docs/video_script.md

Sozai — Product Overview Video

A walkthrough script and narrative guide for a short product overview of Sozai.

Format: ~2.5–3 minutes. Mostly voiceover over screen recordings and a few PowerPoint-style slides. Two narrators (A and B) taking turns — conversational, warm, informative. Not a dramatized film; think "two people who built this giving you the tour." Keep it relaxed and clear.

---

The one-sentence pitch

Sozai turns the photos buried in your camera roll into watercolour memories you develop with your companions: like a shared, modern darkroom.

What the name means: 素材 (sozai) is Japanese for "raw material / ingredients." Your snapshots are the raw material; the memory is what you make from them. Worth saying once in the video — it frames everything.

The core metaphor the whole product runs on: analog film photography. You don't "post," you develop. There's a real darkroom, chemical trays, prints on a line, and you shake each one to reveal it. That metaphor is the brand — lean on it.

---

The two narrators

• A — frames the "why" and the feeling (the camera-roll problem, the togetherness).
• B — drives the "how" and the features (rooms, develop, views, the AI parts).

They trade off naturally. Lines are written to be read plainly, not performed.

---

The script

Each beat = [slide or screen cue] + the spoken lines. Screen names match the real app so you can record straight from it.

1. Open — the problem (slide)

[SLIDE: a dense grid of phone photo thumbnails, greyed out. Sozai bird logo in corner.]

A: Most of us take hundreds of photos on a trip — and then never look at them again. They just sit in the camera roll.

B: That's the problem we kept running into. So we built Sozai — a place to actually do something with those photos, with the people you took them with.

[SLIDE: the word "素材 — raw material" fades in under the Sozai logo.]

A: "Sozai" means "raw material." Your photos are the raw material. The memory is what you make out of them.

2. Getting in — rooms (screen: landing page)

[SCREEN: the Sozai landing page — bird, map lines, "build memories together".]

B: When you open Sozai, you can start a room and invite someone, join a room with a short code, or just go solo.

[SCREEN: click "make a room" → a short share code appears. Cut to a second window: someone types the code into "enter room code" and clicks join. Two named cursors now move on the same screen.]

A: Here's the part we love — it's live. Two people, two cities, same room. You can see each other's cursors and edits as they happen.

B: Or pick "continue without a room" and it's just you. Same app, no invite needed.

3. Adding photos + captions (screen: Develop Photos)

[SCREEN: the "Develop Photos" page — photo stack on the left, Title / Caption / When & Where on the right.]

B: You drop in your photos — up to eight at a time — and give them a title and a caption.

[SCREEN: click the mic icon on the caption field; speak; text appears.]

A: You can type, or just talk. The mic transcribes what you say, so captioning a photo feels more like telling someone about it.

[SCREEN: click "Auto-generate"; the stages tick by: "Reading the light… Finding the words…"; a caption fills in.]

B: Or let the AI take a look and suggest one. It's actually looking at the photo — so the title, caption, and tags match what's really in the picture.

4. Choosing how it renders (screen: Views grid)

[SCREEN: the "Views" picker — 2D Map, 3D Globe, Canvas, Scrapbook, Gallery, Timeline.]

A: Before you develop, you choose how your memories will be laid out — on a map, a spinning globe, a scrapbook, a gallery, or a timeline.

B: And this isn't a one-time choice — you can switch any time. Hold that thought, because it's the part we care about most.

5. The darkroom — the heart (screen: Darkroom)

[SCREEN: click "Develop Photos →". The Darkroom: four glowing trays — Developer, Stop Bath, Fixer, Wash. Status line: "giving your photos an acid rinse…"]

B: When you hit develop, your photos go through a darkroom. This is where each one is painted into a watercolour.

[SCREEN: the prints hang on a line. Status: "shake each print to reveal it." A blank print reads "developed — shake me"; it's shaken and the watercolour appears.]

A: And then you reveal them — you literally shake each print until the picture comes through. It's a small thing, but it makes you slow down and actually look.

[SCREEN: optional — show one original photo next to its finished watercolour.]

