File size: 9,645 Bytes
241eca4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
# Sozai β€” 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** β€” 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 β€” 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 β€” lean on it.

---

## The two narrators

- **A** β€” frames the "why" and the feeling (the camera-roll problem, the togetherness).
- **B** β€” 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 β€” 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 β€” 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 β€” a place to
> actually *do something* with those photos, with the people you took them with.

**[SLIDE: the word "素材 β€” 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 β€” rooms (screen: landing page)

**[SCREEN: the Sozai landing page β€” 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" β†’ 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 β€” 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 β€” photo stack on the left, Title / Caption /
When & Where on the right.]**

> **B:** You drop in your photos β€” up to eight at a time β€” 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 β€” 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 β€” 2D Map, 3D Globe, Canvas, Scrapbook, Gallery,
Timeline.]**

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

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

### 5. The darkroom β€” the heart (screen: Darkroom)

**[SCREEN: click "Develop Photos β†’". The Darkroom: four glowing trays β€”
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 β€” shake me"; it's shaken and the watercolour appears.]**

> **A:** And then you reveal them β€” 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 β€” show one original photo next to its finished watercolour.]**

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

### 6. The payoff β€” 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 β€” 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 β€” 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 β€” and you can move them
> around, zoom in, make it yours.

**[SCREEN: switch views and briefly cycle β€” 3D Globe β†’ Scrapbook scroll β†’ Timeline
β†’ Gallery β€” 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 β€” 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 β€” even after the trip's over.

**[SLIDE: Sozai logo + "develop your memories together" + "素材 β€” the raw material of memory".]**

> **B:** That's Sozai. Bring a few photos back to life β€” 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 β€” good for a "yes, it's real and observable" beat.
- **The hidden pet.** A little sprite that follows your cursor around the scrapbook β€”
  a nice "did you catch that?" detail if the tone allows.

---

## Who it's for (personas β€” 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 β€” the globe view
  shines here.
- **The solo journaler.** "Continue without a room" β€” 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 β†’ suggested**.
Listed by impact. (Not yet applied β€” 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.
   β†’ add: **"Turn your camera roll into watercolour memories β€” 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` β†’ **"develop your memories together"** (*develop* = both
   *film* and *grow*).

3. **Pay off the name once, in-app.** Most people don't know 素材.
   β†’ add somewhere quiet: **"Sozai (素材) β€” the raw material of memory."**

4. **Sharpen the reveal line.** `developed β€” shake me`
   β†’ **"still wet β€” give it a shake."**

5. **Warm up the empty state.** Wherever a room/scrapbook starts empty:
   β†’ **"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 β†’`
   β†’ **"develop on your own β†’."**

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

> Leave the existing playful copy as-is β€” *"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 Β§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 β€” 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.