEXT. WHARF/OCEANLINER - DAYBLACK & WHITE . . .\n\nPauline . . . screaming hysterically.\n\nJuliet on the liner . . . tears streaming down her face.\n\nWIDE SHOT . . . liner steams away from the dock.\n\nCLOSE-UP . . . Pauline, sobbing on the wharf.\n\nCLOSE-UP . . . Juliet, tearful. She slowly raises her hand and waves goodbye to Pauline.\n\nJULIET\n(Whispers) I'm sorry.\n\nPULL OUT . . . Henry and Hilda are standing on either side of Juliet. They both put a comforting arm around their distraught daughter's shoulders.\n\nCLOSE-UP . . . Pauline looking devastated.\n\nSLOWLY PULL OUT from Pauline's face . . . she stands alone, sobbing.\n\nFADE TO BLACK\n\nPAULINE\n(V.O.) (Anguished) No!\n\nSONG FADES UP. . .\n\nMario Lanza singing \"You'll Never Walk Alone.\"\n\nA SERIES OF CARDS explains what happened subsequently:\n\nIn the hours following Honora's murder, a police search of the Rieper house unearthed Pauline's dairies. This resulted in her immediate arrest for the murder her mother.\n\nJuliet was arrested and charged with murder the following day.\n\nAfter Pauline's arrest it was discovered that Honor, HERBERT\nRieper had never married. Pauline was therefore charged under her mother's maiden name of Parker.\n\nIn August 1954, a plea of insanity was rejected by the jury in the Christchurch Supreme Court trial and Pauline Parker and Juliet Hulme were found guilty of murder.\n\nToo young for the death penalty, they were sent to separate prisons to be 'Detained at Her Majesty's Pleasure.'\n\nJuliet was released in November 1959 and immediately left New Zealand to join her mother overseas.\n\nPauline was released two weeks "}, {"role": "assistant", "content": "## SCENE 1 — EXT. WHARF/OCEANLINER - DAYBLACK & WHITE . . .\nACTION: Pauline . . . screaming hysterically. Juliet on the liner . . . tears streaming down her face.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Heavenly Creatures: Pauline . . . screaming hysterically. Juliet on the liner . . . tears streaming down her face., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Pauline . . . screaming hysterically. Juliet on the liner . . . tears streaming down her face. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: He rouses slightly, gripping her tighter.\nSetting: derived from \" T E R M I N A T O R \"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PROCESS SHOT\nACTION: He rouses slightly, gripping her tighter. HIGH ANGLE - MOVING WITH BOTH VEHICLES as Sarah starts to\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \" T E R M I N A T O R \": He rouses slightly, gripping her tighter. HIGH ANGLE - MOVING WITH BOTH VEHICLES as Sarah starts to, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He rouses slightly, gripping her tighter. HIGH ANGLE - MOVING WITH BOTH VEHICLES as Sarah starts to The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Flash Gordon\n\nSCREENPLAY EXCERPT:\nEXT. FOREST - FLASH AND TIGER MAN\n (FLASH, TIGER MAN)\n\n They run through the green tangle. FLASH trips over a vine.\n TIGER MAN picks him up.\n\n TIGER MAN\n Thank you for freeing me.\n 48.\n\n FLASH\n You can speak?\n\n TIGER MAN\n For all the good it does me.\n\n SOUNDS echo: hunting horns, birdcalls of TREE MEN. They start\n running again.\n\n FLASH\n Have we a chance, Tiger Man?\n\n TIGER MAN\n Only if we climb -- out of the green\n light -- to the sky -- Prince Barin's\n Tree Men can't follow us up there!\n\n FLASH\n What do you mean?\n\n TIGER MAN\n Ming's Law -- outside his own kingdom,\n every hunter becomes the hunted!\n\n Ahead is a huge tree made for climbing: branches stick out\n like the rungs of some great green ladder. They leap onto\n the lowest.\n\n ANGLES - THE TREE - FLASH and TIGER MAN\n\n Climbing. Jack and the Beanstalk, up and up, panting, but\n there is never a top in sight. Suddenly the branches end.\n\n FLASH\n Damn. End of the line.\n\n TIGER MAN reacts to something off. Touches FLASH'S shoulder\n and points.\n\n THEIR POV - AT END OF LAST BRANCH\n\n A crazy sort of ELEVATOR is going by. It's a moving. vine\n with LEAFY BASKETS attached, a "}, {"role": "assistant", "content": "## SCENE 1 — EXT. FOREST - FLASH AND TIGER MAN\nACTION: They run through the green tangle. FLASH trips over a vine. TIGER MAN picks him up.\nDIALOGUE — Tiger Man: Thank you for freeing me. 48.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Flash Gordon: They run through the green tangle. FLASH trips over a vine. TIGER MAN picks him up., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They run through the green tangle. FLASH trips over a vine. TIGER MAN picks him up. The camera pushes in with small amplitude at slow speed. Tiger Man (S1) says: [English] Thank you for freeing me. 48. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A Precrime HOVERCRAFT floats high above her house, a dark\nSetting: derived from \"Minority Report\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ANDERSON'S HOUSE\nACTION: A Precrime HOVERCRAFT floats high above her house, a dark moon in the low clouds.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Minority Report\": A Precrime HOVERCRAFT floats high above her house, a dark moon in the low clouds., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A Precrime HOVERCRAFT floats high above her house, a dark moon in the low clouds. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Coraline\n\nSCREENPLAY EXCERPT:\nINT. LIVING ROOM - SAME\n\n\n\n Coraline goes to the little door, still ajar. She peeks\n behind and finds the WALL OF SOLID BRICKS. Yes, it was\n only a dream.\n\n\n\n CORALINE\n\n (SOFTLY)\n Huh?\n\n\n She shakes her head and pushes the door shut. The kettle\n blows from the kitchen.\n\n\n\n\n\n KITCHEN - DAY, GROUND FOG OUT WINDOWS\n\n\n Breakfast. Mel and Charlie finish their \"Go Lean\n Breakfast Twigs\" and mugs of instant coffee. Preoccupied\n with their work, they half-listen as Coraline recounts\n her dream, her breakfast untouched. FOG lies outside the\n windows.\n\n\n\n CORALINE\n It was incredibly real, Mom! Only you\n weren't really you; you were my other\n mother.\n\n\n\n MEL\n\n (DISAPPROVING)\n Buttons for eyes, huh?\n (indicates untouched food)\n Coraline, you only dreamed you ate all\n that chicken. Take your multi-vitamin, at\n least.\n\n\n\n CORALINE\n You were in the dream too, Dad. You had\n wild-looking pajamas and orange monkey\n slippers.\n\n\n\n CHARLIE\n (pretend offense)\n Orange? My monkey slippers are blue.\n\n\n He stan"}, {"role": "assistant", "content": "## SCENE 1 — INT. LIVING ROOM - SAME\nACTION: Coraline goes to the little door, still ajar. She peeks behind and finds the WALL OF SOLID BRICKS. Yes, it was\nDIALOGUE — Coraline: Huh?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Coraline: Coraline goes to the little door, still ajar. She peeks behind and finds the WALL OF SOLID BRICKS. Yes, it was, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Coraline goes to the little door, still ajar. She peeks behind and finds the WALL OF SOLID BRICKS. Yes, it was The camera pushes in with small amplitude at slow speed. Coraline (S1) says: [English] Huh? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: A Nightmare On Elm Street\n\nSCREENPLAY EXCERPT:\nWhere'd you get this snotty old\n thing?\n\n SNORES from ROD. TINA yawns, turns off the light and snuggles\n against ROD, pulling the cover gingerly over herself, too.\n\n\n15. INT. TINA'S BEDROOM. NIGHT 15.\n\n CAMERA MOVES across the room of the original nightmare to find\n NANCY alone in TINA's bed, staring at the slanting ceiling above\n the bed. Thinking. We can just hear her HEART beating. She\n sighs and turns on her side.\n\n Immediately the wall above her head turns a faint reddish hue,\n with a broad yellow smear across its center. All unseen by\n NANCY, the wall begins to pulse in exact time with her heart's\n beat.\n\n CLOSE ON NANCY'S FACE. She closes her eyes.\n\n ANGLE BACK UP ON THE CEILING JUST ABOVE HER HEAD. SOMETHING\n presses against the surface from the inside. The plaster buldges\n out as if suddenly elastic, taking the shape of the thing\n pressing from inside -- taking the shape of a man's face. The\n face opens its mouth. The knives rake through the surface.\n\n ANGLE ON NANCY -- as plaster dust snows down on her.\n\n She jerks awake, sitting bolt upright. The face retracts\n suddenly -- the wall is normal.\n\n ANGLE DOWN ON NANCY as she looks up to the ceiling, touching her\n hair and feeling the plaster dust.\n\n REVERSE IN HER POV TO THE CEILING. There are three parallel cuts\n in the plaster there. About eight inches long. As if cut by\n sharp knives. Nothing else.\n\n Back on NANCY."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Where'd you get this snotty old SNORES from ROD. TINA yawns, turns off the light and snuggles\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from A Nightmare On Elm Street: Where'd you get this snotty old SNORES from ROD. TINA yawns, turns off the light and snuggles, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Where'd you get this snotty old SNORES from ROD. TINA yawns, turns off the light and snuggles The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: B&B walk up. There's a camper in the driveway.\nSetting: derived from \"Beavis And Butt-Head Do America\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. THE ANDERSON'S DRIVEWAY - DUSK\nACTION: B&B walk up. There's a camper in the driveway. Whoa, check it out Beavis. I didn't\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Beavis And Butt-Head Do America\": B&B walk up. There's a camper in the driveway. Whoa, check it out Beavis. I didn't, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. B&B walk up. There's a camper in the driveway. Whoa, check it out Beavis. I didn't The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Neo enters nervously. Beneath his feet is a path of the\nSetting: derived from The Matrix\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SHRINE (MATRIX) - DAY\nACTION: Neo enters nervously. Beneath his feet is a path of the zodiac leading to marbled stairs that rise to a dais and\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Matrix: Neo enters nervously. Beneath his feet is a path of the zodiac leading to marbled stairs that rise to a dais and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Neo enters nervously. Beneath his feet is a path of the zodiac leading to marbled stairs that rise to a dais and The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fade Out:\n\nSCREENPLAY EXCERPT:\nExt. Public Square, Emerald City -- LS -- The Wizard and\nDorothy in the basket of balloon -- Tin Man, Scarecrow and\nLion standing on platform with them -- people of Oz grouped\nabout them -- the Wizard speaks to them as the CAMERA MOVES\nforward -- the people cheer --\n\n WIZARD\n Good people of Oz, this is positively the\n finest exhibition ever to be shown --\n (stammers)\n (MORE)\n (CONTINUED)\n\n 124.\nCONTINUED: (7)\n WIZARD (CONT'D)\n -- yes -- well -- be that as it may -- I,\n your Wizard par ardua ad alta, am about\n to embark upon a hazardous and\n technically unexplainable journey into\n the outer stratosphere.\n\nMCS -- Wizard and Dorothy in the basket -- the Wizard speaks\nto the crowd o.s. -- CAMERA PANS to left to enter the Tin Man\nand Scarecrow, then PANS right as the Wizard points to the\nLion --\n\n WIZARD\n To confer, converse, and otherwise hob-\n nob with my brother wizards. And I\n hereby decree that until what time -- if\n any -- that I return, the Scarecrow, by\n virtue of his highly superior brains,\n shall rule in my stead...assisted by the\n Tin Man, by virtue of his magnificent\n heart...and the Lion -- by virtue of his\n courage! Obey them as you would me! And -\n ah - well, that's all.\n\nCU -- Toto in Dorothy's arms -- he barks at somet"}, {"role": "assistant", "content": "## SCENE 1 — EXT. PUBLIC SQUARE, EMERALD CITY -- LS -- THE WIZARD AND\nACTION: Dorothy in the basket of balloon -- Tin Man, Scarecrow and Lion standing on platform with them -- people of Oz grouped\nDIALOGUE — Wizard: Good people of Oz, this is positively the finest exhibition ever to be shown --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade Out:: Dorothy in the basket of balloon -- Tin Man, Scarecrow and Lion standing on platform with them -- people of Oz grouped, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Dorothy in the basket of balloon -- Tin Man, Scarecrow and Lion standing on platform with them -- people of Oz grouped The camera pushes in with small amplitude at slow speed. Wizard (S1) says: [English] Good people of Oz, this is positively the finest exhibition ever to be shown -- The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: She rejoins the rest of the group. The formation is not\nSetting: derived from The Heavy Metal Movie\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. - THE \"PACIFIC PEARL\"\nACTION: She rejoins the rest of the group. The formation is not quite as tight as before. As the formation flies along,\nDIALOGUE — Cut To: LOW VIEW OF THE \"PACIFIC PEARL\"\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Heavy Metal Movie: She rejoins the rest of the group. The formation is not quite as tight as before. As the formation flies along,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. She rejoins the rest of the group. The formation is not quite as tight as before. As the formation flies along, The camera pushes in with small amplitude at slow speed. Cut To (S1) says: [English] LOW VIEW OF THE \"PACIFIC PEARL\" The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jennifer'S Body\n\nSCREENPLAY EXCERPT:\nEXT. MC CULLUM PARK - CONTINUOUS\n Chip is cutting through the park on his way to the dance.\n He whistles to himself.\n Across the park, he sees a STRAY DOG trotting toward him.\n The dog pants aggressively. CHIP tries to smile, but\n looks uncomfortable.\n A few SQUIRRELS emerge from the trees, gazing curiously\n at him."}, {"role": "assistant", "content": "## SCENE 1 — EXT. MC CULLUM PARK - CONTINUOUS\nACTION: Chip is cutting through the park on his way to the dance. He whistles to himself.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jennifer'S Body: Chip is cutting through the park on his way to the dance. He whistles to himself., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Chip is cutting through the park on his way to the dance. He whistles to himself. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Final Destination 2\n\nSCREENPLAY EXCERPT:\nEXT. GAS STATION - DAY\n\nCLOSE UP - GAS DRIPPING\n\nFrom the nozzle in Thomas' car. Thomas fills the tank, watching for\nsigns of danger. Clear's hand is poised over the gas pumps' EMERGENCY\nCUT-OFF VALVE\n\nSome SKATE RATS exit the mini -mart with a pack of Camels, followed by\nKimberly, who carries a couple of Red Bulls and hands them out.\n\n THOMAS\n New life defeats death? Follow the\n signs? Where the hell did you find\n that guy?\n\n KIMBERLY\n Yeah, I thought he was supposed to\n be helpful.\n\n CLEAR\n He was. If we can use your ability\n to see the signs, we can cheat\n Death long enough to figure out\n what \"new life\" means.\n (off their looks)\n I know, it worries me that I\n\n understand him.\n\nThomas tops off the tank. Some gas spills to the ground. Quick looks\nall around -- sure enough, one of the Skate Rats is about to strike a\nmatch.\n\nCLEAR'S HAND SLAPS the kid in the back of the head, stopping him mid -\nmotion.\n\n CLEAR\n What the fuck are you thinking?\n\nThe startled Skate Rat shrinks back from Clear's rage.\n\n SKATE RAT\n I'm thinkin', suck my junk, bee\n yatch.\n\nThomas chuckles to himself and looks at Kimberly who --\n\nSMASH CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — EXT. GAS STATION - DAY\nACTION: From the nozzle in Thomas' car. Thomas fills the tank, watching for signs of danger. Clear's hand is poised over the gas pumps' EMERGENCY\nDIALOGUE — Thomas: New life defeats death? Follow the signs? Where the hell did you find that guy?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Final Destination 2: From the nozzle in Thomas' car. Thomas fills the tank, watching for signs of danger. Clear's hand is poised over the gas pumps' EMERGENCY, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. From the nozzle in Thomas' car. Thomas fills the tank, watching for signs of danger. Clear's hand is poised over the gas pumps' EMERGENCY The camera pushes in with small amplitude at slow speed. Thomas (S1) says: [English] New life defeats death? Follow the signs? Where the hell did you find that guy? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I wanna remind everyone that the tenth\nSetting: derived from The Revenge\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LAUGHING\nACTION: I wanna remind everyone that the tenth anniversary party for Rick\nDIALOGUE — George: What?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Revenge: I wanna remind everyone that the tenth anniversary party for Rick, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I wanna remind everyone that the tenth anniversary party for Rick The camera pushes in with small amplitude at slow speed. George (S1) says: [English] What? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Event Horizon\"\n\nSCREENPLAY EXCERPT:\nINT. EVENT HORIZON - SECOND CONTAINMENT\n\n Weir gazes up at her, transfixed. He takes her hand and raises\n it to his face. She caresses his cheek. And reaches for his\n eyes..."}, {"role": "assistant", "content": "## SCENE 1 — INT. EVENT HORIZON - SECOND CONTAINMENT\nACTION: Weir gazes up at her, transfixed. He takes her hand and raises it to his face. She caresses his cheek. And reaches for his\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Event Horizon\": Weir gazes up at her, transfixed. He takes her hand and raises it to his face. She caresses his cheek. And reaches for his, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Weir gazes up at her, transfixed. He takes her hand and raises it to his face. She caresses his cheek. And reaches for his The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BabylonFive illustrating the trope 'TheNotLoveInterest'. Babylon 5: Susan Ivanova's relationship with most of the male leadership of the station can be described as a form of very affectionate professionalism. Her relationship with Sheridan, in particular, is one of the most important relationships in the show, and despite the complete lack of sexual tension and it is a complete lack the storyline makes it painfully clear that they do love each other ...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying TheNotLoveInterest.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BabylonFive, TheNotLoveInterest, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for TheNotLoveInterest. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You don't know how glad I am\nSetting: derived from Sunset Boulevard\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You don't know how glad I am Walking out on the mob. What's\nDIALOGUE — Artie: Walking out on the mob. What's the big idea?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Sunset Boulevard: You don't know how glad I am Walking out on the mob. What's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You don't know how glad I am Walking out on the mob. What's The camera pushes in with small amplitude at slow speed. Artie (S1) says: [English] Walking out on the mob. What's the big idea? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: the courtyard is being taken away by the Nothing. Because of Bastians new\nSetting: derived from The Neverending Story\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. VIEW-\nACTION: the courtyard is being taken away by the Nothing. Because of Bastians new found disbelief the Nothing has grown stronger and is now attacking the last remaining part of Fantasia.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Neverending Story: the courtyard is being taken away by the Nothing. Because of Bastians new found disbelief the Nothing has grown stronger and is now attacking the last remaining part of Fantasia., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. the courtyard is being taken away by the Nothing. Because of Bastians new found disbelief the Nothing has grown stronger and is now attacking the last remaining part of Fantasia. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Look down, look down,\nSetting: derived from Les Miserables\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: Look down, look down, Upon your fellow man!\nDIALOGUE — Beggars Chorus: Look down, look down, Upon your fellow man!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Les Miserables: Look down, look down, Upon your fellow man!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Look down, look down, Upon your fellow man! The camera pushes in with small amplitude at slow speed. Beggars Chorus (S1) says: [English] Look down, look down, Upon your fellow man! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Die Hard\"\n\nSCREENPLAY EXCERPT:\nKarl, I know you want him, but the\n police are probably on their way.\n Maybe we can convince them it was all\n a mistake, but not if they hear gunshots!\n If you lock him in he'll be neutralized\n -- now do it! Karl? Karl!\n\n Karl turns off his radio. In the light of their flashlights,\n the two other terrorists look at Karl in stunned disbelief.\n He opens the door to the elevator shaft.\n\n109-B INT. ELEVATOR SHAFT - ON MCCLANE 109-B\n\n He's OVERHEARD enough of this to realize he's in deep shit.\n He backtracks to the air shaft door, strikes a cigarette\n lighter.\n\n110- OUT OUT 110-\n117 117\n\n117-A ELEVATOR SHAFT (OPPOSITE SIDE) 117-A\n\n Karl steps off the ladder to the catwalk, his own gun held *\n ready.