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.

video: "Kokon 0.10" update video (Update010)

Gabriele Lozupone0ab2314d1027
diff --git a/marketing/README.md b/marketing/README.md
index c252989..b70a59b 100644
--- a/marketing/README.md
+++ b/marketing/README.md
@@ -35,6 +35,21 @@ words, Prompt+ writes the prompt, the image hatches) → Region (paint, remove,
Montage (five edits, full frame) → Agent (MCP) → Hardware (Windows, macOS, Linux) → Outro. Each scene is also its own composition, so one can be rendered alone, for example
`npx remotion render Montage out/montage.mp4` (the voice and music play only in `KokonLaunch`).
+**"Kokon 0.10" update video** (`Update010`, ~41 s, for Reddit, YouTube and the site): Prompt+
+(a few words, the prompt written, the image hatching on the drop), "One model. Two jobs." (Prompt+
+runs on the Qwen3-VL 8B text encoder Qwen-Image 2.1 already loads), your own LoRAs (beta: Settings,
+then the LoRA pill), the new look, the end card. Its clock is `src/app/timeline010.ts` (beats,
+shots, voice cues), its scenes `src/scenes/Update010.tsx`, its script `audio/voiceover-0.10.md`.
+Until its own takes exist it plays the launch track's tail and no voice (`voice-0.10.json` holds
+estimated spans, marked `_provisional`).
+
+```powershell
+python audio/gen_music.py update # takes/update-<stamp>-N.wav (ACE-Step API running)
+D:/AI/tts/.venv/Scripts/python audio/gen_voice.py audition # voice/update-0.10/audition-<stamp>/
+D:/AI/tts/.venv/Scripts/python audio/gen_voice.py build --from audio/voice/update-0.10/audition-<stamp> --pick 1.0:11,2.0:11,...
+npx remotion render Update010 out/kokon-0.10-update.mp4 # in video/, 1920×1080
+```
+
`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 --git a/marketing/audio/gen_music.py b/marketing/audio/gen_music.py
index 6d92d9d..786e018 100644
--- a/marketing/audio/gen_music.py
+++ b/marketing/audio/gen_music.py
@@ -2,8 +2,11 @@
Start the server first (D:/AI/ACE-Step-1.5): uv run acestep-api (listens on :8001)
Then: python gen_music.py [style ...] -> takes/<style>-<n>.wav
+ python gen_music.py update -> takes/update-<stamp>-<n>.wav (the "Kokon 0.10" video)
-Every take is instrumental and a little longer than the video (~85 s), so it can fade out.
+Every take is instrumental and a little longer than its video (~85 s, or ~44 s for "update"), so
+it can fade out. Takes of the launch styles keep their old names; "update" takes get a time stamp,
+so no take is ever written over (house-4 is the launch film's music).
"""
import json
import sys
@@ -37,7 +40,17 @@ STYLES = {
118,
"A Minor",
),
+ # The 0.10 update video (~41 s): a short build, the drop at ~13 s (the image hatches), a
+ # bright groove, a lift around 30 s and a clean ending (no fade needed).
+ "update": (
+ "deep melodic house, instrumental, short filtered intro with soft plucks and rising tension, "
+ "big satisfying drop early, warm sub bass, crisp hi-hats, dreamy bright chords, "
+ "confident and playful, a lift in the second half, clean definite ending, modern clean mix",
+ 118,
+ "A Minor",
+ ),
}
+DURATIONS = {"update": 44}
def call(path: str, body: dict | None = None) -> dict:
@@ -59,12 +72,13 @@ def generate(style: str) -> None:
"bpm": bpm,
"key_scale": key,
"time_signature": "4",
- "audio_duration": DURATION,
+ "audio_duration": DURATIONS.get(style, DURATION),
"batch_size": TAKES,
"inference_steps": 8,
"audio_format": "wav",
})
task_id = task["task_id"]
+ stamp = time.strftime("%Y%m%d-%H%M%S")
t0 = time.time()
while True:
time.sleep(3)
@@ -76,12 +90,15 @@ def generate(style: str) -> None:
results = json.loads(item["result"]) if isinstance(item["result"], str) else item["result"]
OUT.mkdir(exist_ok=True)
for n, res in enumerate(results, 1):
- dest = OUT / f"{style}-{n}.wav"
+ dest = OUT / (f"{style}-{stamp}-{n}.wav" if style in DURATIONS else f"{style}-{n}.wav")
urllib.request.urlretrieve(API + res["file"], dest)
print(f"{dest.name} seed {res.get('seed_value', '?')} {res.get('metas', {})}")
print(f"{style}: {len(results)} takes in {time.time() - t0:.0f} s")
if __name__ == "__main__":
- for s in sys.argv[1:] or list(STYLES):
+ # A style must be named: running every style would write over the launch takes.
