Skip to content

armgit / kokon

Public Imported from GitHub
0

Edit photos by describing the change, on your own GPU. A free, open-source alternative to Nano Banana and ChatGPT image editing: photos stay on your computer, no daily limit. One installer for Windows, Mac and Linux sets up the best open model (Qwen-Image 2.1). MCP server for AI agents.

Merge branch 'video-use-cases' into server-0.11

Gabriele Lozuponeb582abdde98b
diff --cc marketing/README.md
index 3b66e4b,f7ceff0..bbc9e15
--- a/marketing/README.md
+++ b/marketing/README.md
@@@ -50,25 -50,34 +50,53 @@@ D:/AI/tts/.venv/Scripts/python audio/ge
npx remotion render Update010 out/kokon-0.10-update.mp4 # in video/, 1920×1080
```
- **"Kokon 0.11" video** (`Update011`, ~76 s): Kokon on other devices. The site's "Other devices"
++**"Kokon 0.11" video** (`Update011`, ~77 s): Kokon on other devices. The site's "Other devices"
+scene (`src/app/Devices011.tsx`: the computer with the GPU, a MacBook and an iPhone, frames from
+Devices.css, MIT, in `src/app/devices.css`), the three setup steps, the phone in use, safe by
+default, the server and Docker, the end card. The app on screen is the app itself: Settings →
+Other devices in the desktop window, and the browser UI in the phone layout (`loaders/phone-container.cjs`
+turns the 720 px media query into a container query; `src/mocks/platform.ts` switches desktop and
+browser per window). The QR codes in the camera shot and on the end card open the setup guide.
+Clock `src/app/timeline011.ts` (bars of the music, shots, voice cues), script `audio/voiceover-0.11.md`,
- music one ACE-Step take played whole at one level (`gen_music.py v011-steady`, steady deep house at
- 118 BPM), new effects `make_sfx.py 011` (three variants each in `public/sfx/011`, the one played is
- picked in `sound.tsx`).
++music the launch film's house-4 played whole at one level (faded out over the end card), new
++effects `make_sfx.py 011` (three variants each in `public/sfx/011`, the one played is picked in
++`sound.tsx`).
+
+```powershell
+D:/AI/tts/.venv/Scripts/python audio/gen_voice.py audition-011 --seeds 3 # voice/update-0.11/audition-<stamp>/
+D:/AI/tts/.venv/Scripts/python audio/gen_voice.py build-011 --from voice/update-0.11/audition-<stamp> --pick 1.0:17,...
- python audio/gen_music.py v011-steady # takes/v011-steady-<stamp>-N.wav (ACE-Step API running)
+npx remotion render Update011 out/kokon-0.11.mp4 # in video/, 1920×1080
+```
+
++
+ **"Use cases" video** (`UseCases`, ~87 s, English, narrated, for the site and YouTube): it opens
+ on why Kokon (private, unlimited), then thirteen everyday edits (photobombers removed, a blink
+ fixed, an old photo restored and coloured, a missing person added, a haircut and a jacket tried on,
+ a room tidied, a home redone, a headshot, an invitation with text), each shown as the before photo,
+ the prompt typed into the prompt box (the app's `.composer` markup and styles), a wipe to the after
+ landing on the spoken word, and a caption. Every image was made with Kokon (`public/img/usecases/`,
+ web copies of `%USERPROFILE%\Pictures\Kokon\use-cases-2k`); the video says so on screen. No camera
+ moves: hard cuts on the beat, wipes and type. Its clock is `src/app/timelineUseCases.ts` (beat
+ lengths, moods, where each voice line starts, the word each wipe lands on), its scene
+ `src/scenes/UseCases.tsx`, its script `audio/voiceover-usecases.md`: one narrator with a tone per
+ section, the launch voice on Chatterbox, and Chatterbox Turbo cloned from it for the playful lines
+ (with its `[chuckle]` and `[happy]` tags). The music is one ACE-Step track with a calm piano part,
+ cut on bar lines so that part covers the serious beats and the groove returns on the flamingo. The
+ YouTube thumbnail is `UseCasesThumb`.
+
+ ```powershell
+ D:/AI/tts/.venv/Scripts/python audio/gen_voice.py audition-usecases --seeds 4 # voice/usecases/audition-<stamp>
+ D:/AI/tts/.venv/Scripts/python audio/gen_voice.py build-usecases --from audio/voice/usecases/audition-<stamp> --pick 1.0:14,...
+ python audio/gen_music.py usecases # takes/usecases-<stamp>-N.wav (ACE-Step API running)
+ D:/AI/tts/.venv/Scripts/python audio/make_update_music.py usecases # video/public/music-usecases.m4a
+ python audio/make_sfx.py pop swipe deep # only this video's effects
+ npx remotion render UseCases out/usecases-1.mp4 # in video/, 1920×1080
+ npx remotion still UseCasesThumb out/usecases-thumb.png # 1280×720
+ ```
+
+ The word each wipe lands on comes from `src/app/voice-usecases-words.json` (Whisper word times of
+ the built lines, whisper-small.en): regenerate it after a new build.
+
`src/app/timeline.ts` is the film's clock: scene lengths, the frame where each voice-over sentence
plays, the frames of the actions the voice names (Remove, Add, the montage cuts) and the music's
beat grid. Retime there and pictures, voice and music move together.
diff --cc marketing/video/src/Root.tsx
index 1fbce8b,b3b3c8f..891dd8b
--- a/marketing/video/src/Root.tsx
+++ b/marketing/video/src/Root.tsx
@@@ -13,9 -13,8 +13,10 @@@ import { Update010, UPDATE_FRAMES } fro
import { LENGTH } from "./app/timeline";
import { REELS, ReelScene, reelFrames } from "./scenes/Reels";
import { AppShot, PH_SLIDES } from "./scenes/ProductHunt";
+import { Probe011 } from "./scenes/Probe011";
+import { Update011, UPDATE011_FRAMES } from "./scenes/Update011";
import { VREELS, VReelScene, vreelFrames } from "./scenes/VReels";
+ import { UseCases, USECASES_FRAMES, USECASES_IMAGES, UseCasesThumb } from "./scenes/UseCases";
const withPreload = (C: ComponentType) => () => (
<>
@@@ -29,8 -28,9 +30,11 @@@ export function Root()
return (
<>
<Composition id="KokonLaunch" component={withPreload(Main)} durationInFrames={MAIN_FRAMES} {...S} />
+ <Composition id="Update011" component={withPreload(Update011)} durationInFrames={UPDATE011_FRAMES} {...S} />
+ <Composition id="Probe011" component={withPreload(Probe011)} durationInFrames={1} {...S} />
+ <Composition id="UseCases" component={() => (<><Preload extra={USECASES_IMAGES} /><UseCases /></>)} durationInFrames={USECASES_FRAMES} {...S} />
+ <Composition id="UseCasesHouse4" component={() => (<><Preload extra={USECASES_IMAGES} /><UseCases music="music-usecases-house4.m4a" /></>)} durationInFrames={USECASES_FRAMES} {...S} />
+ <Composition id="UseCasesThumb" component={() => (<><Preload extra={USECASES_IMAGES} /><UseCasesThumb /></>)} durationInFrames={1} fps={30} width={1280} height={720} />
<Composition id="Update010" component={withPreload(Update010)} durationInFrames={UPDATE_FRAMES} {...S} />
<Composition id="Montage" component={withPreload(Montage)} durationInFrames={MONTAGE_FRAMES} {...S} />
<Composition id="Agent" component={withPreload(Agent)} durationInFrames={AGENT_FRAMES} {...S} />