\n\n118 MCCLANE 118\n\n HEARING Karl's approach, McClane thinks fast, looks down at\n his narrow confines, and then at:\n\n118-A HIS WEAPON 118-A\n\n and its canvas gunsling and metal strap slides.\n\n118-B BACK TO SCENE 118-B\n\n Quickly, McClane lets out all the slack in the sling. Then,\n he BRACES the weapon across the outside opening of the air\n shaft door and lowers himself into the:\n\n119 AIR SHAFT "}, {"role": "assistant", "content": "## SCENE 1 — INT. OPPOSITE SIDE\nACTION: Karl, I know you want him, but the police are probably on their way.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Die Hard\": Karl, I know you want him, but the police are probably on their way., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Karl, I know you want him, but the police are probably on their way. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Jay and Suzanne fall backwards, as Silent Bob pops through.\nSetting: derived from Jay And Silent Bob Strike Back\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SEWER TUNNEL--SAME\nACTION: Jay and Suzanne fall backwards, as Silent Bob pops through.\nDIALOGUE — Jay: INCOMING!!!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jay And Silent Bob Strike Back: Jay and Suzanne fall backwards, as Silent Bob pops through., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jay and Suzanne fall backwards, as Silent Bob pops through. The camera pushes in with small amplitude at slow speed. Jay (S1) says: [English] INCOMING!!! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Buckaroo STILL ACCELERATING, struggling with his mighty\nSetting: derived from Buckaroo Banzai\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. JET CAR - DAY\t51\nACTION: Buckaroo STILL ACCELERATING, struggling with his mighty machine...fierce vibrations, OIL AND SMOKE filling the cockpit...as he\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Buckaroo Banzai: Buckaroo STILL ACCELERATING, struggling with his mighty machine...fierce vibrations, OIL AND SMOKE filling the cockpit...as he, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Buckaroo STILL ACCELERATING, struggling with his mighty machine...fierce vibrations, OIL AND SMOKE filling the cockpit...as he The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: They're shooting at the lights.\nSetting: derived from \"Die Hard\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: They're shooting at the lights. More shots ring out from the building going over the SWAT\nDIALOGUE — Robinson: They're going after the lights!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Die Hard\": They're shooting at the lights. More shots ring out from the building going over the SWAT, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They're shooting at the lights. More shots ring out from the building going over the SWAT The camera pushes in with small amplitude at slow speed. Robinson (S1) says: [English] They're going after the lights! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Futurama\n\nSCREENPLAY EXCERPT:\nFUTURAMA\n\n Episode 206\n\n \"BRANNIGAN, BEGIN AGAIN\"\n\n By\n\n Lewis Morton\n\n Transcribed by Dave, The Neutral Planet\n\n\n\n [Ships Rec Room. Fry and Bender play a futuristic version of\n chess. The pieces are holograms and include an Omicronian, a\n Trisolian, a Horrible Gelatinous Blob, one of Kif's people and\n one of Zoidberg's people.]\n\n\n FRY\n Hmm. Horsey to pointy guy six. Check.\n\n\n\n BENDER\n Hmm. Get him, boys!\n\n\n [Fry screams as Bender's pieces leap off the board and knock\n him to the floor and beat him up.]\n\n\n FRY\n (hoarse) Good move.\n\n\n [Opening Credits. Caption: Not Y3K Compliant.]\n\n\n [Planet Express: Lounge. Fry lies on the couch next to Bender\n and Leela leans against the wall reading Cosmopolitan. The door\n opens.]\n\n\n FARNSWORTH\n Good news, everyone. We have a mission\n to further the noble cause of intergalactic\n peace.\n\n\n BENDER\n Nope. Watching cartoons.\n\n\n FRY\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. HOARSE\nACTION: Transcribed by Dave, The Neutral Planet [Ships Rec Room. Fry and Bender play a futuristic version of\nDIALOGUE — Fry: Hmm. Horsey to pointy guy six. Check.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Futurama: Transcribed by Dave, The Neutral Planet [Ships Rec Room. Fry and Bender play a futuristic version of, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Transcribed by Dave, The Neutral Planet [Ships Rec Room. Fry and Bender play a futuristic version of The camera pushes in with small amplitude at slow speed. Fry (S1) says: [English] Hmm. Horsey to pointy guy six. Check. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Endeavour illustrating the trope 'ComfortingTheWidow'. The killer in \"Home\" kills her husband to invoke this. It doesn't work.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying ComfortingTheWidow.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Endeavour, ComfortingTheWidow, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for ComfortingTheWidow. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Joker\n\nSCREENPLAY EXCERPT:\nWe're gonna go with it.\n\nGene rubs his temples, he doesn't like this, but Murray is\nthe boss.\n\n JOKER\n Thank you Murray.\n\n MURRAY FRANKLIN\n (flashes his bemused\n smile; condescending)\n Couple rules though,-- No cussing,\n no off-color material, we do a\n clean show, okay? You'll be on\n after Dr. Sally. Someone will come\n and get you. Good?\n\nJoker nods good. Smiles back at Murray.\n\nMurray and Gene turn to go, exchanging smirks with each other\nas they walk out, making light of Joker who we see behind\nthem still standing there.\n\n JOKER\n Hey Murray,-- one small thing? When\n you bring me out, can you introduce\n me as \"The Joker\"?\n\nMurray and Gene look back at him\n\n GENE UFLAND\n What? You don't want to use your\n real name?\n\n JOKER\n Honestly, I don't even know what my\n real name is.\n\nJoker smiles, the guys can't tell if he's kidding or not.\n\n JOKER\n Besides, that's what you called me\n on the show, Murray. A joker.\n Remember?\n 102.\n\n\n MURRAY FRANKLIN\n (to Gene; trying not to\n crack up)\n Did I?\n\n GENE UFLAND\n I have no idea.\n\n MURRAY FRANKLIN"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: We're gonna go with it. Gene rubs his temples, he doesn't like this, but Murray is\nDIALOGUE — Joker: Thank you Murray.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Joker: We're gonna go with it. Gene rubs his temples, he doesn't like this, but Murray is, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We're gonna go with it. Gene rubs his temples, he doesn't like this, but Murray is The camera pushes in with small amplitude at slow speed. Joker (S1) says: [English] Thank you Murray. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: We're there Stepek, right on\nSetting: derived from The Heavy Metal Movie\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. - THE \"PACIFIC PEARL\"\nACTION: We're there Stepek, right on target ... Rabual dead ahead!\nDIALOGUE — Navigator: We're there Stepek, right on target ... Rabual dead ahead!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Heavy Metal Movie: We're there Stepek, right on target ... Rabual dead ahead!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We're there Stepek, right on target ... Rabual dead ahead! The camera pushes in with small amplitude at slow speed. Navigator (S1) says: [English] We're there Stepek, right on target ... Rabual dead ahead! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: American History X\n\nSCREENPLAY EXCERPT:\nINT. BEDROOM - TWO SINGLE BEDS\n\nShe sits on the edge of DANNY'S bed. Danny hands her his\nassignment instructions.\n\n DAVINA\nI got shit to do, Dan.\n\n DANNY\nJust read it.\n\nShe sighs and begins to read it out loud.\n\n DAVINA\n\"American History X? Take home paper\nas assigned by Principal Robert\nSweeney?\"\n (to Danny)\nWhy is he giving you homework? What\nhappened to Murray?\n\n DANNY\nAsshole turned me in.\n\nDAVINA\nWhy?\n\n DANNY\nJust read.\n\n DAVINA\n\"Describe in detail your opinion of\n-the historical event that took place\nin the early morning of October 4th,\n1993.\" What's that?\n\n DANNY\nThe night Derek wasted those guys.\n\nDavina takes a moment.\n\n DAVINA\nSweeney gave you this?\n\nDANNY\n\nYep.\n\n DAVINA\n (continuing her reading)\n\"Before and after...how has this event\nhelped or hurt your present\nperspective concerning life in\ncontemporary America. Use the\nstandard five paragraph format, blab\nblab blah--this'll be good for you..\n\n DANNY\nNooo...it'll be good for you.\n\nShe stares at him for a good three seconds.\n\n DAVINA\nEat me, Dan.\n\n DANNY\nCome on! Dick Nixon's playin' at\nCammeron's tonight. You owe me.\n\n DAVINA\nI'm not doing your homework for you!\nI gotta spreadsheet due!\n\n DANNY\nGoddammit!\n\nDavina walks to the door and looks back to her brother.\n\n DAVINA\nWhat's the matter with you? You wanna\nbe a fuckin' loser your whole life?\n\nNot receiving a response, she walks out the door.\n\n DANNY\nAsshole Sweeney"}, {"role": "assistant", "content": "## SCENE 1 — INT. BEDROOM - TWO SINGLE BEDS\nACTION: She sits on the edge of DANNY'S bed. Danny hands her his assignment instructions.\nDIALOGUE — Davina: I got shit to do, Dan.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American History X: She sits on the edge of DANNY'S bed. Danny hands her his assignment instructions., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. She sits on the edge of DANNY'S bed. Danny hands her his assignment instructions. The camera pushes in with small amplitude at slow speed. Davina (S1) says: [English] I got shit to do, Dan. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Madison steps into the hallway, looking around. Turns on the\nSetting: derived from Malignant\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HOME - DOWNSTAIRS HALLWAY - NIGHT\nACTION: Madison steps into the hallway, looking around. Turns on the light. Sees blood on the floor. Her mouth drops open,\nDIALOGUE — Madison: Derek? A DARK FIGURE steps out of the living room into the hallway. This is THE KILLER. We do not see the Killer's's face, only a menacing SILHOUETTE that's...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Malignant: Madison steps into the hallway, looking around. Turns on the light. Sees blood on the floor. Her mouth drops open,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Madison steps into the hallway, looking around. Turns on the light. Sees blood on the floor. Her mouth drops open, The camera pushes in with small amplitude at slow speed. Madison (S1) says: [English] Derek? A DARK FIGURE steps out of the living room into the hallway. This is THE KILLER. We do not see the Killer's's face, only a menacing SILHOUETTE that's... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Glue illustrating the trope 'CrowdSurfing'. Done in episode 2 to James and Rob at a woodland rave, to the former's discomfort and the latter's glee.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying CrowdSurfing.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Glue, CrowdSurfing, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for CrowdSurfing. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: So, er, thanks for dinner. It was great.\nSetting: derived from He Phone Message\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOOKS DISAPPOINTED\nACTION: So, er, thanks for dinner. It was great. Yeah. We should do this again.\nDIALOGUE — George: Yeah. We should do this again.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from He Phone Message: So, er, thanks for dinner. It was great. Yeah. We should do this again., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. So, er, thanks for dinner. It was great. Yeah. We should do this again. The camera pushes in with small amplitude at slow speed. George (S1) says: [English] Yeah. We should do this again. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Second Containment. A lone engineer finishes his check of\nSetting: derived from \"Event Horizon\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. EVENT HORIZON - SECOND CONTAINMENT (FOR VIDEO)\nACTION: Second Containment. A lone engineer finishes his check of the Core. He turns to the camera and gives a self-conscious\nDIALOGUE — Kilpack: When you get this message, God willing, we will reach the solar system of Proxima Centauri...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Event Horizon\": Second Containment. A lone engineer finishes his check of the Core. He turns to the camera and gives a self-conscious, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Second Containment. A lone engineer finishes his check of the Core. He turns to the camera and gives a self-conscious The camera pushes in with small amplitude at slow speed. Kilpack (S1) says: [English] When you get this message, God willing, we will reach the solar system of Proxima Centauri... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Fear & Loathing In Las Vegas\n\nSCREENPLAY EXCERPT:\nINT. BAKER TRUCK STOP - DAY\n\nDUKE pulls out the telegram from his top pocket.\n\n GONZO'S VOICE\n You brainless scumbag! You're\n supposed to be covering the National\n District Attorney's conference! I\n made all the reservations... rented\n a white Cadillac convertible... the\n whole thing is arranged! What the\n hell are you doing out there in the\n middle of the fucking desert?\n\n 62.\n\n\nDUKE stares at the telegram.\n\n DUKE\n Never mind. It's all a big joke.\n I'm actually sitting beside the\n pool at the Flamingo. I'm talking\n from a portable phone. Some dwarf\n brought it out from the casino. I\n have total credit! Can you grasp\n that?\n (shouts)\n Don't come anywhere near this place!\n Foreigners aren't welcome here!\n\nDUKE, breathing heavily, hangs up phone."}, {"role": "assistant", "content": "## SCENE 1 — INT. BAKER TRUCK STOP - DAY\nACTION: DUKE pulls out the telegram from his top pocket. You brainless scumbag! You're\nDIALOGUE — Gonzo'S Voice: You brainless scumbag! You're supposed to be covering the National District Attorney's conference! I made all the reservations... rented a white Cadillac...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fear & Loathing In Las Vegas: DUKE pulls out the telegram from his top pocket. You brainless scumbag! You're, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. DUKE pulls out the telegram from his top pocket. You brainless scumbag! You're The camera pushes in with small amplitude at slow speed. Gonzo'S Voice (S1) says: [English] You brainless scumbag! You're supposed to be covering the National District Attorney's conference! I made all the reservations... rented a white Cadillac... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Sandra splashes water on her face. Stares at herself in the\nSetting: derived from Legion\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. DINER BATHROOM - CONTINUOUS\nACTION: Sandra splashes water on her face. Stares at herself in the mirror. A living wreck.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Legion: Sandra splashes water on her face. Stares at herself in the mirror. A living wreck., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sandra splashes water on her face. Stares at herself in the mirror. A living wreck. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Fred is behind the wheel of his car. Behind him are the\nSetting: derived from As Credits Roll\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. DESERT HIGHWAY - DAY\nACTION: Fred is behind the wheel of his car. Behind him are the CLOSE UP ON FRED'S FACE - through the front windshield. The\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from As Credits Roll: Fred is behind the wheel of his car. Behind him are the CLOSE UP ON FRED'S FACE - through the front windshield. The, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Fred is behind the wheel of his car. Behind him are the CLOSE UP ON FRED'S FACE - through the front windshield. The The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Sponge\n\nSCREENPLAY EXCERPT:\nJERRY\n On the other hand...\n\n\n ELAINE\n What?\n\n\n GEORGE\n Thanks again for last night!\n\n\n ELAINE\n Hey, I didn't even use one.\n\n\n JERRY\n I thought you said it was imminent.\n\n\n\n ELAINE\n Yeah, it was, but then I just couldn't\n decide if he was really sponge-worthy.\n\n\n\n JERRY\n Sponge-worthy?\n\n\n ELAINE\n Yeah, Jerry, I have to conserve these\n sponges.\n\n\n JERRY\n But you like this guy, isn't that what\n the sponges are for?\n\n\n ELAINE\n Yes, yes - before they went off the\n market. I mean, now I've got to re-evaluate\n my whole screening process. I can't\n afford to waste any of 'em.\n\n\n GEORGE\n You know, you're nuts with these sponges.\n George is gettin' frustrated!\n\n\n\n\n Kramer signing in at the AIDS walk.\n\n\n KRAMER\n Uh, Cosmo Kramer?\n\n\n ORGANIZER\n Uh...o.k., you're checked in. Here's\n your AIDS ribbon.\n\n\n KRAMER\n Uh, no thanks.\n\n\n ORGANIZER\n You don't want to wear an AIDS ri"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Thanks again for last night! Hey, I didn't even use one.\nDIALOGUE — Elaine: What?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Sponge: Thanks again for last night! Hey, I didn't even use one., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Thanks again for last night! Hey, I didn't even use one. The camera pushes in with small amplitude at slow speed. Elaine (S1) says: [English] What? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Good News, Bad News\n\nSCREENPLAY EXCERPT:\nJERRY\n \"Laundry-day\" is the only exciting day\n in the live of clothes. It is...no,\n think about it: the washing machine\n is the nightclub of clothes. Ya know,\n it's dark, there's bubbles happening,\n they're all kind a dancing around in\n there...shirt grabs the underwear: \"C'mon\n babe, let's gather\". You come by, you\n open up the lid and they'll: (shows\n how clothes are acting when you open\n the lid)...Socks are the most amazing\n article of clothing. They hate their\n lives, they're in the shoes with stinky\n feet, the boring drawers...the dryer\n is their only chance to escape and they\n all know it. They knew a escape from\n the dryer. They plan it in the hamper,\n the night before: (sock's voice) \"Tomorrow,\n the dryer, I'm goin'...you wait here!\"\n The dryer-door swings open and the sock\n is waiting up against the side wall.\n They hope you don't see him and then\n he goes down the road (shows how the\n sock is going down the road). They got\n buttons sewed on their faces: join the\n puppet show...So they're showing me\n on television the detergent for getting\n out blood-stains...Is this a violent\n image to anybody? Blood-stains? I mean,\n I, come on, you got a T-shirt with"}, {"role": "assistant", "content": "## SCENE 1 — INT. SOCK'S VOICE\nACTION: \"Laundry-day\" is the only exciting day in the live of clothes. It is...no,\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Good News, Bad News: \"Laundry-day\" is the only exciting day in the live of clothes. It is...no,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. \"Laundry-day\" is the only exciting day in the live of clothes. It is...no, The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Heavy Metal Movie\n\nSCREENPLAY EXCERPT:\nEXT. - THE \"PACIFIC PEARL\"\n\n She rejoins the rest of the group. The formation is not\n quite as tight as before. As the formation flies along,\n puffs of cloud pass by, occasionally obscuring a machine.\n\n\n CUT TO\n\n\n LOW VIEW OF THE \"PACIFIC PEARL\"\n\n Under the \"Pacific Pearl's\" belly, we see the bellygunner\n suspended in his small glass pod. A red glow brightens up\n as he drags on a cigar. The rest of the formation are\n visible stretched to the side and behind as puffs of cloud\n waft past. We begin to feel as if the weather might be\n changing.\n\n\n CUT TO\n\n\n THE NOSE OF THE \"PACIFIC PEARL\"\n\n The BOMBARDIER is relaxing inside as he swigs from the bottle\n and sings raucously. From the front we look in on the pilot\n with the dead co-pilot sagging in his seat next to him.\n\n\n BELLYGUNNER\n (V.O.)\n\n Sorry about Elkhorn, skip.\n What happened?\n\n PILOT\n\n I dunno. The shell exploded and he\n collapsed. I think he took it in\n the head.\n\n\n BELLYGUNNER\n (V.O.)\n\n Are ya sure he's dead, Skipper?\n\n\n PILOT\n\n He's dead.\n\n\n BOMBARDIER\n (sings drunkenly)\n\n Oh I put my finger in a woodpecker's\n "}, {"role": "assistant", "content": "## SCENE 1 — EXT. - THE \"PACIFIC PEARL\"\nACTION: She rejoins the rest of the group. The formation is not quite as tight as before. As the formation flies along,\nDIALOGUE — Cut To: LOW VIEW OF THE \"PACIFIC PEARL\"\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Heavy Metal Movie: She rejoins the rest of the group. The formation is not quite as tight as before. As the formation flies along,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. She rejoins the rest of the group. The formation is not quite as tight as before. As the formation flies along, The camera pushes in with small amplitude at slow speed. Cut To (S1) says: [English] LOW VIEW OF THE \"PACIFIC PEARL\" The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Alex has moved closer to her house, remaining in the shadows. He checks for any passers-by. Being sure there are none, he creeps toward Ms. Lewton's car. He visually inspects the exterior. As he kicks the tires... an unmarked sedan SCREECHES up. Doors open...\nSetting: derived from \"Final Destination\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. LEWTON'S HOUSE - NIGHT\nACTION: Alex has moved closer to her house, remaining in the shadows. He checks for any passers-by. Being sure there are none, he creeps toward Ms. Lewton's car. He visually inspects the exterior. As he kicks the tires... an unmarked sedan SCREECHES up. Doors open... Startled, Alex.