+ if not sys.argv[1:]:
+ raise SystemExit(f"usage: python gen_music.py STYLE [...] ({', '.join(STYLES)})")
+ for s in sys.argv[1:]:
generate(s)
diff --git a/marketing/audio/gen_voice.py b/marketing/audio/gen_voice.py
index a9c220e..fbc8310 100644
--- a/marketing/audio/gen_voice.py
+++ b/marketing/audio/gen_voice.py
@@ -8,6 +8,15 @@
D:/AI/tts/.venv/Scripts/python gen_voice.py splice 3 --keep 0:0,4:2 --pick 1:13,2:22 --from voice/redo-<stamp>
-> line 3 rebuilt from kept and picked sentences
+The "Kokon 0.10" update video has its own script (voiceover-0.10.md) and its own files, so the
+launch take is never touched:
+
+ D:/AI/tts/.venv/Scripts/python gen_voice.py audition [--seeds 4]
+ -> voice/update-0.10/audition-<stamp>/: takes of every sentence
+ D:/AI/tts/.venv/Scripts/python gen_voice.py build --from voice/update-0.10/audition-<stamp> --pick 1.0:11,2.0:23,...
+ -> voice/update-0.10/final-<stamp>/NN.wav, copied to
+ video/public/voice-0.10 with video/src/app/voice-0.10.json
+
redo/splice change single sentences of the approved take: `--keep NEW:OLD` reuses old sentence
OLD (cut from voice/final/raw by raw/spans.json) as new sentence NEW; the other sentences of the
line in voiceover.md get takes to audition, and `--pick NEW:SEED` chooses one each (the auditioned file itself). splice backs
@@ -67,8 +76,8 @@ MASTER = "aresample=48000,highpass=f=80,deesser=i=0.4,acompressor=threshold=-20d
LUFS = -16.0
-def script_lines() -> list[str]:
- rows = [l for l in (HERE / "voiceover.md").read_text(encoding="utf-8").splitlines() if re.match(r"^\| \d+ \|", l)]
+def script_lines(script: str = "voiceover.md") -> list[str]:
+ rows = [l for l in (HERE / script).read_text(encoding="utf-8").splitlines() if re.match(r"^\| \d+ \|", l)]
return [r.split("|")[4].strip() for r in rows]
@@ -243,6 +252,75 @@ def splice(n: int, keep: dict[int, int], pick: dict[int, int], src: Path) -> Non
print(f"{n:02d} {pos / SR:4.1f} s at {lufs(folder / f'{n:02d}.wav'):.1f} LUFS (was {target:.1f}); backup in {backup}")
+# The "Kokon 0.10" update video (voiceover-0.10.md): each line's model settings, as in FINAL.
+UPDATE = {"script": "voiceover-0.10.md", "dir": OUT / "update-0.10", "public": "voice-0.10", "json": "voice-0.10.json"}
+UPDATE_TAKES = {1: "calmer", 2: "expressive", 3: "expressive", 4: "expressive", 5: "calmer", 6: "expressive"}
+
+
+def audition(seeds: int, base: int) -> None:
+ """`seeds` takes of every sentence of the update script, read one at a time, for picking by ear."""