B: Same photo, now a painting. That's the "develop."

6. The payoff — one memory, many ways (screen: 2D Map, then cycle)

This is the most important beat in the video. The core feature isn't any single view — it's that the same memories can be reimagined in many ways, interactively. Anchor on the 2D Map (it's the most stable to record), then briefly cycle the rest.

[SCREEN: developed watercolours pinned on the 2D Map — the stable hero shot. Pan and zoom around the pins; drag a polaroid to reposition it.]

A: Once they're developed, your memories land in the view you picked. Here they're pinned on the map, right where you took them — and you can move them around, zoom in, make it yours.

[SCREEN: switch views and briefly cycle — 3D Globe → Scrapbook scroll → Timeline → Gallery — the same photos re-laid-out each time, a beat on each.]

B: And this is really the heart of Sozai: the same memories, reimagined as many ways as you like. Spin them on a globe, scroll them as a scrapbook, walk them down a timeline, browse them as a gallery — one click each.

A: Nothing's locked in. You keep rearranging until it feels like yours.

7. Together + close

[SCREEN: the share code is sent to a third person; a new named cursor joins the room.]

A: And because it's a room, you can hand the code to someone else and keep building it together — even after the trip's over.

[SLIDE: Sozai logo + "develop your memories together" + "素材 — the raw material of memory".]

B: That's Sozai. Bring a few photos back to life — together.

---

Optional add-ons (for a longer / more technical cut)

• Under the hood slide. One slide for a technical audience: photos are painted by FLUX.2-klein with a watercolour style, captions come from the MiniCPM-V vision model, voice is transcribed by NVIDIA NeMo, and every uploaded photo passes a safety check first.
• LLM Trace panel. Sozai has a built-in trace view (Phoenix) showing every model call, its latency, and tokens — good for a "yes, it's real and observable" beat.
• The hidden pet. A little sprite that follows your cursor around the scrapbook — a nice "did you catch that?" detail if the tone allows.

---

Who it's for (personas — for framing, not necessarily on screen)

• Two friends after a trip (the hero use case; our sample photos are a Toronto trip). They develop the trip together from two cities.
• Family across distance. One shared album spanning continents — the globe view shines here.
• The solo journaler. "Continue without a room" — a quiet afternoon turning last month's photos into a watercolour scrapbook.

---

High-leverage copy recommendations

Small wording changes that tighten the story. Each is current → suggested. Listed by impact. (Not yet applied — review first.)

1. Add a one-line "what is this" under the landing tagline. Today the landing shows only Sozai / build memories together with no explanation. → add: "Turn your camera roll into watercolour memories — together."

2. Make "develop" the single signature verb. The app already lives on it (Develop Photos, The Darkroom, developing…). Consider the tagline: build memories together → "develop your memories together" (develop = both film and grow).

3. Pay off the name once, in-app. Most people don't know 素材. → add somewhere quiet: "Sozai (素材) — the raw material of memory."

4. Sharpen the reveal line. developed — shake me → "still wet — give it a shake."

5. Warm up the empty state. Wherever a room/scrapbook starts empty: → "Nothing developed yet. Drop in a few photos and let's bring them back."

6. Keep the verb on the solo path. CONTINUE WITHOUT A ROOM → → "develop on your own →."

7. Give the full view a closing line. On the globe/scrapbook once populated: → "Every pin is a place you've been."

Leave the existing playful copy as-is — "giving your photos an acid rinse," "Reading the light… / Letting it soak… / Finding the words…", and the tray names. That voice is the brand; the changes above just extend it.

---

Maps

Decided: the 2D Map is the anchor for the §6 payoff (most stable to record). The 3D Globe and the other views are shown as a quick cycle right after, to make the point that the same memories can be reimagined many ways — the interactivity is the core feature, not any single view.

Still to finalize when map footage arrives:

• the recording pace of the cycle (how long to hold each view),
• whether the pin-drop / drag-a-polaroid-on-the-map interaction gets its own spoken line or just plays under A's "make it yours" line.