\nDIALOGUE — Schreck: What are you doing?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Final Destination\": Alex has moved closer to her house, remaining in the shadows. He checks for any passers-by. Being sure there are none, he creeps toward Ms. Lewton's car. He visually inspects the exterior. As he kicks, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Alex has moved closer to her house, remaining in the shadows. He checks for any passers-by. Being sure there are none, he creeps toward Ms. Lewton's car. He visually inspects the exterior. As he kicks the tires... an unmarked sedan SCREECHES up. Doors open... Startled, Alex. The camera pushes in with small amplitude at slow speed. Schreck (S1) says: [English] What are you doing? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The door is heard opening and closing-a customer leaving.\nSetting: derived from Int: Bedroom. Early-Morning Hours\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: The door is heard opening and closing-a customer leaving. How much money did you leave up\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int: Bedroom. Early-Morning Hours: The door is heard opening and closing-a customer leaving. How much money did you leave up, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The door is heard opening and closing-a customer leaving. How much money did you leave up The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Revenge\n\nSCREENPLAY EXCERPT:\nLEVITAN\n I wanna remind everyone that the tenth\n anniversary party for Rick\n\n\n Barr Properties is gonna be Wednesday afternoon at four o'clock\n in\n\n\n Lasky's Bar, Madison 48th. I want all of you to be there. This\n really\n\n\n means a lot to me. Is that Costanza over there? What are you\n doing\n\n\n here?\n\n\n GEORGE\n What?\n\n\n LEVITAN\n Am I crazy, or didn't you quit?\n\n\n GEORGE\n When?\n\n\n LEVITAN\n Friday.\n\n\n GEORGE\n Oh, what? What? That? Are you kidding?\n I didn't quit. What? You\n\n\n took that seriously?\n\n\n LEVITAN\n You mean, laughingstock? All that stuff?\n\n\n\n GEORGE\n Come on. Will you stop it.\n\n\n LEVITAN\n No brains? No ability?\n\n\n GEORGE\n Teasing.\n\n\n LEVITAN\n Okay. I want you outta here.\n\n\n GEORGE\n I don't know where you're getting this\n from. I....you're serious\n\n\n aren't you? Oh, (laughing) you see?\n\n\n LEVITAN\n You can't win. You can't beat me. That's\n why I'm here and you're\n\n\n th"}, {"role": "assistant", "content": "## SCENE 1 — INT. LAUGHING\nACTION: I wanna remind everyone that the tenth anniversary party for Rick\nDIALOGUE — George: What?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Revenge: I wanna remind everyone that the tenth anniversary party for Rick, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I wanna remind everyone that the tenth anniversary party for Rick The camera pushes in with small amplitude at slow speed. George (S1) says: [English] What? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Inglourious Basterds\n\nSCREENPLAY EXCERPT:\nSo young lady, you are to close\n your cinema tonight, and have\n a private screening me.\n Francesca interprets...\n\n GOEBBELS\n What German films do you have?\n Francesca asks..\n\n SHOSANNA\n My cinema , on German night, tends\n to show older German classics.\n Francesca interprets...\n\n GOEBBELS\n Why not my films?\n Francesca asks...\n\n\n\n\n\n\n\n\n\n 6Z.\n\n SHOSANNA\n I draw a older German audience in\n my cinema, that appreciate the\n nostalgia of the earlier time.\n Francesca interprets...\n\n GOEBBELS\n That's nonsense fraulein. Us Germans\n are looking forward, not backwards.\n That era of German cinema is dead.\n The German cinema I create, will not\n only be thee cinema of Europe. But\n the worlds only alternative to the\n degenerate Jewish influence of\n Hollywood.\n Fredrick Jumps in...\n\n FREDRICK\n Along with being a cinema owner,\n Emmanuelle is quite a formidable\n film critic.\n He chuckles, but alone.\n\n GOEBBELS\n WSo it would appear. Unfortunately\n for the fraulein, I've outlawed\n film criticism.\n Zoller, thinking fast, says;\n\n FREDRICK\n Why don't you screen \"Lucky Kids\"?\n I'm sure Emmanuelle hasn't seen. it.\n And it's so funny, I've been meaning\n to recommend it to her, for her\n German night. That'"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: So young lady, you are to close your cinema tonight, and have\nDIALOGUE — Goebbels: What German films do you have? Francesca asks..\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Inglourious Basterds: So young lady, you are to close your cinema tonight, and have, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. So young lady, you are to close your cinema tonight, and have The camera pushes in with small amplitude at slow speed. Goebbels (S1) says: [English] What German films do you have? Francesca asks.. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The top floor of Bean Annex. The room is filled with boxes of\nSetting: derived from Fantastic Mr. Fox\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ATTIC. DAY\nACTION: The top floor of Bean Annex. The room is filled with boxes of Christmas ornaments, old sports equipment, two stained\nDIALOGUE — Kristofferson: Could I have a cup of water, please? Kristofferson waits for a reply, but no one answers. He whistles to himself for a minute. He clears his throat again. He...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fantastic Mr. Fox: The top floor of Bean Annex. The room is filled with boxes of Christmas ornaments, old sports equipment, two stained, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The top floor of Bean Annex. The room is filled with boxes of Christmas ornaments, old sports equipment, two stained The camera pushes in with small amplitude at slow speed. Kristofferson (S1) says: [English] Could I have a cup of water, please? Kristofferson waits for a reply, but no one answers. He whistles to himself for a minute. He clears his throat again. He... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Mr. Eddy rotates his neck - shaking out some adrenaline\nSetting: derived from As Credits Roll\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. MR. EDDY'S 600 PULLMAN - DAY\nACTION: Mr. Eddy rotates his neck - shaking out some adrenaline remnants. Pete is visibly shaken by what he has seen.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from As Credits Roll: Mr. Eddy rotates his neck - shaking out some adrenaline remnants. Pete is visibly shaken by what he has seen., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Mr. Eddy rotates his neck - shaking out some adrenaline remnants. Pete is visibly shaken by what he has seen. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: American History X\n\nSCREENPLAY EXCERPT:\nINT. BURGER SHOP - CAMHERON AND SETH\n\nCUT TO:\n\nSeth eats a burger and chili fries at the same time.\nCameron smokes across from him and watches Seth eat like\na pig. He takes a look at the homeless man outside who\nstill lays motionless. Cam feels his BANDAGED FACE.\n\n CAMMERON\nI think I need to go to the hospital.\n\n SETH\nSeriously?\n\n CAMMERON\nI don't know. I think so.\n\n SETH\nWhere's Stacey?\n\n CAMMERON\nWho gives a fuck?\n\nSeth takes a bite and chews.\n\n SETH\nDerek's a fuckin' traitor pussy.\n\n CAMMERON\nWell...we might be pussies too if we\ngot treated the way he did.\n\n SETH\nWhat do you mean?\n\n CAMMERON\nStevie McCormick called...old school\nVenice bro. He's at Chino...doin'\nlife. He hated Derek.\n (exhaling)\nSaid he was a fucking blow-up doll in\nthere.\n\n SETH\n (overwhelmed)\nFucking A.\n\nSeth continues to eat. Cammeron looks out the window and\nhates what he sees.\n\n CAMMERON\nOh Christ. You gotta be kidding me?\n\nSETH\n\nWhat?\n\nCAMMERON'S POV. A large, mature and sharp-looking BLACK\nMAN helps A BEAUTIFUL BLOND out of a RED BMW. THEY KISS.\nCammeron is at a loss for words. Cam puts out his\ncigarette as the couple ENTERS.\n\n CAMMERON\n (looking at the two)\nIs there anything sacred in this\ncountry anymore? Jesus Christ. Hey.\nStop feeding your face for a second\nand look behind you.\n\nSeth turns his head and stares at the interracial couple.\nHe turns back to Cam and sings a verse from the famous\natevie Wonder song, EBONY AND IVORY. Camero"}, {"role": "assistant", "content": "## SCENE 1 — INT. BURGER SHOP - CAMHERON AND SETH\nACTION: Seth eats a burger and chili fries at the same time. Cameron smokes across from him and watches Seth eat like\nDIALOGUE — Cammeron: I think I need to go to the hospital.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American History X: Seth eats a burger and chili fries at the same time. Cameron smokes across from him and watches Seth eat like, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Seth eats a burger and chili fries at the same time. Cameron smokes across from him and watches Seth eat like The camera pushes in with small amplitude at slow speed. Cammeron (S1) says: [English] I think I need to go to the hospital. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Neverending Story\n\nSCREENPLAY EXCERPT:\nINT. VIEW- Ivory Tower- The Empresses court\n\n\n\nThe door to the Empress's room opens into the courtyard. Lots of creatures from Fantasia await help from the Empress. The Empress's head servant comes out of her room and speaks to them all.\n\n\n\nCarion:\n\n Friends, I know why you are all here. The nothing is destroying our world. I also know that you have come to beseech the Empress for help. But I, I have terrible news. The Empress herself has become deathly ill. There seems to be a mysterious link between her illness and the nothing. She is dying. So she cannot save us...\n\n\n\nThere is much commotion among the creatures gathered there, but he continues.\n\n\n\n But there just might be one chance. The plains people who\n\n hunt the purple buffalo have among them a great warrior and\n\nhe alone has a chance to fight the nothing and save us. He is\n\nour only hope. His name is Atreyu.\n\n\n\nCUT TO: School attic- Bastian\n\n\n\nView of Bastian. He looks up from the\n\nbook and grabs his school bag. He looks\n\nat the Indians on the front.\n\n\n\nBastian:\n\n Atreyu..\n\n\n\nBastian:\n\nThe Empress had already sent for the great warrior. When he finally style=\"mso-spacerun: yes\"> appeared on the terrace of the Ivory Tower, he carried with him the hopes of all Fantasia.\"\n\n\n\nView of a young Indian boy walking up the stairs to the old man.\n\n\n\nBastian:\n\n A little boy\n\n\n\nClose up of Atreyu as the old man speaks to him.\n\n\n\nCarion:\n\nI'm sorry, but this is not the time nor place for children.\n\nI must ask you to leave.\n\n\n\nAtreyu:\n\n If you don't want me here,\n\nyou shouldn't have s"}, {"role": "assistant", "content": "## SCENE 1 — INT. VIEW- IVORY TOWER- THE EMPRESSES COURT\nACTION: The door to the Empress's room opens into the courtyard. Lots of creatures from Fantasia await help from the Empress. The Empress's head servant comes out of her room and speaks to them all. Friends, I know why you are all here. The nothing is destroying our world. I also know.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Neverending Story: The door to the Empress's room opens into the courtyard. Lots of creatures from Fantasia await help from the Empress. The Empress's head servant comes out of her room and speaks to them all. Friends, , detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The door to the Empress's room opens into the courtyard. Lots of creatures from Fantasia await help from the Empress. The Empress's head servant comes out of her room and speaks to them all. Friends, I know why you are all here. The nothing is destroying our world. I also know. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Yes ... No ... send that back ...\nSetting: derived from 1 Ext. Cityscape Sunset 1\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Yes ... No ... send that back ... wrong department ... of course ... of\nDIALOGUE — Expediter 4: ... about these requisitions ...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from 1 Ext. Cityscape Sunset 1: Yes ... No ... send that back ... wrong department ... of course ... of, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Yes ... No ... send that back ... wrong department ... of course ... of The camera pushes in with small amplitude at slow speed. Expediter 4 (S1) says: [English] ... about these requisitions ... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: 5A 5A\nSetting: derived from Crank\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TV SCREEN\nACTION: 5A 5A It's RICKY VERONA on the SCREEN, a young, irritatingly slick\nDIALOGUE — Verona: What's shaking, douchebag? Thought I'd give you the heads up. You're dead.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Crank: 5A 5A It's RICKY VERONA on the SCREEN, a young, irritatingly slick, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. 5A 5A It's RICKY VERONA on the SCREEN, a young, irritatingly slick The camera pushes in with small amplitude at slow speed. Verona (S1) says: [English] What's shaking, douchebag? Thought I'd give you the heads up. You're dead. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Hackers\n\nSCREENPLAY EXCERPT:\nINT. GRAND CENTRAL PUBLIC PHONES.\n\n The hackers are now setting up laptops at a bank of pay\n phones. Dade is wearing a \"Pirate Eye\" eyepiece.\n\n KATE\n Now listen up, use your best viruses to buy us\n time, we have to get into Plague's file and\n copy the worm.\n\n Cereal screams.\n\n CEREAL\n Ai! Boom boom aiaiaiaiaee! Alright, that was\n a little tension breaker, that had to be done,\n alright?\n\n KATE\n Cereal.\n\n CEREAL\n Yeah?\n\n KATE\n Go fix the phones.\n\n CEREAL\n Roger.\n\n Cereal takes off to fix the phones.\n\n KATE\n Joey, take his place.\n\n JOEY\n What, me?\n\n KATE\n Take his place, man, do it. You can do it.\n\n Joey takes Cereal's place.\n\n KATE\n Ready?\n\n DADE\n Yeah.\n\n KATE\n Alright, let's boot up.\n\n The four boot up their machines. Various vanity screens\n come up on the laptops. They begin to hack. We see the\n inside of the Gibson. Viruses of all kinds begin to pour\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. GRAND CENTRAL PUBLIC PHONES.\nACTION: The hackers are now setting up laptops at a bank of pay phones. Dade is wearing a \"Pirate Eye\" eyepiece.\nDIALOGUE — Kate: Now listen up, use your best viruses to buy us time, we have to get into Plague's file and copy the worm.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hackers: The hackers are now setting up laptops at a bank of pay phones. Dade is wearing a \"Pirate Eye\" eyepiece., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The hackers are now setting up laptops at a bank of pay phones. Dade is wearing a \"Pirate Eye\" eyepiece. The camera pushes in with small amplitude at slow speed. Kate (S1) says: [English] Now listen up, use your best viruses to buy us time, we have to get into Plague's file and copy the worm. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Legion\n\nSCREENPLAY EXCERPT:\nEXT. HIGHWAY - CONTINUOUS\n\n The cruiser PUNCHES through the burning \"Paradise\" sign,\n smashing it into fiery pieces!\n The cruiser charges on down the endless highway, burning bits\n of sign swirling in its wake."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HIGHWAY - CONTINUOUS\nACTION: The cruiser PUNCHES through the burning \"Paradise\" sign, smashing it into fiery pieces!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Legion: The cruiser PUNCHES through the burning \"Paradise\" sign, smashing it into fiery pieces!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The cruiser PUNCHES through the burning \"Paradise\" sign, smashing it into fiery pieces! The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Joel and Clementine sit and eat dinner in front of the TV.\nSetting: derived from Int. Publishing House Reception Area - Day\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. JOEL'S APARTMENT - NIGHT\nACTION: Joel and Clementine sit and eat dinner in front of the TV. It's hard to make out what they're watching.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int. Publishing House Reception Area - Day: Joel and Clementine sit and eat dinner in front of the TV. It's hard to make out what they're watching., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Joel and Clementine sit and eat dinner in front of the TV. It's hard to make out what they're watching. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A deserted access street, sidewalks lined with the husks of\nSetting: derived from Batman\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. SIDE STREET - THAT MOMENT - NIGHT\nACTION: A deserted access street, sidewalks lined with the husks of stripped-down cars. MOM, DAD, and JIMMY take a deep breath\nDIALOGUE — Voice: Hey, mister. Gimme a dollar?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman: A deserted access street, sidewalks lined with the husks of stripped-down cars. MOM, DAD, and JIMMY take a deep breath, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A deserted access street, sidewalks lined with the husks of stripped-down cars. MOM, DAD, and JIMMY take a deep breath The camera pushes in with small amplitude at slow speed. Voice (S1) says: [English] Hey, mister. Gimme a dollar? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: 1 Ext. Cityscape Sunset 1\n\nSCREENPLAY EXCERPT:\nSam Lowry. Hello. This sounds insane,\n I know, but I've been dreaming about\n you. Even before I saw you you were\n in my dreams. Weird isn't it. I mean\n ... I don't know what it means ...\n but it might mean something ...\n mightn't it? I hope so. Anyway you're\n in danger and I think we should get\n out of here, now, quick! Come on!\n\n Still JILL does nothing. SAM pulls down roller blind over.\n window.\n\n SAM"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Sam Lowry. Hello. This sounds insane, I know, but I've been dreaming about\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from 1 Ext. Cityscape Sunset 1: Sam Lowry. Hello. This sounds insane, I know, but I've been dreaming about, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sam Lowry. Hello. This sounds insane, I know, but I've been dreaming about The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The PENGUIN stands over his indoor penguin pool. He's wearing\nSetting: derived from Batman 2\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PENGUIN'S LAIR - DAY\nACTION: The PENGUIN stands over his indoor penguin pool. He's wearing rubber gloves, feeding LIVE FISH from an ice chest to his\nDIALOGUE — Selina: He's just another rich idiot. The odd thing is, he didn't seem a bit concerned.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman 2: The PENGUIN stands over his indoor penguin pool. He's wearing rubber gloves, feeding LIVE FISH from an ice chest to his, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The PENGUIN stands over his indoor penguin pool. He's wearing rubber gloves, feeding LIVE FISH from an ice chest to his The camera pushes in with small amplitude at slow speed. Selina (S1) says: [English] He's just another rich idiot. The odd thing is, he didn't seem a bit concerned. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: A Nightmare On Elm Street\n\nSCREENPLAY EXCERPT:\nWIDER, ON NANCY alone in the room. She turns and looks at the\n bed. MUSIC slips in and builds. The bed has changed color.\n It's now an ash-darkened red and yellow.\n\n207. CLOSER ON NANCY from the direction of the bed. MUSIC SUDDENLY 207.\n STOPS, and the surface of the red and yellow bed gets a bump in\n its center that keeps raising, raising until it's a hump that's a\n head and shoulders, still raising until it looms over NANCY.\n\n Then FRED KRUEGER's entire shape sweeps up into the yellow and\n red mass -- and the garish head, smoking and seething, pops\n through.\n\n NEW ANGLE -- KRUEGER, a burned, sizzling black hump of a killer,\n clumps onto the floor between NANCY and the door.\n\n NANCY falls absoltely still, and her face goes through a\n strange, almost sublime transformation.\n\n NANCY"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: WIDER, ON NANCY alone in the room. She turns and looks at the bed. MUSIC slips in and builds. The bed has changed color.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from A Nightmare On Elm Street: WIDER, ON NANCY alone in the room. She turns and looks at the bed. MUSIC slips in and builds. The bed has changed color., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. WIDER, ON NANCY alone in the room. She turns and looks at the bed. MUSIC slips in and builds. The bed has changed color. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I'm okay, really. Listen, I\nSetting: derived from \" T E R M I N A T O R \"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. V.O.\nACTION: I'm okay, really. Listen, I want you to pack some stuff\nDIALOGUE — Sarah: Thanks a lot.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \" T E R M I N A T O R \": I'm okay, really. Listen, I want you to pack some stuff, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm okay, really. Listen, I want you to pack some stuff The camera pushes in with small amplitude at slow speed. Sarah (S1) says: [English] Thanks a lot. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You can have a new identity\nSetting: derived from \"Total Recall\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You can have a new identity for the duration of the trip.\nDIALOGUE — Quail: \"Intelligence Agent\"... wouldn't that be dangerous? I might attack....