+ folder = UPDATE["dir"] / f"audition-{time.strftime('%Y%m%d-%H%M%S')}"
+ folder.mkdir(parents=True)
+ for n, line in enumerate(script_lines(UPDATE["script"]), 1):
+ take = UPDATE_TAKES.get(n, "expressive")
+ for i, text in enumerate(sentences(line)):
+ for k in range(seeds):
+ seed = base + 10 * i + k
+ a = say(take, text, None, seed)
+ raw = folder / f"{n:02d}-s{i}-seed{seed}.raw.wav"
+ sf.write(raw, a, SR)
+ master(raw, folder / f"{n:02d}-s{i}-seed{seed}.wav")
+ print(f"{n:02d} s{i} seed {seed} {a.size / SR:4.1f} s {text}", flush=True)
+ print(f"takes in {folder}")
+
+
+def build(src: Path, pick: dict[str, int]) -> None:
+ """The update's lines from the picked sentence takes (`pick` "LINE.SENTENCE" -> seed), 0.4 s
+ between sentences, mastered like the launch take (-16 LUFS over all lines). Each build goes to
+ a new folder; the video gets the files and their sentence spans."""
+ lines = script_lines(UPDATE["script"])
+ missing = [f"{n}.{i}" for n, l in enumerate(lines, 1) for i in range(len(sentences(l))) if f"{n}.{i}" not in pick]
+ if missing:
+ raise SystemExit(f"no take picked for {', '.join(missing)}")
+ folder = UPDATE["dir"] / f"final-{time.strftime('%Y%m%d-%H%M%S')}"
+ (folder / "raw").mkdir(parents=True)
+ spans: dict[int, dict] = {}
+ files = []
+ for n, line in enumerate(lines, 1):
+ chunks, marks, pos = [], [], 0
+ parts = sentences(line)
+ for i in range(len(parts)):
+ c, sr = sf.read(src / f"{n:02d}-s{i}-seed{pick[f'{n}.{i}']}.raw.wav", dtype="float32")
+ assert sr == SR
+ chunks.append(c)
+ marks.append([round(pos / SR, 3), round((pos + c.size) / SR, 3)])
+ pos += c.size
+ if i < len(parts) - 1:
+ chunks.append(np.zeros(int(SENTENCE_PAUSE * SR), dtype=np.float32))
+ pos += chunks[-1].size
+ spans[n] = {"dur": round(pos / SR, 3), "sentences": marks}
+ raw = folder / "raw" / f"{n:02d}.wav"
+ sf.write(raw, np.concatenate(chunks), SR)
+ master(raw, folder / f"{n:02d}.wav")
+ files.append(folder / f"{n:02d}.wav")
+ print(f"{n:02d} {pos / SR:4.1f} s {line}", flush=True)
+ gap = np.zeros(int(PAUSE * 48000), dtype=np.float32)
+ preview = np.concatenate([x for f in files for x in (sf.read(f, dtype="float32")[0], gap)])
+ sf.write(folder / "all.wav", preview, 48000)
+ gain = LUFS - lufs(folder / "all.wav")
+ for f in [*files, folder / "all.wav"]:
+ tmp = f.with_suffix(".tmp.wav")
+ master(f, tmp, gain)
+ tmp.replace(f)
+ (folder / "picks.json").write_text(json.dumps({"from": str(src), "pick": pick}, indent=1) + "\n", encoding="utf-8")
+ dst = VIDEO / "public" / UPDATE["public"]
+ dst.mkdir(parents=True, exist_ok=True)
+ for f in files:
+ shutil.copy(f, dst / f.name)
+ (VIDEO / "src" / "app" / UPDATE["json"]).write_text(json.dumps(spans, indent=1) + "\n", encoding="utf-8")
+ print(f"{folder}: all.wav {preview.size / 48000:.1f} s at {lufs(folder / 'all.wav'):.1f} LUFS; copied to video/public/{UPDATE['public']}, spans in video/src/app/{UPDATE['json']}")
+
+
if __name__ == "__main__":
ap = argparse.ArgumentParser()
ap.add_argument("takes", nargs="*", default=list(TAKES))
@@ -254,6 +332,10 @@ if __name__ == "__main__":
ap.add_argument("--take", help="redo: model settings (default: the line's approved take in FINAL)")
ap.add_argument("--from", dest="src", type=Path)
a = ap.parse_args()
+ if a.takes[:1] == ["audition"]:
+ raise SystemExit(audition(a.seeds, a.seed))
+ if a.takes[:1] == ["build"]:
+ raise SystemExit(build(a.src, {k: int(v) for k, v in (p.split(":") for p in a.pick.split(","))}))
if a.takes[:1] == ["redo"]:
n = int(a.takes[1])
raise SystemExit(redo(n, pairs(a.keep), a.seeds, a.take or FINAL[n][0], a.seed))
diff --git a/marketing/audio/voiceover-0.10.md b/marketing/audio/voiceover-0.10.md
new file mode 100644
index 0000000..9e2e9f9
--- /dev/null
+++ b/marketing/audio/voiceover-0.10.md
@@ -0,0 +1,23 @@
+# Voiceover script: "Kokon 0.10" update video (2026-10-20)
+
+About 40 s, for Reddit (it autoplays muted, so the on-screen text carries every message on its
+own), YouTube and the site. Same voice and delivery as the launch take (Chatterbox base, one
+sentence at a time, 0.4 s between sentences). "Prompt Plus" is spelled out for the voice; on
+screen it is "Prompt+". Honest claims only: Prompt+ uses Qwen3-VL 8B, the text encoder
+Qwen-Image 2.1 already loads (0.10.0 still offers an engine update, so no "nothing to
+download"); LoRA is beta, tested on Windows, native Qwen-Image 2.1 LoRAs only, no catalog.
