Spaces:
Configuration error
Configuration error
Commit History
-
fix(denoise): rewrite step 9 β extract audio only then remux, avoids ffmpeg 6.1 filter_complex video mapping bug
AIGoose15ce6d7 -
fix(step9): split denoise+silenceremove into 2 passes β anlmdn chained filter crashes ffmpeg 6.1.1
AIGoosed09c1d0 -
fix(v3): burn captions in ffmpeg, denoise voice, jump-cut pauses, title card before footage, fix headline overlap
AIGoosee7002b8 -
fix: merge duplicate env: blocks in Upload step β was causing startup_failure
AIGooseefd8a15 -
feat: Bright Studio style β Space Grotesk 100px top, captions, trim, music, navy border rule
AIGooseb10b75c -
perf: 720x1280@24fps + chrome-headless-shell + each clip on own track + system fonts (target render 5-8min)
AIGoosedf74a69
-
fix: HyperFrames needs project directory with index.html, not direct file path
AIGoose0ddf953 -
feat: add HyperFrames testimonial pipeline β composition, workflow, patcher, README
AIGoose0624752 -
feat: add .gitignore
AIGoose96ae60b -
feat: hyperframes-video skill v3 β watch_gdrive.py based on bradautomates/claude-videoβ
Dee FerdinandClaude can now WATCH Google Drive videos: - Auto-scaled frame extraction (same budget logic as claude-video) - Groq whisper-large-v3 -> OpenAI -> HF Inference transcription waterfall - Indonesian filler word cleaning (eh, umm, itu, ya, kan, gitu) - 6-marker quote scoring (outcome/surprise/transform/recommend) - Outputs frame paths for Claude Read tool + timestamped transcript0084100 -
feat: hyperframes-video skill v2.0.0β
Dee Ferdinand- Act-first pipeline: scan β transcribe β analyze β brief β confirm β build - Whisper transcription + Indonesian filler word cleaning + best quote extraction - 3 visual styles: Dark Premium / Warm Documentary / Bold Editorial - Auto-style assignment by content energy + client type - Direct GitHub push + HF render integration via Composio - Zero question-asking before scanning assets - Duration intelligence based on raw footage lengthef517dd -
chore: trigger HF space rebuild for chrome-headless-shell + video prefetch fix
AIGooseba8e3d7 -
fix: pre-download remote video URLs before render; use chrome-headless-shell if available
AIGooseef62cee -
fix: install chrome-headless-shell for HyperFrames optimized render path
AIGoose2d3b83d -
feat: hyperframes-video skill v2.0 β GitHub/HF pipeline + video transcription
AIGoose9563efc -
feat: fix black frames + ElevenLabs lo-fi music generation + YuE fallback + complete skill updateβ
Dee FerdinandBlack frame fix: - Root cause: tl.to({},{duration:TOTAL},0) conflicts with GSAP β use tl.set({},{},TOTAL) instead - Add explicit data-duration on root div - Add 0.5s safety buffer: scenes end 0.3s before next starts (no overlap) - Volume levels from coleam00 audio-design.md research: music=0.07-0.12, narration=1.0 ElevenLabs music (primary): - client.music.compose() with model_id='music_v2' - Prompt: 'Lo-fi upbeat instrumental, jazzy piano, warm vinyl crackle, positive energy, 85-95 BPM' - music_length_ms = duration * 1000 + 5000 (always generate 5s extra for safety) - Requires ELEVENLABS_API_KEY env var in HF Space secrets YuE fallback: - Lightweight API wrapper if ElevenLabs key not available - Direct HTTP to YuE inference endpoint Transition rules from coleam00 transitions.md: - EXIT ANIMATIONS BANNED (except final scene) β transitions ARE the exit - Scene content must be fully visible when transition starts - 0.3-0.5s transitions for medium energy (corporate) - autoAlpha not opaβ¦1f05b41 -
feat: complete skill rewrite from 3 repos + audio fix + Dee brand fonts + 2x kinetic captionsβ