\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Total Recall\": You can have a new identity for the duration of the trip., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You can have a new identity for the duration of the trip. The camera pushes in with small amplitude at slow speed. Quail (S1) says: [English] \"Intelligence Agent\"... wouldn't that be dangerous? I might attack.... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Kaamelott illustrating the trope 'OtherMeAnnoysMe'. Kaamelott: A minor example, where Perceval somehow manages to clone himself with Merlin's magic. The Stinger has him introduce the clone to Arthur saying \"But for some reason, he's a real dumbass.\"\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying OtherMeAnnoysMe.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Kaamelott, OtherMeAnnoysMe, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for OtherMeAnnoysMe. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Sunset Boulevard\n\nSCREENPLAY EXCERPT:\nYou don't know how glad I am\n to see youl\n\n ARTIE\n Walking out on the mob. What's\n the big idea?\n\n GILLIS\n I'm sorry about New Year's. Would\n you believe me if I said I had\n to be with a sick friend?\n\n ARTIE\n Someone in the formal set, no\n doubt, with a ten-carat kidney\n stone.\n\n BETTY\n Stop it, Artie, will you?"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You don't know how glad I am Walking out on the mob. What's\nDIALOGUE — Artie: Walking out on the mob. What's the big idea?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Sunset Boulevard: You don't know how glad I am Walking out on the mob. What's, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You don't know how glad I am Walking out on the mob. What's The camera pushes in with small amplitude at slow speed. Artie (S1) says: [English] Walking out on the mob. What's the big idea? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Apparently I'm Thomas Wayne's son!\nSetting: derived from Joker\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: Apparently I'm Thomas Wayne's son! How could you keep that from me?\nDIALOGUE — Joker: Apparently I'm Thomas Wayne's son! How could you keep that from me?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Joker: Apparently I'm Thomas Wayne's son! How could you keep that from me?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Apparently I'm Thomas Wayne's son! How could you keep that from me? The camera pushes in with small amplitude at slow speed. Joker (S1) says: [English] Apparently I'm Thomas Wayne's son! How could you keep that from me? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Titles: \"E.T. - The Extra-Terrestial\"\n\nSCREENPLAY EXCERPT:\nTITLES: \"E.T. - THE EXTRA-TERRESTIAL\"\n\n[The letters are in soft-purple against a black background. Purple is\ntraditionally the color of that which is sacred.]\n\nEXT: NIGHT SKY: NIGHT\n\nThe black screen becomes a night sky. The camera angles lowers to show a\nforest against the night sky.\n\nEXT: LANDING SITE: NIGHT\n\nIn an opening in the forest stands a spacecraft. The view of the craft is\nobscured by tree branches. The atmosphere is misty, with blue lights coming\nfrom the spacecraft.\n\n[The opening scene is misty and diffused. This forces the audience to pay\nclose attention to the images on the screen. The characters are not clearly\nseen. This engages the audience, as they attempt to see what the aliens\nreally look like.]\n\nOne creature walks up the gang blank and into the ship.\n\nINSERT: ALIEN HAND\n\nA strange hand, with two long and slender fingers protruding, move aside a\nbranch that obstructs the view.\n\n[This concentrates the audience's attention. The creature going into the\nship is being observed by another creature. Who are they? What's going on?\nThis is another technique that forces the audience to focus on the action.]\n\nINT: SPACECRAFT'S GREENHOUSE: NIGHT\n\nThe inside of the ship appears to be a greenhouse. There are sounds of\nwater dripping. Cone shaped objects (possibly alien plants) sit among earth\nplants. Vapors flow up from the plants.\n\n[These images all appear non-threatening. The aliens are inferred to be\ncollecting vegetation, and are thereby inferred to be harmless.]\n\n[Like many of Spielberg's other films, the opening sequences contain almost\nno dialogue. The story i"}, {"role": "assistant", "content": "## SCENE 1 — INT. POSSIBLY ALIEN PLANTS\nACTION: [The letters are in soft-purple against a black background. Purple is traditionally the color of that which is sacred.]\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Titles: \"E.T. - The Extra-Terrestial\": [The letters are in soft-purple against a black background. Purple is traditionally the color of that which is sacred.], detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. [The letters are in soft-purple against a black background. Purple is traditionally the color of that which is sacred.] The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DowntonAbbey illustrating the trope 'SarcasticClapping'. During the 2013 National Television Awards, Martin Freeman did a hilarious version of this when Downton Abbey beat Sherlock (in which he stars) for Best Drama. He wasn't even clapping, he was airclapping!\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying SarcasticClapping.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DowntonAbbey, SarcasticClapping, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for SarcasticClapping. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Hannibal illustrating the trope 'EroticEating'. During the Ko No Mono episode of Hannibal, there is a scene where Hannibal Lecter and Will Graham are eating whole roasted ortolan birds. The entire meal is shot as erotically as possible to imply mutual seduction, complete with extreme close-ups of tiny birds between open lips, swallowing throats, and an orgasmic haze.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying EroticEating.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Hannibal, EroticEating, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for EroticEating. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jurassic Park\n\nSCREENPLAY EXCERPT:\nYes, yes, yes.\n\nIn the corner, Hammond lifts his head and Ellie gently helps him to a\nsitting position.\n\n HAMMOND\n Do you mean those raptor\ns could be loose?\n\n ARNOLD\n I'm afraid so.\n\n MULDOON\n No, no way. Last time, it took 'em two\n days to chew through that mesh. But give\n 'em enough time, they'll get out. We\n better get that power on now.\n\n HAMMOND\n Regardless, check the pit. Make sure\n there' five of them there.\n\n\nEXT VISITOR'S CENTER - AFTERNOON\n\nMuldoon and Arnold run across the lawn. Muldoon carries his rocket\nlauncher at a ready position. In the distance, the alarm still SOUNDS,\nlights FLASH a continued warning.\n\nThey reach the maintenance building where the generator is housed. Its\nfamiliar him has been replaced by its own alarms. The closer they\napproach, the louder and brighter these alarms are. Grey steam pours\nout of the vents of the building, obscuring everything.\n\n ARNOLD\n What's that steam coming out of my\n generator room?\n\n\n MULDOON\n Relax, it's just refrigerated air.\n There's a leaky cooling pipe. I fixed it\n a couple times before but the original\n installation was at too sharp an angle.\n\nArnold accepts this. He is about to enter when Muldoon instructs him.\n\n MULDOON\n Just go and turn on your generator, then\n get right back to t"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: In the corner, Hammond lifts his head and Ellie gently helps him to a Do you mean those raptor\nDIALOGUE — Hammond: Do you mean those raptor s could be loose?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jurassic Park: In the corner, Hammond lifts his head and Ellie gently helps him to a Do you mean those raptor, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. In the corner, Hammond lifts his head and Ellie gently helps him to a Do you mean those raptor The camera pushes in with small amplitude at slow speed. Hammond (S1) says: [English] Do you mean those raptor s could be loose? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The Cruiser spins out of control, swerving off the road and\nSetting: derived from Legion\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. RED ROCK PARK - CONTINUOUS\nACTION: The Cruiser spins out of control, swerving off the road and careening into a large rock formation.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Legion: The Cruiser spins out of control, swerving off the road and careening into a large rock formation., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The Cruiser spins out of control, swerving off the road and careening into a large rock formation. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jurassic Park\n\nSCREENPLAY EXCERPT:\n'Fraid of heights.\n\nTimmy climbs tentatively. A dinosaur ROARS in the distance.\n\nCLOSE ON - Timmy's foot loses its grip.\n\nTimmy hangs for a moment, held only by his straining hands.\n\nAll three stop. Wait. Timmy regains his balance. Slowly, they proceed\nup and over.\n\nOn the other side, Grant looks around. More high trees. Beyond a field.\n\n GRANT\n A herbivore grazing area. Excellent.\n\nGrant starts to walk toward it. Timmy stops. Lets sits on a tree root.\n\n TIM\n We gonna walk all night?\n\nGrant looks at his watch.\n\n GRANT\n Twelve hours before that ship reaches the\n mainland.\n\n TIM\n We need sleep.\n\nTrue, and Grant's thinking the same thing. He looks aro\nund, surveying\nthe misty forest. He looks straight up into the beautiful, tall trees\nthat ring the wide-open grassy meadow.\n\n GRANT\n Up there.\n\n TIM\n Climb? Again? I can't climb again?\n\nGrant smiles.\n\n GRANT\n Sure you can.\n\nGrant checks out and rejects tree after tree. Near the meadow he finds\nit. A good, basic climbing tree.\n\n\nEXT UPPER BRANCHES OF THE TREE\n\nGrant helps Lex get comfortable on his lap. Timmy tucks himself in\nbetween Grant's outstretched legs. Grant looks out. Back to kids.\n\nTimmy and Lex are already snoring. Lex turns and hugs Grant close to\nher. She murmurs in her sleep. Grant starts to pat her shoulder when\nTimmy, without waking, re"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Timmy climbs tentatively. A dinosaur ROARS in the distance. CLOSE ON - Timmy's foot loses its grip.\nDIALOGUE — Grant: A herbivore grazing area. Excellent.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jurassic Park: Timmy climbs tentatively. A dinosaur ROARS in the distance. CLOSE ON - Timmy's foot loses its grip., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Timmy climbs tentatively. A dinosaur ROARS in the distance. CLOSE ON - Timmy's foot loses its grip. The camera pushes in with small amplitude at slow speed. Grant (S1) says: [English] A herbivore grazing area. Excellent. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in FlandersCompany illustrating the trope 'ReadingsBlewUpTheScale'. Caleb builds a \"Millstone Detector\" to alleviate a bit Hippolyte's work. That's when Maxence and Déborah show up; the detector quite naturally explodes.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying ReadingsBlewUpTheScale.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for FlandersCompany, ReadingsBlewUpTheScale, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for ReadingsBlewUpTheScale. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in Charmed1998 illustrating the trope 'DefectorFromDecadence'. Charmed: Billie thought she was this when Christy turned her against the Charmed Ones.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying DefectorFromDecadence.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for Charmed1998, DefectorFromDecadence, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for DefectorFromDecadence. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: He moves away into the next room. She hesitates a\nSetting: derived from The Bodyguard\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: He moves away into the next room. She hesitates a moment, then follows.\nDIALOGUE — Frank: No, it isn't. He moves away into the next room. She hesitates a moment, then follows.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Bodyguard: He moves away into the next room. She hesitates a moment, then follows., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He moves away into the next room. She hesitates a moment, then follows. The camera pushes in with small amplitude at slow speed. Frank (S1) says: [English] No, it isn't. He moves away into the next room. She hesitates a moment, then follows. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jennifer'S Body\n\nSCREENPLAY EXCERPT:\nINT. MC CULLUM PARK - CONTINUOUS\n Needy runs into the park and scans the dark horizon. Her\n bare feet are bloody from running and her face is shiny\n with sweat.\n NEEDY\n (YELLING)\n Chip!\n A terrified SCREAM in the distance.\n Needy whips around, hikes up her skirt and starts running\n toward Murphy Pool.\n\n She reaches the fence and starts climbing it. She slips\n and cuts her hand on a broken link.\n CHIP screams again on the other side of the fence.\n\n\n (CONTINUED)\n\n 101.\n CONTINUED:\n\n NEEDY (CONT'D)\n I'm coming!\n She flings herself over the top of the fence, her body\n slamming against the concrete pool deck. Her ARM snaps.\n It's a compound fracture, the bone thrusting through the\n flesh of her arm as she screams."}, {"role": "assistant", "content": "## SCENE 1 — INT. MC CULLUM PARK - CONTINUOUS\nACTION: Needy runs into the park and scans the dark horizon. Her bare feet are bloody from running and her face is shiny\nDIALOGUE — Needy: Chip! A terrified SCREAM in the distance. Needy whips around, hikes up her skirt and starts running toward Murphy Pool.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jennifer'S Body: Needy runs into the park and scans the dark horizon. Her bare feet are bloody from running and her face is shiny, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Needy runs into the park and scans the dark horizon. Her bare feet are bloody from running and her face is shiny The camera pushes in with small amplitude at slow speed. Needy (S1) says: [English] Chip! A terrified SCREAM in the distance. Needy whips around, hikes up her skirt and starts running toward Murphy Pool. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Untitled\n\nSCREENPLAY EXCERPT:\nThis is gonna hurt. But every pok�mon is fighting each other,\nyou're unfair.\n\nCopy Meowth\nMeow meow...\n\nMeowth\nWhat? I'm more unfair? It's because I'm not fighting with you?\nYour nails would hurt more wouldn't they?\n\nCopy Meowth\nmeow meow meow\n\nMeowth\nYou said tonight's moon would be round? You're right, it must be\na full moon.\n\nCopy Meowth\nmeow meow\n\nMeowth\nTalking about the moon at a time like this is tasteful. Its\nphilosophical.\n\t(Meanwhile Mew and Mewtwo are still bouncing around\nfighting. And Ash finally makes it back to the ground.)\n\nAsh"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: This is gonna hurt. But every pok�mon is fighting each other, What? I'm more unfair? It's because I'm not fighting with you?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: This is gonna hurt. But every pok�mon is fighting each other, What? I'm more unfair? It's because I'm not fighting with you?, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. This is gonna hurt. But every pok�mon is fighting each other, What? I'm more unfair? It's because I'm not fighting with you? The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ------------------------------------------------------------------------\nSetting: derived from \"Hunter\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/17/86 81\nDIALOGUE — Schaefer: Get back, into the rocks!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hunter\": ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/17/86 81, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ------------------------------------------------------------------------ REVISED - \"HUNTER\" - 4/17/86 81 The camera pushes in with small amplitude at slow speed. Schaefer (S1) says: [English] Get back, into the rocks! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Don't forget: It's Johnny Horne's\nSetting: derived from T W I N P E A K S\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Don't forget: It's Johnny Horne's 167. INT. LAURA'S ROOM - MINUTES LATER\nDIALOGUE — Laura: I'm nearly out.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T W I N P E A K S: Don't forget: It's Johnny Horne's 167. INT. LAURA'S ROOM - MINUTES LATER, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Don't forget: It's Johnny Horne's 167. INT. LAURA'S ROOM - MINUTES LATER The camera pushes in with small amplitude at slow speed. Laura (S1) says: [English] I'm nearly out. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Big Lebowski\n\nSCREENPLAY EXCERPT:\nCome and get it.\n\n\t\t\t\tDIETER\n\t\tVEE FUCK YOU UP, MAN!\n\n\t\t\t\tWALTER\n\t\tCome and get it. Fucking nihilist.\n\n\t\t\t\tDIETER\n\t\tI FUCK YOU! I FUCK YOU!\n\n\t\t\t\tWALTER\n\t\tShow me what you got. Nihilist.\n\t\tDipshit with a nine-toed woman.\n\nIn a rage, Dieter charges.\n\n\t\t\t\tDIETER\n\t\tI FUCK YOU! I FUCK YOU!\n\nWALTER\n\nhurls his leather satchel.\n\nKIEFFER\n\nWatching Dieter's charge, is caught off-guard. The bowling\nball thuds into his chest and lifts him off his feet.\n\nHe falls back, his uzi clattering away.\n\nWALTER\n\ntwists away as Dieter reaches him; grabs Dieter's head in\nboth hands; draws Dieter's head up to his mouth, which closes\non Dieter's ear.\n\nDUDE\n\nHe rushes Franz but draws up short as Franz sends out karate\nkicks, his leather pants squeaking and popping. Franz gives\na loud cry with each kick; the Dude leans back, throwing his\narms up, evading the kicks.\n\nWALTER\n\nHis jaw is still clamped on Dieter's ear. Dieter draws his\nsaber against Walter's side, drawing blood.\n\nWalter doesn't react to the wound. Growling as Dieter\nscreams, he worries his ear, waggling his head with his jaws\nclamped.\n\nTHE SABER\n\nDieter drops it.\n\nDUDE\n\nAwkwardly circling, evading Franz's kicks.\n\nWALTER\n\nstill worrying the ear. With a tearing sound his head and\nDieter's separate.\n\nDIETER, EARLESS, SCREAMS:\n\n\t\t\t\tDIETER\n\t\tI FUCK YOU! YOU CANNOT HURT ME! I\n\t\tBELIEF IN NUSSING!\n\nWalter spits his ear into his face.\n\nDUDE\n\nThe Dude and Franz, both now panting heavily, have yet to\nestablish body contact. Franz continues to kick.\n\n\t\t\t\tFRANZ\n\t\tVEAKLING!\n\nWALTER\n\ndraws back his fist.\n\n\t\t\t\tDIETER\n\t\tNUSSING!\n\n\t\t\t\tWALTER\n\t\t"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Come and get it. Fucking nihilist. Show me what you got. Nihilist.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Big Lebowski: Come and get it. Fucking nihilist. Show me what you got. Nihilist., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Come and get it. Fucking nihilist. Show me what you got. Nihilist. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Heavy Metal Movie\n\nSCREENPLAY EXCERPT:\nEXT. POLICE STATION\n\n HARRY, carrying the GIRL, comes out of the police station, as\n a patrol car screeches to a halt in front of him. Two cops\n emerge, dragging a green alien with four arms.\n\n\n GREEN ALIEN\n\n I tell you I'm an American citizen.\n I just lost my papers.\n\n\n COP\n (to himself)\n\n Goddamn illegal aliens.\n\n\n HARRY heads toward his cab.\n\n\n HARRY\n (V.O.)\n\n So here I was, stuck with this beautiful\n girl. I knew she was gonna be nuthin'\n but trouble ...\n\n\n He looks around at the sleazy neighbourhood. Every doorway\n and alley seems to have a rapist or thug lurking.\n\n\n HARRY\n (V.O.)\n\n ... but I didn't feel right just\n leaving her for the sickos. I\n dunno, maybe I'm gettin' sentimental\n or something ...\n\n\n DISSOLVE TO"}, {"role": "assistant", "content": "## SCENE 1 — EXT. POLICE STATION\nACTION: HARRY, carrying the GIRL, comes out of the police station, as a patrol car screeches to a halt in front of him. Two cops\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Heavy Metal Movie: HARRY, carrying the GIRL, comes out of the police station, as a patrol car screeches to a halt in front of him. Two cops, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. HARRY, carrying the GIRL, comes out of the police station, as a patrol car screeches to a halt in front of him. Two cops The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ------------------------------------------------------------------------\nSetting: derived from \"Hunter\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: ------------------------------------------------------------------------ 120 CONTINUED: 120\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hunter\": ------------------------------------------------------------------------ 120 CONTINUED: 120, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ------------------------------------------------------------------------ 120 CONTINUED: 120 The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: T W I N P E A K S\n\nSCREENPLAY EXCERPT:\nDon't forget: It's Johnny Horne's\n birthday today.\n\n167. INT. LAURA'S ROOM - MINUTES LATER\n\n Laura's hands are shaking as she does a line of cocaine. It is the\n last stuff she has -- she puts the empty packet into her diary.\n\n168. INT. HIGH SCHOOL - DAY\n\n Laura corners Bobby in a hallway.\n\n LAURA\n I'm nearly out.\n\n BOBBY\n It's taken care of, babe. You and I are\n going to make a big score tonight."