+
+| # | Shot | Window | Line |
+|---|---|---|---|
+| 1 | Title (the mark, "0.10") | 0:00–0:03 | Kokon zero point ten is here. |
+| 2 | Prompt+ (type, write, hatch) | 0:03–0:13 | Type a few words. Prompt Plus writes the full prompt. Then Kokon makes the image. |
+| 3 | One model, two jobs | 0:13–0:21 | One model, two jobs. The model that reads your prompt now writes it, too. No second model to download. |
+| 4 | LoRA (beta) | 0:21–0:27.5 | Your own LoRAs, in beta. Load one in Settings. Turn it on with one click. |
+| 5 | The new look | 0:27.5–0:33.5 | And a calmer new look, spun from silk. |
+| 6 | End card | 0:33.5–0:40.7 | Free, for Windows, Mac and Linux. Get it at kokon dot app. |
+
+Takes: `gen_voice.py audition` (every sentence, a few seeds each, in voice/update-0.10/audition-<stamp>),
+then `gen_voice.py build --from <that folder> --pick LINE.SENTENCE:SEED,...`, which writes
+voice/update-0.10/final-<stamp>/ and copies the lines to video/public/voice-0.10 with their
+sentence spans in video/src/app/voice-0.10.json. The launch take (voice/final, public/voice,
+voice.json) is never touched.
diff --git a/marketing/video/src/Root.tsx b/marketing/video/src/Root.tsx
index 3eade85..a24b5c4 100644
--- a/marketing/video/src/Root.tsx
+++ b/marketing/video/src/Root.tsx
@@ -9,6 +9,7 @@ import { MAIN_FRAMES, Main } from "./scenes/Main";
import { MONTAGE_FRAMES, Montage } from "./scenes/Montage";
import { Hook, Intro } from "./scenes/Opening";
import { Banner } from "./scenes/Banner";
+import { Update010, UPDATE_FRAMES } from "./scenes/Update010";
import { LENGTH } from "./app/timeline";
const withPreload = (C: ComponentType) => () => (
@@ -23,6 +24,7 @@ export function Root() {
return (
<>
<Composition id="KokonLaunch" component={withPreload(Main)} durationInFrames={MAIN_FRAMES} {...S} />
+ <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} />
<Composition id="Hardware" component={withPreload(Hardware)} durationInFrames={HARDWARE_FRAMES} {...S} />
@@ -32,7 +34,7 @@ export function Root() {
<Composition id="Create" component={withPreload(Create)} durationInFrames={CREATE_FRAMES} {...S} />
<Composition id="Region" component={withPreload(Region)} durationInFrames={REGION_FRAMES} {...S} />
<Composition id="Versions" component={withPreload(Versions)} durationInFrames={VERSIONS_FRAMES} {...S} />
- <Composition id="Layout" component={Layout} durationInFrames={4} {...S} />
+ <Composition id="Layout" component={Layout} durationInFrames={8} {...S} />
<Composition id="Banner" component={withPreload(Banner)} durationInFrames={1} fps={30} width={2560} height={1440} />
</>
);
diff --git a/marketing/video/src/app/AppWindow.tsx b/marketing/video/src/app/AppWindow.tsx
index 4af6802..05ead9e 100644
--- a/marketing/video/src/app/AppWindow.tsx
+++ b/marketing/video/src/app/AppWindow.tsx
@@ -2,8 +2,10 @@
// frame. Kokon draws its own title bar (the top bar holds the window's buttons), so the window
// is one surface, as in the app.