Dee FerdinandAudio root causes fixed: 1. Video audio: remove `muted` from video element + use data-volume β HyperFrames needs unmuted to extract audio during render 2. Music: replace dead FMA URLs with working ccMixter/bensound direct URLs 3. Silence fallback: generate proper silence mp3 not empty file Brand updates: - Fonts: Space Grotesk (headings) + Plus Jakarta Sans (body) via Google Fonts - Caption sizes 2x: hook=128-160px, energy=120-144px, CTA=112px - Kinetic motion: varied eases, stagger, scale pops, y-slide variety - Dark Premium palette: #0d0d1f base, #7C6FE0 accent, #9EF0C8 energy green Skill learned from: - nateherkai/hyperframes-student-kit (13 projects, motion principles, visual styles) - heygen-com/hyperframes-launch-video (production patterns, audio HANDOFF fix) - robonuggets/hyperframes-helper (glass cards, composition template, recipes)c0f1ad6 -
feat: complete rewrite β correct HyperFrames compositions with GSAP + class=clip + window.__timelinesβ
Dee Ferdinand- Add SKILL.md: HyperFrames production skill with all rules, patterns, vocabulary - Rewrite composer.js: generates proper HyperFrames HTML (GSAP timelines, clip classes, registered timelines) - Testimonial: 30s workflow with hook/proof/story/energy/cta/brand structure - All 5 hard rules enforced: class=clip, paused timeline, no media control, no Math.random, tl.set extend - GSAP CDN loaded from jsdelivr, all animations use correct easing vocabulary - Video wrapped in divs (never animate video element directly) - Lower thirds, stat counter, brand end card patternsa2e4ffd
-
fix: correct hyperframes CLI β render <dir> -o <output>, PRODUCER_HEADLESS_SHELL_PATH for Chrome, workers=1 for cpu-basic
Dee Ferdinandd3cc583 -
fix: run hyperframes from inside compDir with relative paths (cwd=compDir, ./index.html, -o ./render.mp4)
Dee Ferdinand5a4b7f7 -
fix: correct HyperFrames CLI syntax β render takes file path + -o flag, not --width/--height args
Dee Ferdinand017f3f3 -
fix: music URLs (use GitHub-hosted royalty-free MP3s, Pixabay CDN blocks server requests) + fix HyperFrames CLI pathβ
Dee Ferdinand- Replace Pixabay CDN URLs with GitHub-hosted ccMixter/Free Music Archive tracks - Fallback: generate silent audio with ffmpeg if all downloads fail - Fix renderer to use correct npx hyperframes render path - Add --chrome-path flag for Chromium in Dockerb132b92 -
fix: pass HF_TOKEN as Space secret in deploy workflow so uploader.py can authenticate
Dee Ferdinandd83ad14 -
feat: upload rendered MP4s to HuggingFace Dataset repo instead of local downloadβ
Dee Ferdinand- Renders push to AIgoose/video-renders dataset via huggingface_hub - Studio UI shows HF link + preview instead of download button - CommitScheduler auto-batches uploads every 30s - requirements.txt updated with huggingface_hub - lib/uploader.js handles the Python bridge upload3cd9013 -
docs: update README β rename Space to hyperframes-video-studio, add live URL
Dee Ferdinand0eb1d42 -
ci: update HF Space name to hyperframes-video-studio
Dee Ferdinandc718a43 -
ci: add HuggingFace auto-deploy workflow + git configs
Dee Ferdinanddeb2826 -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)410447e -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)f4ac095 -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)b84a129 -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)f88cf50 -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)63afd10 -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)1b305e8 -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)b41b9bd -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)10992d3 -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)345fbd8 -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)a60741e -
feat: init Dee Video Studio β HyperFrames cloud render for testimonial/teaser/trailer videosβ
Dee Ferdinand- 4 workflows: testimonial (75s), teaser (30s), trailer (60s), community (60s) - 6 royalty-free Pixabay music tracks auto-selected by workflow - WebSocket real-time render progress - Dark-mode Studio UI with drag-drop upload - HuggingFace Docker Space ready (port 7860)aace184