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Don't forget: It's Johnny Horne's 167. INT. LAURA'S ROOM - MINUTES LATER\nDIALOGUE — Laura: I'm nearly out.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T W I N P E A K S: Don't forget: It's Johnny Horne's 167. INT. LAURA'S ROOM - MINUTES LATER, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Don't forget: It's Johnny Horne's 167. INT. LAURA'S ROOM - MINUTES LATER The camera pushes in with small amplitude at slow speed. Laura (S1) says: [English] I'm nearly out. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages":[{"role":"system","content":"You write ONE MiniMax-H3 FL2VA scene beat for Backlot."},{"role":"user","content":"Premise: A rain-soaked cyclist opens a black umbrella beside a silver bicycle.\nWrite SCENE 1 (single continuous FL2VA path)."},{"role":"assistant","content":"## SCENE 1 — EXT. WET STREET - DAY\nACTION: A cyclist raises and opens a black umbrella, stepping under it as the camera pulls out slightly.\nSHOT: medium shot, pull out with small amplitude at slow speed\nSTORYBOARD_PROMPT: rain-soaked cyclist with closed black umbrella beside silver bicycle, wet street, overcast daylight, cinematic\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a rain-soaked cyclist begins in the position and framing established by Picture 1, holding a closed black umbrella beside a silver bicycle. The camera pulls out with small amplitude at slow speed as she releases the bicycle handle, raises the umbrella, and presses the runner upward until the canopy opens. Water rolls from the expanding fabric while she steps beneath it and settles into the pose and composition established by Picture 2.\n\noverall_soundscape: Rain falls steadily on pavement, followed by the metallic click of the umbrella runner and the soft snap of the canopy opening. Water drips from the bicycle frame; distant traffic passes.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: rain, umbrella snap\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Django Unchained\n\nSCREENPLAY EXCERPT:\nYou know I'm not on the manifest? All of you\n know I'm not suppose to be on this trip.\n But those four men, are still back there.\n They're still wanted. And that eleven thousand\n five hundred is still up for grabs. And the\n last thing they'd expect is y'all rid.in back\n and gittin it.\n Django is damn convincing.\n\n ROY\n What's your deal? You tell us who they are\n and we let ya go?\n\n DJANGO\n I ain't tellin' who they are. But, you give\n me a pistol, and a horse, and five hundred\n dollars of that eleven thousand five hundred,\n and I'll point 'em out to you.\n He's got these greedy sonsabitches right on the hook...he just needs\n one little push.\n\n\n\n\n\n\n\n\n\n 1Q7\n\n DJANGO\n Y'all wanna ask somebody if I'm tellin the\n truth, ask them mandingos. You can't put\n me in the same cage with them without them\n killin me. Why ya think that is? Ask them\n .am I a Candyland slave, or did I ride in\n there on a horse, with a white man, yesterday?\n\n CUT TO\n\n ROY AND FLOYD\n go to the Cage Wagon to talk with the three mandingos. As they walk,\n Roy continues to study the handbill.\n\n ROY"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You know I'm not on the manifest? All of you know I'm not suppose to be on this trip.\nDIALOGUE — Roy: What's your deal? You tell us who they are and we let ya go?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Django Unchained: You know I'm not on the manifest? All of you know I'm not suppose to be on this trip., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You know I'm not on the manifest? All of you know I'm not suppose to be on this trip. The camera pushes in with small amplitude at slow speed. Roy (S1) says: [English] What's your deal? You tell us who they are and we let ya go? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Hellboy thrusts his hand into the SECOND IMPRINT. CLACK! He\nSetting: derived from \"Hellboy\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CATACOMBS - NIGHT\nACTION: Hellboy thrusts his hand into the SECOND IMPRINT. CLACK! He Myers spots the broken rosary and reaches for it --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hellboy\": Hellboy thrusts his hand into the SECOND IMPRINT. CLACK! He Myers spots the broken rosary and reaches for it --, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hellboy thrusts his hand into the SECOND IMPRINT. CLACK! He Myers spots the broken rosary and reaches for it -- The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A1 TONY carries the TESSERACT CASE through THE LOBBY,\nSetting: derived from Avengers: Endgame\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. STARK TOWER, LOBBY - DAY (AVENGERS 1)\nACTION: A1 TONY carries the TESSERACT CASE through THE LOBBY, followed by A1 THOR and LOKI.\nDIALOGUE — Tony: Okay, Thumbelina, you're a go.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Avengers: Endgame: A1 TONY carries the TESSERACT CASE through THE LOBBY, followed by A1 THOR and LOKI., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A1 TONY carries the TESSERACT CASE through THE LOBBY, followed by A1 THOR and LOKI. The camera pushes in with small amplitude at slow speed. Tony (S1) says: [English] Okay, Thumbelina, you're a go. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Untitled\n\nSCREENPLAY EXCERPT:\n--------------------------------------------------------------------\nAt first we see an eye. Closed and slowly it opens up and looks\naround. Later we get a full view of a man lying on the floor\nin a strange-looking room. 4 walls, 1 ceiling, 1 floor. 6 sides\njust like a cube. There is a kind of door on each side. Each\ndoor has a handle. On the walls are weird-looking pictures.\nThey doesn't seem to look like anything seen before. The man\nstands up in the middle of the room. The camera pans around him.\nThen he makes some sounds and opens a door in one of the walls.\nAs he looks in to the open door, he sees a new room like the one\nhe is located in, but the light in the new room is blue. He\nturns around and the door closes by itself. Then he opens the\ndoor in in the floor. Under that door, there's a new room\nexcatly like the first, but in this room the light is red.\nThe door closes. He opens a new door in one of the walls.\nThe light in there is red. He looks back and then crawls into\nthe new room. As he lands on the floor in the new room it makes\na loud sound, and then the door closes behind him. He takes one\nstep, and then there is a loud sound. The man stands still.\nSlowly he begins to bleed, and slowly his body begins to fall\napart. As the last parts of his body hits the floor, we see\na sort of trelliswork. First it moves to the floor, and\nafterwards it is being folded up.\n--------------------------------------------------------------------\n\n--------------------------------------------------------------------\nTITLE AND INTRO\n----------------------------------------------------"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: -------------------------------------------------------------------- At first we see an eye. Closed and slowly it opens up and looks\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: -------------------------------------------------------------------- At first we see an eye. Closed and slowly it opens up and looks, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. -------------------------------------------------------------------- At first we see an eye. Closed and slowly it opens up and looks The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Saw\n\nSCREENPLAY EXCERPT:\nAll right, that�s it! (He leans over\n and picks up his saw, sits down and\n starts trying to saw through his chains\n again, grunting. Adam, in the meantime,\n just sits back against the wall, breathing\n heavily, looking dazed. Lawrence continues\n to franticly saw, but it doesn�t work.\n He gets pissed off and throws the blade\n down.) Ah, Damn it!!\n\n Back to Adam, still looking dazed. Suddenly, a look of realization\n dawns upon his face. The electrocution has jogged his memory.\n\n\n ADAM\n (wide-eyed) I remember everything now.\n (Lawrence glances over at him.) I remember\n how I got here.\n\n Flashback to Adam walking up the stairs in his apartment building,\n a camera around his neck, smoking a cigarette and listening to\n music on his headphones. He walks with an uncaring, slow swagger.\n He gets to the top of the stairs and walks down a red-lit hallway\n to his apartment. He goes inside, closing the door behind him\n then moving into the kitchen area, throwing his keys onto the\n table and opening the fridge. The entire apartment is very rundown,\n decrepit, dingy and old; truly a shithole apartment as Adam described\n it earlier. There is gra"}, {"role": "assistant", "content": "## SCENE 1 — INT. WIDE-EYED\nACTION: All right, that�s it! (He leans over and picks up his saw, sits down and\nDIALOGUE — Adam: (wide-eyed) I remember everything now. (Lawrence glances over at him.) I remember how I got here.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Saw: All right, that�s it! (He leans over and picks up his saw, sits down and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. All right, that�s it! (He leans over and picks up his saw, sits down and The camera pushes in with small amplitude at slow speed. Adam (S1) says: [English] (wide-eyed) I remember everything now. (Lawrence glances over at him.) I remember how I got here. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Buckaroo Banzai\n\nSCREENPLAY EXCERPT:\nINT. THE PIT - NIGHT\t376\n\nThe Secretary of Defense still on the floor, spotting the OVERTHRUSTER\nin Penny's purse, taking it just as Buckaroo and New Jersey rush into\nthe room...lights swaying...shadows dancing...\n\n...the Secretary struggling to his feet as Buckaroo cuts Penny\ndown...spraying her with a white insecticide, New Jersey checking her\nvital signs, Buckaroo shocking New Jersey accidentally with a jolt of\nSTATIC ELECTRICITY, as the Secretary slips the OVERTHRUSTER under his\ncoat...\n\nSECRETARY OF DEFENSE\n(to no one in particular)\nIt's not over yet...I'll bounce back...always have. I'll come through\nthis thing smelling like a goddamn rose-just watch me.\n\nThe Secretary limping out of the room...New Jersey trying to get\npenny's pulse while Buckaroo peels back her eyelids...\n\nNEW JERSEY\nLeave her to me. You take care of business.\n\nBUCKAROO BANZAI\nThanks.\n\nBuckaroo looking at New Jersey, knowing he can trust him, rummaging\nthrough Penny's purse...no Overthruster..."}, {"role": "assistant", "content": "## SCENE 1 — INT. THE PIT - NIGHT\t376\nACTION: The Secretary of Defense still on the floor, spotting the OVERTHRUSTER in Penny's purse, taking it just as Buckaroo and New Jersey rush into\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Buckaroo Banzai: The Secretary of Defense still on the floor, spotting the OVERTHRUSTER in Penny's purse, taking it just as Buckaroo and New Jersey rush into, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The Secretary of Defense still on the floor, spotting the OVERTHRUSTER in Penny's purse, taking it just as Buckaroo and New Jersey rush into The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Okay Dude. Have it your way.\nSetting: derived from The Big Lebowski\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Okay Dude. Have it your way. The Dude unclips the phone from the battery pack.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Big Lebowski: Okay Dude. Have it your way. The Dude unclips the phone from the battery pack., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Okay Dude. Have it your way. The Dude unclips the phone from the battery pack. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Sunset Boulevard\n\nSCREENPLAY EXCERPT:\nGive me a pack of those Turkish\n cigarettes -- Melachrinos.\n\n The girl opens the glass showcase to locate the fancy\n brand. From OFF comes\n\n ARTIE'S VOICE\n Stick 'em up, Gillis, or I'll\n let you have it!\n\n Gillis turns.\n\n\n D-5 AT THE SODA FOUNTAIN\n\n Artie Green and Betty Schaefer sit having a sandwich\n and a milk shake. With his forefinger and a sound\n effect, Artie riddles Gillis' body. Gillis walks\n INTO THE SHOT.\n\n GILLIS\n Hello, Artie. Good evening,\n Miss Schaefer.\n\n BETTY"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Give me a pack of those Turkish cigarettes -- Melachrinos.\nDIALOGUE — Artie'S Voice: Stick 'em up, Gillis, or I'll let you have it!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Sunset Boulevard: Give me a pack of those Turkish cigarettes -- Melachrinos., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Give me a pack of those Turkish cigarettes -- Melachrinos. The camera pushes in with small amplitude at slow speed. Artie'S Voice (S1) says: [English] Stick 'em up, Gillis, or I'll let you have it! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Fox, Kylie, Ash, and Kristofferson ride down a country road.\nSetting: derived from Fantastic Mr. Fox\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ROAD. DAY\nACTION: Fox, Kylie, Ash, and Kristofferson ride down a country road. Kylie sees something across the meadow. He says warily:\nDIALOGUE — Kylie: Don't turn around!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fantastic Mr. Fox: Fox, Kylie, Ash, and Kristofferson ride down a country road. Kylie sees something across the meadow. He says warily:, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Fox, Kylie, Ash, and Kristofferson ride down a country road. Kylie sees something across the meadow. He says warily: The camera pushes in with small amplitude at slow speed. Kylie (S1) says: [English] Don't turn around! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You know your weapons, buddy.\nSetting: derived from \" T E R M I N A T O R \"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You know your weapons, buddy. Terminator examines each in turn, working the actions with\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \" T E R M I N A T O R \": You know your weapons, buddy. Terminator examines each in turn, working the actions with, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You know your weapons, buddy. Terminator examines each in turn, working the actions with The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in HorribleHistories illustrating the trope 'BringMeMyBrownPants'. Horrible Histories:\n After being intimidated by a Viking warlord online, Ethelred the Unready decides he might as well go clothes shopping as well... and posts an order for brown leggings.\n Henry VIII bursts in on the matchmaking consultation of the Earl of Arran and the infant Mary, Queen of Scots, and declares war on Scotland. One of the dating agency employees notes that someone needs changing ...\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying BringMeMyBrownPants.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for HorribleHistories, BringMeMyBrownPants, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for BringMeMyBrownPants. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: What I'm saying is... I don't\nSetting: derived from \"Total Recall\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: What I'm saying is... I don't remember you. I don't remember\nDIALOGUE — Melina: [And I thought...]\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Total Recall\": What I'm saying is... I don't remember you. I don't remember, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What I'm saying is... I don't remember you. I don't remember The camera pushes in with small amplitude at slow speed. Melina (S1) says: [English] [And I thought...] The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Black Screen\n\nSCREENPLAY EXCERPT:\nINT. BACKSTAGE\n\n\t\t\t\tSLOAN\n\t\t\tWe both lied to him.\n\n\t\t\t\tDELACROIX\n\t\t\tWhat do you want me to say?\n\n\t\t\t\tSLOAN\n\t\t\tJust don't lie to me."}, {"role": "assistant", "content": "## SCENE 1 — INT. BACKSTAGE\nACTION: What do you want me to say? Just don't lie to me.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Black Screen: What do you want me to say? Just don't lie to me., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What do you want me to say? Just don't lie to me. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: You wait here -- I'll let you in.\nSetting: derived from Mrs. Teasdale\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You wait here -- I'll let you in. Chico walks right in through the door, closes it behind him\nDIALOGUE — Vera: Psst...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Mrs. Teasdale: You wait here -- I'll let you in. Chico walks right in through the door, closes it behind him, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You wait here -- I'll let you in. Chico walks right in through the door, closes it behind him The camera pushes in with small amplitude at slow speed. Vera (S1) says: [English] Psst... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Minority Report\"\n\nSCREENPLAY EXCERPT:\nINT. OUTSIDE WITWER'S OFFICE -- MOMENTS LATER\n\n Anderson looks in the door Of Witwer's empty office. He\n takes a step inside.\n\n Witwer's booming voice sounds from behind him, startling\n him.\n\n WITWER\n Breaking and entering. That'll get\n you five to ten, hard.\n\n Witwer immediately scans his old partner's ashen face.\n\n WITWER (CONT'D)\n What's wrong?\n\n Anderson can hardly bear to meet his friend's eyes. He\n REACHES into his pocket, as if to lift the disc into the\n light. If he could just do that, show it to Witwer.\n\n \t\t\t WITWER\n Paul?\n\n Anderson's hand comes out of his pocket, EMPTY.\n\n ANDERSON\n Take over for me today?\n\n WITWER\n You sick?\n\n ANDERSON\n Yeah.\n\n Witwer makes a show of backing away.\n\n WITWER\n Don't give it to me. You probably\n have that Trans-10 virus going around.\n A stomach thing. I hate stomach\n things.\n\n Anderson Almost smiles.\n\n ANDERSON\n Ed.\n\n---------------"}, {"role": "assistant", "content": "## SCENE 1 — INT. OUTSIDE WITWER'S OFFICE -- MOMENTS LATER\nACTION: Anderson looks in the door Of Witwer's empty office. He Witwer's booming voice sounds from behind him, startling\nDIALOGUE — Witwer: Breaking and entering. That'll get you five to ten, hard.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Minority Report\": Anderson looks in the door Of Witwer's empty office. He Witwer's booming voice sounds from behind him, startling, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Anderson looks in the door Of Witwer's empty office. He Witwer's booming voice sounds from behind him, startling The camera pushes in with small amplitude at slow speed. Witwer (S1) says: [English] Breaking and entering. That'll get you five to ten, hard. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Mrs. Teasdale\n\nSCREENPLAY EXCERPT:\nYou wait here -- I'll let you in.\n\n Chico walks right in through the door, closes it behind him\n and then appears at the window. He raises it from the inside\n and Harpo climbs in through the window and joins Chico.\n\n CUT TO INSIDE. They are walking cautiously across the room\n when they hear the following coming over scene -- \"Psst...\n Psst...\" They stop and look around, trying to locate the\n source of the sound. We CUT TO the balcony and see Vera\n Trentino, still dressed in her negligee. Once more she tries\n to attract them.\n\n VERA\n Psst...\n\n The boys look up and see her.\n\n VERA\n (In a half whisper -\n mysteriously)\n In the safe -- behind the drapes --\n (Pointing in the\n direction of the\n safe)\n\n CHICO"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: You wait here -- I'll let you in. Chico walks right in through the door, closes it behind him\nDIALOGUE — Vera: Psst...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Mrs. Teasdale: You wait here -- I'll let you in. Chico walks right in through the door, closes it behind him, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. You wait here -- I'll let you in. Chico walks right in through the door, closes it behind him The camera pushes in with small amplitude at slow speed. Vera (S1) says: [English] Psst... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I don't know what you came here for, but\nSetting: derived from Script Date 5/25/94\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: I don't know what you came here for, but we won't have any part of it.\nDIALOGUE — Keaton: Edie, please.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Script Date 5/25/94: I don't know what you came here for, but we won't have any part of it., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I don't know what you came here for, but we won't have any part of it. The camera pushes in with small amplitude at slow speed. Keaton (S1) says: [English] Edie, please. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: They're quite exquisite!\nSetting: derived from The Puffy Shirt\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. MESMERIZED\nACTION: They're quite exquisite! (Mesmerized) Extraordinary! Have you\nDIALOGUE — George: What about them?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Puffy Shirt: They're quite exquisite! (Mesmerized) Extraordinary! Have you, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. They're quite exquisite! (Mesmerized) Extraordinary! Have you The camera pushes in with small amplitude at slow speed. George (S1) says: [English] What about them? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: He and Gus sit there with Lt. Walker, Harrigan, Andrews and\nSetting: derived from Basic Instinct\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LT. WALKER'S OFFICE\nACTION: He and Gus sit there with Lt. Walker, Harrigan, Andrews and Sixteen stab wounds to the chest\nDIALOGUE — Harrigan: Sixteen stab wounds to the chest and neck. No usable prints, no forcible entry, nothing missing. No prints on the ice pick, either -- it's available at any...\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Basic Instinct: He and Gus sit there with Lt. Walker, Harrigan, Andrews and Sixteen stab wounds to the chest, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. He and Gus sit there with Lt. Walker, Harrigan, Andrews and Sixteen stab wounds to the chest The camera pushes in with small amplitude at slow speed. Harrigan (S1) says: [English] Sixteen stab wounds to the chest and neck. No usable prints, no forcible entry, nothing missing. No prints on the ice pick, either -- it's available at any... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Rear Window\"\n\nSCREENPLAY EXCERPT:\nINT. APARTMENT - DAY - CLOSEUP\n\n Jeff's eyes drop from the ballet dancer's room to the one\n underneath."}, {"role": "assistant", "content": "## SCENE 1 — INT. APARTMENT - DAY - CLOSEUP\nACTION: Jeff's eyes drop from the ballet dancer's room to the one\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Rear Window\": Jeff's eyes drop from the ballet dancer's room to the one, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jeff's eyes drop from the ballet dancer's room to the one The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in FamilyTies illustrating the trope 'HippieParents'. On Family Ties Steven and Elyse tried to keep the spirit of the hippie movement alive while still living their upper middle class lifestyle.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying HippieParents.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for FamilyTies, HippieParents, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for HippieParents. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: (The phone rings. He picks it up and\nSetting: derived from Good News, Bad News\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SOMEBODY KNOCKS AT THE DOOR\nACTION: (The phone rings. He picks it up and says:) If you know what happened in\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Good News, Bad News: (The phone rings. He picks it up and says:) If you know what happened in, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. (The phone rings. He picks it up and says:) If you know what happened in The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Beavis And Butt-Head Do America\"\n\nSCREENPLAY EXCERPT:\nINT. HALLWAY OUTSIDE OF OVAL OFFICE - DAY\n\n\tA Secret Service guard is talking on radio/phone.\n\n\t\t\t\t\tGUARD\n\t\t\tEvacuation?... Probably just another bomb\n\t\t\tthreat or something... OK.\n\n\tThe guard walks off down the hallway, leaving his post.\n\n\tFrom the other end of the hallway we see Beavis/Cornholio enter,\n\tstill babbling."}, {"role": "assistant", "content": "## SCENE 1 — INT. HALLWAY OUTSIDE OF OVAL OFFICE - DAY\nACTION: A Secret Service guard is talking on radio/phone. Evacuation?... Probably just another bomb\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Beavis And Butt-Head Do America\": A Secret Service guard is talking on radio/phone. Evacuation?... Probably just another bomb, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A Secret Service guard is talking on radio/phone. Evacuation?... Probably just another bomb The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Final Destination\"\n\nSCREENPLAY EXCERPT:\nEXT. BACKYARD - CLEAR RIVER'S HOUSE - NIGHT - CLEAR'S POV\n\nAn old circular clothesline turns like a pinwheel in the wind. In the center of the yard is an oblong, tarped above ground pool. In the furthest corner of the yard, her dog is chained to a tree, BARKING AND FREAKED OUT. An old tire swing rocks in the storm. From the manner in which the house is situated and the length of the snapped power lines, one is capable of reaching the backyard.\n\nThe power line STRIKES the ground, \"closing in on\" the dog."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BACKYARD - CLEAR RIVER'S HOUSE - NIGHT - CLEAR'S POV\nACTION: An old circular clothesline turns like a pinwheel in the wind. In the center of the yard is an oblong, tarped above ground pool. In the furthest corner of the yard, her dog is chained to a tree, BARKING AND FREAKED OUT. An old tire swing rocks in the storm. From the manner in.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Final Destination\": An old circular clothesline turns like a pinwheel in the wind. In the center of the yard is an oblong, tarped above ground pool. In the furthest corner of the yard, her dog is chained to a tree, BARKI, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. An old circular clothesline turns like a pinwheel in the wind. In the center of the yard is an oblong, tarped above ground pool. In the furthest corner of the yard, her dog is chained to a tree, BARKING AND FREAKED OUT. An old tire swing rocks in the storm. From the manner in. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in ATouchOfCloth illustrating the trope 'RunningGag'. So you're gay? Anne Bi, Jack. Cloth No, don't go, it's none of my business.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying RunningGag.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for ATouchOfCloth, RunningGag, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for RunningGag. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Jurassic Park\n\nSCREENPLAY EXCERPT:\n- you gotta agree it's funny! These two,\n they dig up dinosaurs! It's wonderful,\n isn't it?\n\nNedry pats Grant on his shoulder.\n\n NEDRY\n Dr. Bones, you're going to love this place.\n\nNedry bursts out laughing again as he heads out the helicopter door.\n\nA smile comes across Gennaro's face. As he smiles he motions with his\nhands he doesn't mean any harm. Grant and Ellie stare at him.\n\n\n PILOT\n Come on folks. Gotta get back, there's a\n storm alert.\n\nROTORS TURN. OUTSIDE, a man reaches the copter. He wears a baseball\ncap over short red hair and he's dressed in phony safari garb. He\nshakes Gennaro's hand. This is ED REGIS, 35, head of Public Relations.\nHe throws open the copter door next to Grant. Big, cheerful smile.\n\n REGIS\n Hi! Ed Regis. Real big welcome to Isla\n Nublar, Dr. Grant, Dr. Sattler. Little\n tough landing here, I know. But you did\n it! Come on down, we're so happy to have\n you. Now, watch your step.\n\nEllie and Grant jump into the world of Jurassic Park.\n\n\nEXT LUSH TROPICAL FOREST - MORNING\n\nGrant takes in the beautiful tropical terrain. This place is the\nopposite of the Badlands. There is elaborate planting everywhere:\nhuge, hairy ferns; exotic, spiked flowers; berries of every color;\nrushing vines. Peeking through the thick greenery are beautiful birds\na\nnd flying squirrels. The strange, prehistoric world impresses Grant\nand Ellie. Even Nedry and Gennaro take in the vegetal wonder.\n"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: - you gotta agree it's funny! These two, they dig up dinosaurs! It's wonderful,\nDIALOGUE — Nedry: Dr. Bones, you're going to love this place.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Jurassic Park: - you gotta agree it's funny! These two, they dig up dinosaurs! It's wonderful,, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. - you gotta agree it's funny! These two, they dig up dinosaurs! It's wonderful, The camera pushes in with small amplitude at slow speed. Nedry (S1) says: [English] Dr. Bones, you're going to love this place. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: On her hands and knees, she enters the TUNNEL, not waiting\nSetting: derived from Fade In:\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TUNNEL/CUBBYHOLE\nACTION: On her hands and knees, she enters the TUNNEL, not waiting for him to respond. She knows what he'll do... Catherine and\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fade In:: On her hands and knees, she enters the TUNNEL, not waiting for him to respond. She knows what he'll do... Catherine and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. On her hands and knees, she enters the TUNNEL, not waiting for him to respond. She knows what he'll do... Catherine and The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BattlestarGalactica2003 illustrating the trope 'BoobsAndButtPose'. Six from Battlestar Galactica.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying BoobsAndButtPose.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BattlestarGalactica2003, BoobsAndButtPose, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for BoobsAndButtPose. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: American Psycho\n\nSCREENPLAY EXCERPT:\nEXT. STREET - NIGHT\n\nBATEMAN is trying to hail a cab. Owen's body is at his\nfeet. Luis Carruthers and a Japanese girl walk up to him.\n\nCARRUTHERS\nPatrick? Is that you?\n\nBATEMAN\nNo, Luis. It's not me. You're mistaken.\n\nCARRUTHERS\nThis is Gwendolyn Ichiban. This is my very good\nfriend Patrick Bateman. Where are you going? We're going to\nNell's. Gwendolyn's father's buying it.\n(Looking down at the body)\nWhere did you get your overnight bag?\n\nBATEMAN\nCommes des Garcon.\n\nA cab stops. BATEMAN opens the door and manages to get\nOwen's body into the backseat.\n\nBateman gets into the cab.\n\nCARRUTHERS\nCall me please, Patrick.\n\nBATEMAN\nJesus lives, Luis."}, {"role": "assistant", "content": "## SCENE 1 — EXT. STREET - NIGHT\nACTION: BATEMAN is trying to hail a cab. Owen's body is at his feet. Luis Carruthers and a Japanese girl walk up to him.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from American Psycho: BATEMAN is trying to hail a cab. Owen's body is at his feet. Luis Carruthers and a Japanese girl walk up to him., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. BATEMAN is trying to hail a cab. Owen's body is at his feet. Luis Carruthers and a Japanese girl walk up to him. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Well there�s...there�s not much to tell,\nSetting: derived from Saw\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOOKING AT HIM STRANGELY\nACTION: Well there�s...there�s not much to tell, really. (Across the room, Adam gives\nDIALOGUE — Adam: First dead body I�ve ever seen. Look different in real life. They don�t move.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Saw: Well there�s...there�s not much to tell, really. (Across the room, Adam gives, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Well there�s...there�s not much to tell, really. (Across the room, Adam gives The camera pushes in with small amplitude at slow speed. Adam (S1) says: [English] First dead body I�ve ever seen. Look different in real life. They don�t move. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: CLOSE ON: Dan, naked, on his knees in the center of an\nSetting: derived from Watchmen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. DAN'S NIGHTMARE - NIGHT\nACTION: CLOSE ON: Dan, naked, on his knees in the center of an INFINITE FIELD OF BLACKNESS. No corners, no exits. In the\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Watchmen: CLOSE ON: Dan, naked, on his knees in the center of an INFINITE FIELD OF BLACKNESS. No corners, no exits. In the, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CLOSE ON: Dan, naked, on his knees in the center of an INFINITE FIELD OF BLACKNESS. No corners, no exits. In the The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Natural Born Killers\"\n\nSCREENPLAY EXCERPT:\nINT. COURTROOM - DAY\n\n COURT SKETCH - CU of Grace staring intently. CAMERA widens to\n include Bisbing standing before the bench as Mickey sits with\n Mallory at the defence table. Judge Steinsma looks down at\n Grace. The distinct sounds of a court in session can be heard\n in the B.G.\n\n The color of the scene FADES to Black & White as the CAMERA\n cranes up to reveal the COURT ARTIST drawing while court is in\n session.\n\n OFF SCREEN: Bisbing is wrapping up her examination of Grace who\n is at the witness stand. Judge Steinsma resides at his bench.\n\n CAMERA dollies into a MED. OVERHEAD TWO-SHOT of Mickey and\n Mallory who are sketching on a pad of paper. CAMERA reveals the\n drawing of a man stabbing a woman. Mickey is doing the actual\n drawing, but every once and a while Mallory takes to pencil and\n adds some touches, as Grace finishes her tearful testimony.\n\n BISBING (O.S.)\n No further questions, your honor.\n\n Emotional pause, the silence.\n\n JUDGE STEINSMA (O.S.)\n Would you care to cross examine the\n witness, Mr. Knox?\n\n Mickey scratches out the drawing. Still holding the pencil he\n stands and leaves frame.\n\n MICKEY (O.S)\n As a matter of fact, your honor, I\n would.\n\n NOTE: The following scene is to be played with EXTREME CLOSE\n UPS. Not once is the courtroom seen in detail. However\n courtroom sound can be heard vivi"}, {"role": "assistant", "content": "## SCENE 1 — INT. COURTROOM - DAY\nACTION: COURT SKETCH - CU of Grace staring intently. CAMERA widens to include Bisbing standing before the bench as Mickey sits with\nDIALOGUE — Bisbing: No further questions, your honor.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Natural Born Killers\": COURT SKETCH - CU of Grace staring intently. CAMERA widens to include Bisbing standing before the bench as Mickey sits with, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. COURT SKETCH - CU of Grace staring intently. CAMERA widens to include Bisbing standing before the bench as Mickey sits with The camera pushes in with small amplitude at slow speed. Bisbing (S1) says: [English] No further questions, your honor. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Alien\n\nSCREENPLAY EXCERPT:\nI don't know... I'm getting these\n CRAMPS!\n\n The others stare at him in alarm. Another GROAN is torn from his lips.\n He clutches the edge of the table with his hands, his knuckles\n whitening.\n\n STANDARD\n Breathe deeply.\n\n BROUSSARD"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I don't know... I'm getting these The others stare at him in alarm. Another GROAN is torn from his lips.\nDIALOGUE — Standard: Breathe deeply.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Alien: I don't know... I'm getting these The others stare at him in alarm. Another GROAN is torn from his lips., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I don't know... I'm getting these The others stare at him in alarm. Another GROAN is torn from his lips. The camera pushes in with small amplitude at slow speed. Standard (S1) says: [English] Breathe deeply. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: It is a typical '50s family practice right out of Norman\nSetting: derived from Pleasantville\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. DOCTOR'S OFFICE. DAY.\nACTION: It is a typical '50s family practice right out of Norman Rockwell. There is a jar of tongue depressors on the counter\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Pleasantville: It is a typical '50s family practice right out of Norman Rockwell. There is a jar of tongue depressors on the counter, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. It is a typical '50s family practice right out of Norman Rockwell. There is a jar of tongue depressors on the counter The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Batman 2\n\nSCREENPLAY EXCERPT:\nINT. POLICE OBSERVATION ROOM - NIGHT\n\n COMMISSIONER GORDON and another cop, LT. EDDIE BULLOCK, are in\n darkened antechamber adjacent to an interrogation room.\n\n BULLOCK\n It wasn't about the drugs. Whoever it was\n left thirty kilos sitting on the docks.\n\n They're watching, through a two-way glass panel, as a\n terrified man with a heavily-bandaged face tells his story.\n It's the lone survivor of the dock massacre, JULIO, and his\n voice is audible over a concealed intercom:\n\n JULIO (filter)\n Un silueta negra -- con colmillos, y\n garras -- el demonio. El murcielago.\n MURCIELAGO!\n\n GORDON\n What's that he keeps saying?\n\n BULLOCK\n \"Murcielago.\" -- Bat.\n\n GORDON\n Nonsense. That dock looked like a\n slaughter-house. Batman's never committed\n murder.\n\n BULLOCK\n We did find this in his back pocket.\n\n BULLOCK hands GORDON a CARD. It reads: \"THOSE WHO FEED ON THE\n SOUL OF GOTHAM WILL SUFFER MY WRATH\" -- and in lieu of a\n signature, there's a little black BAT-EMBLEM in the bottom\n corner.\n\n While GORDON's staring at it, a POLICEMAN pokes his head in:\n\n POLICEMAN\n Commissioner? We've got Barrett."}, {"role": "assistant", "content": "## SCENE 1 — INT. POLICE OBSERVATION ROOM - NIGHT\nACTION: COMMISSIONER GORDON and another cop, LT. EDDIE BULLOCK, are in darkened antechamber adjacent to an interrogation room.\nDIALOGUE — Bullock: It wasn't about the drugs. Whoever it was left thirty kilos sitting on the docks.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman 2: COMMISSIONER GORDON and another cop, LT. EDDIE BULLOCK, are in darkened antechamber adjacent to an interrogation room., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. COMMISSIONER GORDON and another cop, LT. EDDIE BULLOCK, are in darkened antechamber adjacent to an interrogation room. The camera pushes in with small amplitude at slow speed. Bullock (S1) says: [English] It wasn't about the drugs. Whoever it was left thirty kilos sitting on the docks. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Beavis And Butt-Head Do America\"\n\nSCREENPLAY EXCERPT:\nINT. TOUR BUS - DAY\n\n\tButt-Head sits. Beavis pauses, still standing.\n\n\t\t\t\t\tBEAVIS\n\t\t\tHey wait a minute. What's going on? Why\n\t\t\tare we getting back on the bus?\n\n\t\t\t\t\tOLD GUY\n\t\t\tIt's time to go son.\n\n\t\t\t\t\tBEAVIS\n\t\t\tWe can't leave! We never met that chick!\n\t\t\tDammit!!! We were supposed to get some!\n\n\t\t\t\t\tBUTT-HEAD\n\t\t\tHuh huh huh. Settle down Beavis.\n\n\t\t\t\t\tBEAVIS\n\t\t\tOh yeah,...I mean no. NO! I won't settle\n\t\t\tdown! Not this time!...\n\n\tBeavis is shaking, fed up. He delivers the speech of his life.\n\n\t\t\t\t\tBEAVIS (CONT.)\n\t\t\tDammit, this always happens! I think I'm\n\t\t\tgonna score and then I never score! It's\n\t\t\tnot fair! We've traveled a hundred miles\n\t\t\t'cause we thought we were gonna score, but\n\t\t\tnow it's not gonna happen!\n\n\t\t\t\t\tBUS DRIVER\n\t\t\t\t\t(yelling from his seat)\n\t\t\tHey buddy, sit down! Now!\n\n\t\t\t\t\tBEAVIS\n\t\t\tSHUT UP! (continuing) I'm sick and tired\n\t\t\tof this! We're never gonna score! It's\n\t\t\tjust not gonna happen! We're just gonna\n\t\t\tget old like these people, but they've\n\t\t\tprobably scored!\n\n\t\t\t\t\tBUS DRIVER\n\t\t\t\t\t(standing)\n\t\t\tHey! I'm warning you! Sit down!\n\n\t\t\t\t\tBEAVIS\n\t\t\tIt's like this chick's a slut (motioning\n\t\t\tto Martha)... and look at this guy!...\n\t\t\tHe's old but he's probably scored a\n\t\t\tmillion times!\n\n\t\t\t\t\tOLD GUY\n\t\t\t\t\t(nods in agreement)\n\t\t\tOhh yeah.\n\n\t\t\t\t\tBEAVIS\n\t\t\tBut not us! We're never gonna score! WE'RE\n\t\t\tNEVER GONNA SCORE!!! AAGGHHHH!!!\n\n\tThe bus driver tackles Beavis."}, {"role": "assistant", "content": "## SCENE 1 — INT. TOUR BUS - DAY\nACTION: Butt-Head sits. Beavis pauses, still standing. Hey wait a minute. What's going on? Why\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Beavis And Butt-Head Do America\": Butt-Head sits. Beavis pauses, still standing. Hey wait a minute. What's going on? Why, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Butt-Head sits. Beavis pauses, still standing. Hey wait a minute. What's going on? Why The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Futurama\n\nSCREENPLAY EXCERPT:\nFUTURAMA\n\n Episode 220\n\n \"ANTHOLOGY OF INTEREST I\"\n\n By\n\n Eric Rogers, Ken Keeler & David X. Cohen\n\n Transcribed by Dave, The Neutral Planet\n\n\n\n\n [Opening Credits. Caption: Painstakingly Drawn Before A Live\n Audience.]\n\n\n [Planet Express: Farnsworth's Lab. The staff are assembled around\n a table.]\n\n\n FARNSWORTH\n Good news, everyone! I've invented a\n device that allows you to operate equipment\n from great distances. I call it \"the\n fing-longer\". Observe. And, here we\n go. There!\n\n\n FRY\n Ooh!\n\n\n AMY\n Wow!\n\n\n ZOIDBERG\n Ah!\n\n\n FARNSWORTH\n Pretty long, eh?\n\n\n FRY\n Yeah, it's really long. But what did\n you just turn on with it?\n\n\n FARNSWORTH\n Oh, that's just the What-If machine\n I invented. You pose it a What-If question\n and it generates a video simulation\n of what"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Eric Rogers, Ken Keeler & David X. Cohen Transcribed by Dave, The Neutral Planet\nDIALOGUE — Farnsworth: Good news, everyone! I've invented a device that allows you to operate equipment from great distances. I call it \"the fing-longer\". Observe. And, here we go....\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Futurama: Eric Rogers, Ken Keeler & David X. Cohen Transcribed by Dave, The Neutral Planet, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Eric Rogers, Ken Keeler & David X. Cohen Transcribed by Dave, The Neutral Planet The camera pushes in with small amplitude at slow speed. Farnsworth (S1) says: [English] Good news, everyone! I've invented a device that allows you to operate equipment from great distances. I call it \"the fing-longer\". Observe. And, here we go.... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: CAMERA, FACING the ocean, moves along the shore revealing a lone image sitting on the beach, looking into the Atlantic.\nSetting: derived from \"Final Destination\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. BEACH - NIGHT\nACTION: CAMERA, FACING the ocean, moves along the shore revealing a lone image sitting on the beach, looking into the Atlantic. The waves are small bur rhythmic, an enternal metronome. Numb, Alex Browning sits in the sand searching for an answer somewhere in the darkness of the evening.\nDIALOGUE — Alex: Are they up there?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Final Destination\": CAMERA, FACING the ocean, moves along the shore revealing a lone image sitting on the beach, looking into the Atlantic. The waves are small bur rhythmic, an enternal metronome. Numb, Alex Browning sit, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CAMERA, FACING the ocean, moves along the shore revealing a lone image sitting on the beach, looking into the Atlantic. The waves are small bur rhythmic, an enternal metronome. Numb, Alex Browning sits in the sand searching for an answer somewhere in the darkness of the evening. The camera pushes in with small amplitude at slow speed. Alex (S1) says: [English] Are they up there? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Ext. Manhattan Island - Day\n\nSCREENPLAY EXCERPT:\nEXT. FIREHOUSE - DAY\n\nThe old, dilapidated Ghostbusters logo comes crashing to the ground,\ndropped by a pair of workmen on a ladder, and the new logo is hoisted\ninto place over the main entrance. It's just like the original \"No\nGhosts\" logo, but in the new one the ghost in the red circle is holding\nup two fingers. Venkman enters the firehouse."}, {"role": "assistant", "content": "## SCENE 1 — EXT. FIREHOUSE - DAY\nACTION: The old, dilapidated Ghostbusters logo comes crashing to the ground, dropped by a pair of workmen on a ladder, and the new logo is hoisted\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Ext. Manhattan Island - Day: The old, dilapidated Ghostbusters logo comes crashing to the ground, dropped by a pair of workmen on a ladder, and the new logo is hoisted, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The old, dilapidated Ghostbusters logo comes crashing to the ground, dropped by a pair of workmen on a ladder, and the new logo is hoisted The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Beavis And Butt-Head Do America\"\n\nSCREENPLAY EXCERPT:\nINT. HOTEL/CASINO - DUSK\n\n\tON B&B, still staring at the statue.\n\n\t\t\t\t\tB&B\n\t\t\t\t\t(in awe)\n\t\t\tHuh huh huh huh huh.\n\n\tA security guard comes and drags B&B away.\n\n\tON THE LOUNGE BAND, continuing the song we've been hearing."}, {"role": "assistant", "content": "## SCENE 1 — INT. HOTEL/CASINO - DUSK\nACTION: ON B&B, still staring at the statue. A security guard comes and drags B&B away.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Beavis And Butt-Head Do America\": ON B&B, still staring at the statue. A security guard comes and drags B&B away., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ON B&B, still staring at the statue. A security guard comes and drags B&B away. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Rear Window\"\n\nSCREENPLAY EXCERPT:\nEXT. NEIGHBORHOOD - NIGHT - MEDIUM SHOT\n\n We get a better view of the salesman waiting while his\n connection is being made. Whoever he has called answers. And\n instantly there is a marked change in his attitude. He\n relaxes, smiles, is warm. He talks softly, perhaps guardedly,\n with an occasional glance at the bedroom door. In the bedroom,\n his wife has become aware of the call.\n\n Quietly she moves the tray, gets out of bed, and goes to the\n bedroom door to listen. The wall hides her from our view.\n\n Then suddenly, she apparently opens the door, because the\n living room, we see her arm suddenly appear, pointing at the\n man and the telephone. He speaks quickly into the phone, and\n hangs up. His face is flushed and angry as he goes toward\n the bedroom. In the bedroom his wife appears walking back to\n the bed, followed by the husband. She is laughing, and he is\n answering her in angry tones. She climbs in bed laughing.\n The more she laughs, the more angry he gets, and the harder\n she laughs. Finally, he leaves the room, goes into the living\n room, back into the kitchen and has another drink. He stands\n there, controlling an outburst of emotion, and seems almost\n to be crushing the shot glass in his clenched fist."}, {"role": "assistant", "content": "## SCENE 1 — EXT. NEIGHBORHOOD - NIGHT - MEDIUM SHOT\nACTION: We get a better view of the salesman waiting while his connection is being made. Whoever he has called answers. And\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Rear Window\": We get a better view of the salesman waiting while his connection is being made. Whoever he has called answers. And, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. We get a better view of the salesman waiting while his connection is being made. Whoever he has called answers. And The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"The Thing\"\n\nSCREENPLAY EXCERPT:\nINT. STORAGE ROOM\n\n Inside are several marijuana plants. Sun lamps beam down on\n them. Childs inspects them with a wide grin.\n\n CHILDS\n How my brothers and sister doing\n today? Doin' fine.\n\n He moves over to a tape deck, selects a cassette, grins back\n at the plants and turns it on.\n\n CHILDS\n What say to some nice Al Green for\n my babies, huh?\n\n He waters them carefully, as Al Green sings softly. He hears\n a panting and turns around to see the dog. His bandage is\n gone.\n\n CHILDS\n What you...? You get the hell on out\n of here.\n\n The dog is shooed off. Childs turns back grumbling.\n\n CHILDS\n ...Comin' in here... goin' to urinate\n on my babies."}, {"role": "assistant", "content": "## SCENE 1 — INT. STORAGE ROOM\nACTION: Inside are several marijuana plants. Sun lamps beam down on them. Childs inspects them with a wide grin.\nDIALOGUE — Childs: How my brothers and sister doing today? Doin' fine.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"The Thing\": Inside are several marijuana plants. Sun lamps beam down on them. Childs inspects them with a wide grin., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Inside are several marijuana plants. Sun lamps beam down on them. Childs inspects them with a wide grin. The camera pushes in with small amplitude at slow speed. Childs (S1) says: [English] How my brothers and sister doing today? Doin' fine. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: That's not too bad. It's not like a\nSetting: derived from The Suicide\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: That's not too bad. It's not like a Sunny von Bulow comma. The doctor said\nDIALOGUE — Gina: You know why he did this? Because I told him it was over. I did not want to see him anymore.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Suicide: That's not too bad. It's not like a Sunny von Bulow comma. The doctor said, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. That's not too bad. It's not like a Sunny von Bulow comma. The doctor said The camera pushes in with small amplitude at slow speed. Gina (S1) says: [English] You know why he did this? Because I told him it was over. I did not want to see him anymore. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The small, spartan hovel is cluttered with desert junk but\nSetting: derived from Star Wars\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. KENOBI'S DWELLING\nACTION: The small, spartan hovel is cluttered with desert junk but still manages to radiate an air of time-worn comfort and\nDIALOGUE — Luke: No, my father didn't fight in the wars. He was a navigator on a spice freighter.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Star Wars: The small, spartan hovel is cluttered with desert junk but still manages to radiate an air of time-worn comfort and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The small, spartan hovel is cluttered with desert junk but still manages to radiate an air of time-worn comfort and The camera pushes in with small amplitude at slow speed. Luke (S1) says: [English] No, my father didn't fight in the wars. He was a navigator on a spice freighter. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in BronBroen illustrating the trope 'CutApart'. Used in the third season when the SWAT team raid Aleks's house looking for Petersson, but Aleks and Petersson are in a different building, and the killer shoots Aleks and re-kidnaps Petersson.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying CutApart.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for BronBroen, CutApart, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for CutApart. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Empire Strikes Back\n\nSCREENPLAY EXCERPT:\nEXT. SPACE - PLANET DAGOBAH\n\nLuke's tiny X-wing rockets away from the green planet of Dagobah and\noff into space."}, {"role": "assistant", "content": "## SCENE 1 — EXT. SPACE - PLANET DAGOBAH\nACTION: Luke's tiny X-wing rockets away from the green planet of Dagobah and\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Empire Strikes Back: Luke's tiny X-wing rockets away from the green planet of Dagobah and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Luke's tiny X-wing rockets away from the green planet of Dagobah and The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Batman 2\n\nSCREENPLAY EXCERPT:\nINT. PENGUIN'S LAIR - DAY\n\n The PENGUIN stands over his indoor penguin pool. He's wearing\n rubber gloves, feeding LIVE FISH from an ice chest to his\n arctic birds. The vents in the windows are open, and the\n climate in the room is downright icy as SELINA's wrapped in\n fur, stroking her pet cat:\n\n SELINA\n He's just another rich idiot.\n (chuckling to herself)\n The odd thing is, he didn't seem a bit\n concerned.\n\n PENGUIN\n Then he is an idiot.\n\n SELINA\n He lives in some big sprawling manor.\n I'll have to get inside, scope it out...\n see where he's got the raven stashed.\n\n PENGUIN\n How do you plan to do that?\n\n SELINA\n How do you think!\n\n A feline smile from SELINA. The PENGUIN chuckles to himself,\n lobs a FISH out over the pool. A swooping GULL snatches it\n out of the air before it hits the water. FRICK arrives in the\n doorway.\n\n FRICK\n It's Mr. Provost, sir. He's planning to\n embark on an unscheduled Christmas\n vacation.\n\n PENGUIN\n Good! That should save us a trip to the\n bank.\n\n CUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — INT. PENGUIN'S LAIR - DAY\nACTION: The PENGUIN stands over his indoor penguin pool. He's wearing rubber gloves, feeding LIVE FISH from an ice chest to his\nDIALOGUE — Selina: He's just another rich idiot. The odd thing is, he didn't seem a bit concerned.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman 2: The PENGUIN stands over his indoor penguin pool. He's wearing rubber gloves, feeding LIVE FISH from an ice chest to his, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The PENGUIN stands over his indoor penguin pool. He's wearing rubber gloves, feeding LIVE FISH from an ice chest to his The camera pushes in with small amplitude at slow speed. Selina (S1) says: [English] He's just another rich idiot. The odd thing is, he didn't seem a bit concerned. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I don't know... I'm getting these\nSetting: derived from Alien\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I don't know... I'm getting these The others stare at him in alarm. Another GROAN is torn from his lips.\nDIALOGUE — Standard: Breathe deeply.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Alien: I don't know... I'm getting these The others stare at him in alarm. Another GROAN is torn from his lips., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I don't know... I'm getting these The others stare at him in alarm. Another GROAN is torn from his lips. The camera pushes in with small amplitude at slow speed. Standard (S1) says: [English] Breathe deeply. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Jack sits at the table, sips coffee. He's pale, dazed,\nSetting: derived from F I G H T C L U B\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. KITCHEN - MOMENTS LATER\nACTION: Jack sits at the table, sips coffee. He's pale, dazed, seems broken. He hears the faint SOUND of SAWING and\nDIALOGUE — Marla: I'll be out of your way in a sec.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from F I G H T C L U B: Jack sits at the table, sips coffee. He's pale, dazed, seems broken. He hears the faint SOUND of SAWING and, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack sits at the table, sips coffee. He's pale, dazed, seems broken. He hears the faint SOUND of SAWING and The camera pushes in with small amplitude at slow speed. Marla (S1) says: [English] I'll be out of your way in a sec. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Jack's leaving, looks like a war casualty, passing COWORKERS\nSetting: derived from F I G H T C L U B\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HALLWAY - SAME\nACTION: Jack's leaving, looks like a war casualty, passing COWORKERS who coldly stare at him. His face is totally passive.\nDIALOGUE — Jack: I got right in everyone's hostile little face. Yes, these are bruises from fighting. I'm comfortable with them. I am enlightened.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from F I G H T C L U B: Jack's leaving, looks like a war casualty, passing COWORKERS who coldly stare at him. His face is totally passive., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Jack's leaving, looks like a war casualty, passing COWORKERS who coldly stare at him. His face is totally passive. The camera pushes in with small amplitude at slow speed. Jack (S1) says: [English] I got right in everyone's hostile little face. Yes, these are bruises from fighting. I'm comfortable with them. I am enlightened. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: The Marine Biologist\n\nSCREENPLAY EXCERPT:\nELAINE\n What, how did you know that?\n\n\n JERRY\n Because I know who has it.\n\n\n ELAINE\n What, how did you find it?\n\n\n JERRY\n Because the woman who got hit in the\n head with it found my name in\n\n\n it, called me up, and we met!\n\n\n ELAINE\n Well where is it, give it to me!\n\n\n JERRY\n I don't have it!\n\n\n ELAINE\n Why not.\n\n\n JERRY\n Because she's not returning it until\n she gets the money back for the\n\n\n hospital bill.\n\n\n ELAINE\n But I didn't do Testikov did it, he\n should have to pay for it!\n\n\n JERRY\n How much is Testikov getting from Pendman\n for this book?\n\n\n ELAINE\n One million.\n\n\n JERRY\n Well that's a start."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: What, how did you know that? Because I know who has it.\nDIALOGUE — Jerry: Because I know who has it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Marine Biologist: What, how did you know that? Because I know who has it., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What, how did you know that? Because I know who has it. The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] Because I know who has it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Bride Of Leprechaun\n\nSCREENPLAY EXCERPT:\nI'm not giving you anything 'til I\n get Bridget. You'll have to kill\n me, if you want your stinking piece\n of gold.\n\n LEPRECHAUN\n So be it.\n\nThe Leprechaun hits the gas and heads for Cody. Cody tries to\nget to his feet, but his foot is stuck. The impact is\nimminent. Cody lets out a kamikaze yell!\n\nThe go cart hits Cody, but... passes riqht throuqh him.\n\nThe Leprechaun spins and screeches to a stop, facing Cody.\nCody frees his leg and stands, unscathed. He turns towards\nthe Leprechaun, in disbelief. The Leprechaun looks very\nangry.\n 83.\n\n\n\n CODY\n I'm not dead.\n\n LEPRECHAUN\n You will be if you don't give me\n the coin!\n\n The Leprechaun doesn't move. Cody takes the coin out of his\n pocket and looks at it, comprehending its power.\n\n CODY\n You can't do it, can you?\n\n Cody holds up the coin.\n\n CODY (CONT'D)\n As long as I have this!\n\n The Leprechaun growls at Cody.\n\n LEPRECHAUN\n You have to give it to me!!\n\n CODY"}, {"role": "assistant", "content": "## SCENE 1 — INT. CONT'D\nACTION: I'm not giving you anything 'til I get Bridget. You'll have to kill\nDIALOGUE — Leprechaun: So be it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bride Of Leprechaun: I'm not giving you anything 'til I get Bridget. You'll have to kill, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm not giving you anything 'til I get Bridget. You'll have to kill The camera pushes in with small amplitude at slow speed. Leprechaun (S1) says: [English] So be it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: The Psychiatrist listens, paralyzed. Horrified.\nSetting: derived from Watchmen\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. PRISON PSYCH ROOM - DAY\nACTION: The Psychiatrist listens, paralyzed. Horrified. A tremor of impact shook my arm. Warm\nDIALOGUE — Rorschach: A tremor of impact shook my arm. Warm blood splashed my face. It was Walter Kovacs who screamed, who closed his eyes-- it was Rorschach who opened them again....\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Watchmen: The Psychiatrist listens, paralyzed. Horrified. A tremor of impact shook my arm. Warm, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The Psychiatrist listens, paralyzed. Horrified. A tremor of impact shook my arm. Warm The camera pushes in with small amplitude at slow speed. Rorschach (S1) says: [English] A tremor of impact shook my arm. Warm blood splashed my face. It was Walter Kovacs who screamed, who closed his eyes-- it was Rorschach who opened them again.... The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A warning bell rings at a railroad crossing.\nSetting: derived from Asteroid City\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. HIGHWAY. DAY\nACTION: A warning bell rings at a railroad crossing. The train slowly rattles over a blacktop interstate which\nDIALOGUE — Ramp Closed Indefinitely: Department of Roads and Rails 5.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Asteroid City: A warning bell rings at a railroad crossing. The train slowly rattles over a blacktop interstate which, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. A warning bell rings at a railroad crossing. The train slowly rattles over a blacktop interstate which The camera pushes in with small amplitude at slow speed. Ramp Closed Indefinitely (S1) says: [English] Department of Roads and Rails 5. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Blade\n\nSCREENPLAY EXCERPT:\nINT. TEMPLE SANCTUARY - NIGHT\n\n\t-- the invocation of LaMagra reaches critical mass. Frost is consumed\n\tby a hellish force, unable to withstand the onslaught of spirit\n\tenergy any longer.\n\n\tCUT TO:"}, {"role": "assistant", "content": "## SCENE 1 — INT. TEMPLE SANCTUARY - NIGHT\nACTION: -- the invocation of LaMagra reaches critical mass. Frost is consumed by a hellish force, unable to withstand the onslaught of spirit\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade: -- the invocation of LaMagra reaches critical mass. Frost is consumed by a hellish force, unable to withstand the onslaught of spirit, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. -- the invocation of LaMagra reaches critical mass. Frost is consumed by a hellish force, unable to withstand the onslaught of spirit The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: T W I N P E A K S\n\nSCREENPLAY EXCERPT:\nI'm Sam Stanley. If you ever need me.\n\n DESMOND\n Thanks, Sam, for the good work. You\n have a good eye for detail.\n\n STANLEY\n We do notice things, don't we, Agent\n Desmond? Are you going back to the\n trailer park for the Blue rose?\n\n Desmond smiles at Stanley as they shake.\n\n DESMOND'S POV\n\n As he enters the van, Desmond can barely make out what Stanley says\n as he points to the driver.\n\n STANLEY"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I'm Sam Stanley. If you ever need me. Thanks, Sam, for the good work. You\nDIALOGUE — Desmond: Thanks, Sam, for the good work. You have a good eye for detail.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from T W I N P E A K S: I'm Sam Stanley. If you ever need me. Thanks, Sam, for the good work. You, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I'm Sam Stanley. If you ever need me. Thanks, Sam, for the good work. You The camera pushes in with small amplitude at slow speed. Desmond (S1) says: [English] Thanks, Sam, for the good work. You have a good eye for detail. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in DieSendungMitDerMaus illustrating the trope 'PercussiveMaintenance'. In a cartoon in Die Sendung mit der Maus the mouse wants to make a photo of herself. All the photo machine produces are elephants. This trope doesn't help a bit. Why is revealed when the mouse opens the maintenance lid: the machine is a fake and the little blue elephant inside does the pics with crayons.\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying PercussiveMaintenance.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for DieSendungMitDerMaus, PercussiveMaintenance, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for PercussiveMaintenance. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Hellboy\"\n\nSCREENPLAY EXCERPT:\nINT. CATACOMBS - NIGHT\n\n Hellboy thrusts his hand into the SECOND IMPRINT. CLACK! He\n turns it.\n\n Myers spots the broken rosary and reaches for it --"}, {"role": "assistant", "content": "## SCENE 1 — INT. CATACOMBS - NIGHT\nACTION: Hellboy thrusts his hand into the SECOND IMPRINT. CLACK! He Myers spots the broken rosary and reaches for it --\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Hellboy\": Hellboy thrusts his hand into the SECOND IMPRINT. CLACK! He Myers spots the broken rosary and reaches for it --, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Hellboy thrusts his hand into the SECOND IMPRINT. CLACK! He Myers spots the broken rosary and reaches for it -- The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I miss Doc. I miss that ol' rip.\nSetting: derived from Tombstone\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I miss Doc. I miss that ol' rip. A handsomely appointed saloon. At a corner table, the pot is\nDIALOGUE — Wyatt: I miss Doc. I miss that ol' rip.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Tombstone: I miss Doc. I miss that ol' rip. A handsomely appointed saloon. At a corner table, the pot is, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I miss Doc. I miss that ol' rip. A handsomely appointed saloon. At a corner table, the pot is The camera pushes in with small amplitude at slow speed. Wyatt (S1) says: [English] I miss Doc. I miss that ol' rip. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ECKHARDT'S SWAT TEAM goes wide-eyed as a CONVOY OF POLICE\nSetting: derived from Batman\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. ACE CHEMICAL - NIGHT\nACTION: ECKHARDT'S SWAT TEAM goes wide-eyed as a CONVOY OF POLICE BLACK-AND-WHITES roars into the Ace parking lot. UNIFORMED\nDIALOGUE — Eckhardt: What are you trying to do, blow the collar?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Batman: ECKHARDT'S SWAT TEAM goes wide-eyed as a CONVOY OF POLICE BLACK-AND-WHITES roars into the Ace parking lot. UNIFORMED, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ECKHARDT'S SWAT TEAM goes wide-eyed as a CONVOY OF POLICE BLACK-AND-WHITES roars into the Ace parking lot. UNIFORMED The camera pushes in with small amplitude at slow speed. Eckhardt (S1) says: [English] What are you trying to do, blow the collar? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Austin Powers: International Man Of Mystery\n\nSCREENPLAY EXCERPT:\nINT. CASINO\n\n Austin and Vanessa join the high-rollers table. Number Two\n is there, complete with eyepatch. On one side of him is a\n beautiful ITALIAN WOMAN (a la SOPHIA LOREN) in a white dress\n with a white kerchief on her head. On the other side of him\n is an extremely large-breasted BIMBO.\n\n AUSTIN\n Do you mind if I join you?\n\n NUMBER TWO\n Not at all.\n\n The DEALER deals.\n\n DEALER\n Seventeen.\n\n Zoom in on Number Two's eyepatch.\n\n NUMBER TWO'S MONOCULAR POV\n\n GRAPHIC: \"X-RAY EYEPATCH\". We see everyone at the casino\n in their underwear. He looks at the next card in the shoe.\n It is a 4.\n\n NUMBER TWO\n Hit me.\n\n DEALER\n You have seventeen, sir. The book\n says not to, sir.\n\n NUMBER TWO\n I like to live dangerously.\n\n The dealer draws a card from the card shoe.\n\n DEALER\n Four. Twenty-one.\n\n Everyone at the table applauds. The dealer deals to Austin\n and Number Two.\n\n DEALER\n (t"}, {"role": "assistant", "content": "## SCENE 1 — INT. CASINO\nACTION: Austin and Vanessa join the high-rollers table. Number Two is there, complete with eyepatch. On one side of him is a\nDIALOGUE — Austin: Do you mind if I join you?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Austin Powers: International Man Of Mystery: Austin and Vanessa join the high-rollers table. Number Two is there, complete with eyepatch. On one side of him is a, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Austin and Vanessa join the high-rollers table. Number Two is there, complete with eyepatch. On one side of him is a The camera pushes in with small amplitude at slow speed. Austin (S1) says: [English] Do you mind if I join you? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: ANGLE - BARIN advancing by himself, HEARING the triumphant\nSetting: derived from Flash Gordon\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — EXT. FOREST -FEATURE BARIN\nACTION: ANGLE - BARIN advancing by himself, HEARING the triumphant Prince Barin! We have him!\nDIALOGUE — Tree Man: Prince Barin! We have him!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Flash Gordon: ANGLE - BARIN advancing by himself, HEARING the triumphant Prince Barin! We have him!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. ANGLE - BARIN advancing by himself, HEARING the triumphant Prince Barin! We have him! The camera pushes in with small amplitude at slow speed. Tree Man (S1) says: [English] Prince Barin! We have him! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: What, how did you know that?\nSetting: derived from The Marine Biologist\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: What, how did you know that? Because I know who has it.\nDIALOGUE — Jerry: Because I know who has it.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Marine Biologist: What, how did you know that? Because I know who has it., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. What, how did you know that? Because I know who has it. The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] Because I know who has it. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Austin Powers: International Man Of Mystery\n\nSCREENPLAY EXCERPT:\nINT. DR. EVIL'S PRIVATE QUARTERS\n\n Austin and Vanessa are seated at a table with Frau. WAITERS\n serve food.\n\n MUSIC: Sexy Matt Helm-type theme\n\n DR. EVIL\n I think you'll enjoy the food. I\n have the best chef in the world.\n His name is Ezekial. He's made of\n seventy-five percent plastic.\n\n Scott enters.\n\n DR. EVIL\n Scott my boy, come here. How was\n your day?\n\n SCOTT EVIL\n Well, me and a buddy went to the\n video arcade in town and, like, they\n don't speak English right, and so my\n buddy gets into a fight, and he goes\n 'hey, quit hassling me cause I don't\n speak French or whatever', and the\n other guy goes something in Paris\n talk, and I go 'um, just back off'\n and he goes 'get out' and I go 'make\n me'.\n\n DR. EVIL\n (trying to hide\n contempt)\n Fascinating. What are your plans\n for this evening?\n\n SCOTT EVIL\n Thought I'd stay in. There"}, {"role": "assistant", "content": "## SCENE 1 — INT. DR. EVIL'S PRIVATE QUARTERS\nACTION: Austin and Vanessa are seated at a table with Frau. WAITERS MUSIC: Sexy Matt Helm-type theme\nDIALOGUE — Dr. Evil: I think you'll enjoy the food. I have the best chef in the world. His name is Ezekial. He's made of seventy-five percent plastic.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Austin Powers: International Man Of Mystery: Austin and Vanessa are seated at a table with Frau. WAITERS MUSIC: Sexy Matt Helm-type theme, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Austin and Vanessa are seated at a table with Frau. WAITERS MUSIC: Sexy Matt Helm-type theme The camera pushes in with small amplitude at slow speed. Dr. Evil (S1) says: [English] I think you'll enjoy the food. I have the best chef in the world. His name is Ezekial. He's made of seventy-five percent plastic. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: -- the invocation of LaMagra reaches critical mass. Frost is consumed\nSetting: derived from Blade\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. TEMPLE SANCTUARY - NIGHT\nACTION: -- the invocation of LaMagra reaches critical mass. Frost is consumed by a hellish force, unable to withstand the onslaught of spirit\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade: -- the invocation of LaMagra reaches critical mass. Frost is consumed by a hellish force, unable to withstand the onslaught of spirit, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. -- the invocation of LaMagra reaches critical mass. Frost is consumed by a hellish force, unable to withstand the onslaught of spirit The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: CLOSE ON the clock radio from which the voice emanates.\nSetting: derived from Hitchhiker'S Guide To The Galaxy\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. ARTHUR DENT'S BEDROOM - MORNING 2\nACTION: CLOSE ON the clock radio from which the voice emanates. A hand shuts it off. ARTHUR DENT is awake. 30's. Kind\nDIALOGUE — Arthur: What do you mean he's gone out to lunch? You just said he was still at breakfast!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Hitchhiker'S Guide To The Galaxy: CLOSE ON the clock radio from which the voice emanates. A hand shuts it off. ARTHUR DENT is awake. 30's. Kind, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. CLOSE ON the clock radio from which the voice emanates. A hand shuts it off. ARTHUR DENT is awake. 30's. Kind The camera pushes in with small amplitude at slow speed. Arthur (S1) says: [English] What do you mean he's gone out to lunch? You just said he was still at breakfast! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Ride through every village and town\nSetting: derived from Mrs. Teasdale\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Ride through every village and town wake every citizen up hill and down\nDIALOGUE — Groucho: There'll be two lamps in the steeple if they're coming by land and one lamp if they're coming by sea.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Mrs. Teasdale: Ride through every village and town wake every citizen up hill and down, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Ride through every village and town wake every citizen up hill and down The camera pushes in with small amplitude at slow speed. Groucho (S1) says: [English] There'll be two lamps in the steeple if they're coming by land and one lamp if they're coming by sea. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: \"Die Hard\"\n\nSCREENPLAY EXCERPT:\nAnytime you want to go home,\n Sergeant...consider yourself dismissed.\n\n They lock eyes.\n\n POWELL\n No Sir. You couldn't drag me away.\n\n HANS' VOICE"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Anytime you want to go home, Sergeant...consider yourself dismissed.\nDIALOGUE — Powell: No Sir. You couldn't drag me away.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"Die Hard\": Anytime you want to go home, Sergeant...consider yourself dismissed., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Anytime you want to go home, Sergeant...consider yourself dismissed. The camera pushes in with small amplitude at slow speed. Powell (S1) says: [English] No Sir. You couldn't drag me away. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: I want to see Moses and Rafe. And\nSetting: derived from Untitled\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: I want to see Moses and Rafe. And Masha. I think that will help me.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Untitled: I want to see Moses and Rafe. And Masha. I think that will help me., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. I want to see Moses and Rafe. And Masha. I think that will help me. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Diane DeConn? You saw Diane DeConn!\nSetting: derived from The Marine Biologist\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Diane DeConn? You saw Diane DeConn! Yeah! How'd she look.\nDIALOGUE — Jerry: Something huh?\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from The Marine Biologist: Diane DeConn? You saw Diane DeConn! Yeah! How'd she look., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Diane DeConn? You saw Diane DeConn! Yeah! How'd she look. The camera pushes in with small amplitude at slow speed. Jerry (S1) says: [English] Something huh? The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: MacReady, cursing, rips the speakers off the wall.\nSetting: derived from \"The Thing\"\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. REC ROOM\nACTION: MacReady, cursing, rips the speakers off the wall. What are they doing out there?!\nDIALOGUE — Macready: What are they doing out there?!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from \"The Thing\": MacReady, cursing, rips the speakers off the wall. What are they doing out there?!, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. MacReady, cursing, rips the speakers off the wall. What are they doing out there?! The camera pushes in with small amplitude at slow speed. Macready (S1) says: [English] What are they doing out there?! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Dark City\n\nSCREENPLAY EXCERPT:\nINT. BATHROOM\n\n Sachiko locks the door behind her and steps to the middle of\n the room, lifts a small round rug off the tiled floor.\n\n Underneath is a TRAP-DOOR. Opening it, she climbs down a\n hole.\n\n UNDER THE FLOOR\n\n In the semi-darkness she moves to a WOODEN BOX. She cranks a\n small metal handle on the box which makes the lid open. Music\n plays softly.\n\n She looks into the yawning black mouth of the box. A WORM-LIKE\n apparition threads its way out of the hole, wraps about her\n thigh. She breaths deeply, shuts her eyes."}, {"role": "assistant", "content": "## SCENE 1 — INT. BATHROOM\nACTION: Sachiko locks the door behind her and steps to the middle of the room, lifts a small round rug off the tiled floor.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Dark City: Sachiko locks the door behind her and steps to the middle of the room, lifts a small round rug off the tiled floor., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Sachiko locks the door behind her and steps to the middle of the room, lifts a small round rug off the tiled floor. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Script Date 5/25/94\n\nSCREENPLAY EXCERPT:\nYes... No... Not until I put someone on\n him. Listen, I need you to send me\n someone who can speak Hungarian. He's\n awake and talking like a Thai hooker...\n How should I know? Get me someone who can\n talk to him -\n\nBaer is suddenly distracted by something Kovash has said. In\nthe middle of a long string of unintelligible dialect, he has\nspouted two words that have gotten Baer's attention.\n\nHe turns and looks down at the tattered man in the bed.\nKovash realizes Baer is listening and says the two words\nagain.\n\n KOVASH\n Keyser - Soze.\n\n BAER\n What?\n\nHe waves his hand, gesturing for Kovash to say it again.\n\n KOVASH\n Keyser - Soze.\n\n BAER\n No shit?"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Yes... No... Not until I put someone on him. Listen, I need you to send me\nDIALOGUE — Kovash: Keyser - Soze.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Script Date 5/25/94: Yes... No... Not until I put someone on him. Listen, I need you to send me, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Yes... No... Not until I put someone on him. Listen, I need you to send me The camera pushes in with small amplitude at slow speed. Kovash (S1) says: [English] Keyser - Soze. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Joel finds himself eating Chinese food and sitting across\nSetting: derived from Int. Publishing House Reception Area - Day\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. CHINESE RESTAURANT - NIGHT\nACTION: Joel finds himself eating Chinese food and sitting across from Clementine. He is ragged and jarred.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int. Publishing House Reception Area - Day: Joel finds himself eating Chinese food and sitting across from Clementine. He is ragged and jarred., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Joel finds himself eating Chinese food and sitting across from Clementine. He is ragged and jarred. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: A scene in IZombie illustrating the trope 'IronicName'. Liv Moore in iZombie is a zombie named \"Live More\".\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. SETTING - DAY\nACTION: Characters enact a brief beat embodying IronicName.\nSHOT: medium shot, static shot\nSTORYBOARD_PROMPT: cinematic still for IZombie, IronicName, live-action\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, a medium shot introduces the situation for IronicName. The camera holds a static shot as the beat resolves into the final composition of Picture 2.\n\noverall_soundscape: Soft room tone.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Blade Runner\n\nSCREENPLAY EXCERPT:\nINT. INSPECTOR BRYANT'S OFFICE - NIGHT 11\n\n The INSPECTOR is in his fifties. The deep creases in\n his face, the broken capillaries in his nose say\n brawler, spoiler, drinker, but the diplomas on the\n wall say something else. Bryant's kneeled at his safe\n trying to open it. Deckard it sitting on the edge of\n the desk reading the print-out.\n\n DECKARD (V.O.)\n They escaped from the colonies\n two weeks ago. Killed twenty-\n three people and jumped a shuttle.\n An aerial patrol found the ship\n in the desert. No crew.\n\n Bryant gets the safe open and brings out a bottle of\n whiskey.\n\n DECKARD (V.O.)\n Bryant's got a liver problem. A\n couple years back he handed me a\n bottle and said have a drink for\n another man. I been drinking\n for him ever since.\n\n Deckard sets down the report and takes the shot Bryant\n just poured for him.\n\n DECKARD\n Six, huh?\n\n BRYANT\n Five. Three nights ago one of\n them managed to break into the\n Tyrell Corporation. Killed two\n guards and got as far as the\n Genetic Sector before he got\n fried going through an electro-\n field.\n\n "}, {"role": "assistant", "content": "## SCENE 1 — INT. INSPECTOR BRYANT'S OFFICE - NIGHT 11\nACTION: The INSPECTOR is in his fifties. The deep creases in his face, the broken capillaries in his nose say\nDIALOGUE — Deckard: They escaped from the colonies two weeks ago. Killed twenty- three people and jumped a shuttle. An aerial patrol found the ship in the desert. No crew.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Blade Runner: The INSPECTOR is in his fifties. The deep creases in his face, the broken capillaries in his nose say, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. The INSPECTOR is in his fifties. The deep creases in his face, the broken capillaries in his nose say The camera pushes in with small amplitude at slow speed. Deckard (S1) says: [English] They escaped from the colonies two weeks ago. Killed twenty- three people and jumped a shuttle. An aerial patrol found the ship in the desert. No crew. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: DUKE enters. The door hits something with a thud.\nSetting: derived from Fear & Loathing In Las Vegas\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. HOTEL FLAMINGO SUITE - AFTERNOON\nACTION: DUKE enters. The door hits something with a thud. A 16-year-old GIRL with the aura of an angry Pit Bull.\nDIALOGUE — Duke: You degenerate pig!\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Fear & Loathing In Las Vegas: DUKE enters. The door hits something with a thud. A 16-year-old GIRL with the aura of an angry Pit Bull., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. DUKE enters. The door hits something with a thud. A 16-year-old GIRL with the aura of an angry Pit Bull. The camera pushes in with small amplitude at slow speed. Duke (S1) says: [English] You degenerate pig! The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Rewrite this screenplay beat as a single ~5s MiniMax-H3 FL2VA scene for Backlot (first storyboard panel → last panel).\n\nSOURCE TITLE: Bride Of Leprechaun\n\nSCREENPLAY EXCERPT:\nAye. Cry as you may, cry as you\n might. It's going to be one hell of\n a wedding night.\n\n The Leprechaun steps forward to take Bridget's hand, when...\n\n Bridget is pushed to the side and Cody comes out of the\n shadows, behind her. He holds his wrought iron spear in front\n of him. With one smooth, forceful thrust, he jabs home the\n weapon, skewering a surprised Leprechaun through the heart.\n\n CODY\n In your dreams, buddy.\n\n134 INT. INNER LAIR - FOLLOWING 134\n\n The Leprechaun staggers back, in shock, into the inner lair.\n He conjures up his gold coin and looks at it, questioningly.\n He looks back at Cody. The Leprechaun begins to glow and\n vibrate. The wrought iron burns the Leprechaun's insides. He\n writhes and twists in pain, staggering back.\n\n LEPRECHAUN"}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Aye. Cry as you may, cry as you might. It's going to be one hell of\nDIALOGUE — Cody: In your dreams, buddy.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Bride Of Leprechaun: Aye. Cry as you may, cry as you might. It's going to be one hell of, detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Aye. Cry as you may, cry as you might. It's going to be one hell of The camera pushes in with small amplitude at slow speed. Cody (S1) says: [English] In your dreams, buddy. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}
{"messages": [{"role": "system", "content": "You write ONE MiniMax-H3 FL2VA scene beat for Backlot. Output fields: ACTION, SHOT, STORYBOARD_PROMPT, H3_MODE, H3_VIDEO_PROMPT (alignment line + integrated_multimodal_description + overall_soundscape + non_diegetic_music), LORA, AUDIO, DURATION 5. Prefer a single continuous shot from Picture 1 to Picture 2. Dialogue only inside [Language] ...."}, {"role": "user", "content": "Premise: Cute cat. What's its name?\nSetting: derived from Int: Bedroom. Early-Morning Hours\nTone: cinematic\nWrite SCENE 1 now (H3 FL2VA, DURATION 5)."}, {"role": "assistant", "content": "## SCENE 1 — INT. LOCATION - DAY\nACTION: Cute cat. What's its name? The CUSTOMER lets it sink in, and then leaves in a huff.\nSHOT: medium shot, push in with small amplitude at slow speed\nSTORYBOARD_PROMPT: cinematic still from Int: Bedroom. Early-Morning Hours: Cute cat. What's its name? The CUSTOMER lets it sink in, and then leaves in a huff., detailed environment, live-action, film lighting\nH3_MODE: FL2VA\nH3_VIDEO_PROMPT:\nHow the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, medium shot establishing the scene. Cute cat. What's its name? The CUSTOMER lets it sink in, and then leaves in a huff. The camera pushes in with small amplitude at slow speed. The framing begins on the composition of Picture 1 and continuously evolves until it settles into the composition of Picture 2.\n\noverall_soundscape: Room tone and soft environmental ambience matching the location.\n\nnon_diegetic_music: N/A\nLORA: none\nAUDIO: none\nDURATION: 5"}]}