import type { CSSProperties, ReactNode } from "react";
-import { useCurrentFrame } from "remotion";
+import { useEffect, useState } from "react";
+import { continueRender, delayRender, useCurrentFrame } from "remotion";
import { Composer } from "../../../../src/components/Composer";
+import { Settings } from "../../../../src/components/Settings";
import { Sidebar } from "../../../../src/components/Sidebar";
import { Stage } from "../../../../src/components/Stage";
import { TopBar } from "../../../../src/components/TopBar";
@@ -19,8 +21,9 @@ import "./theme.css";
if (typeof document !== "undefined") document.documentElement.dataset.theme = "dark";
/** `split` is the before/after slider position for this frame (0–1); `card` the clock of the
- * job card in the stage (cocoon, Prompt+ text, hatching). */
-export function AppWindow({ state, width, height, overlay, hideComposer, split = 0.5, card }: { state: State; width: number; height: number; overlay?: ReactNode; hideComposer?: boolean; split?: number; card?: CardMotion }) {
+ * job card in the stage (cocoon, Prompt+ text, hatching); `className` "silk" shows the thread
+ * that runs round the composer while you write (app/theme.css). */
+export function AppWindow({ state, width, height, overlay, hideComposer, split = 0.5, card, className }: { state: State; width: number; height: number; overlay?: ReactNode; hideComposer?: boolean; split?: number; card?: CardMotion; className?: string }) {
const f = useCurrentFrame();
setFrameState(state);
setWordPace(card?.cps ?? 60);
@@ -31,7 +34,7 @@ export function AppWindow({ state, width, height, overlay, hideComposer, split =
<FrameStateContext.Provider value={state}>
<CompareSplitContext.Provider value={split}>
<CardMotionContext.Provider value={card ?? null}>
- <div className="kokon-window" style={{ width, height, ...vt }}>
+ <div className={className ? `kokon-window ${className}` : "kokon-window"} style={{ width, height, ...vt }}>
<div className="app">
{state.sidebar && <Sidebar />}
<div className="workspace">
@@ -42,6 +45,12 @@ export function AppWindow({ state, width, height, overlay, hideComposer, split =
</div>
{!hideComposer && <Composer />}
</div>
+ {state.overlay === "settings" && (
+ <>
+ <Settings />
+ <WaitFor selector=".modal" />
+ </>
+ )}
</div>
</div>
</CardMotionContext.Provider>
@@ -49,3 +58,15 @@ export function AppWindow({ state, width, height, overlay, hideComposer, split =
</FrameStateContext.Provider>
);
}
+
+/** Holds the frame until `selector` is on the page: Settings shows once its config has arrived. */
+function WaitFor({ selector }: { selector: string }) {
+ const [handle] = useState(() => delayRender(`Waiting for ${selector}`));
+ useEffect(() => {
+ let raf = 0;
+ const check = () => (document.querySelector(selector) ? continueRender(handle) : (raf = requestAnimationFrame(check)));
+ check();
+ return () => cancelAnimationFrame(raf);
+ }, [handle, selector]);
+ return null;
+}
diff --git a/marketing/video/src/app/fixtures.ts b/marketing/video/src/app/fixtures.ts
index e0eaac1..9bdc72f 100644
--- a/marketing/video/src/app/fixtures.ts
+++ b/marketing/video/src/app/fixtures.ts
@@ -2,7 +2,7 @@
// public/img. Everything a scene shows goes through the app's own components.
import { emptyComposer, type Composer } from "../../../../src/lib/logic";
import { getState, type SessionItem, type State } from "../mocks/store";
-import type { GenerationRecord, GenerationRequest, InpaintAction, JobView, ResolvedMode } from "../../../../src/lib/types";
+import type { GenerationRecord, GenerationRequest, InpaintAction, JobView, LoraItem, ResolvedMode } from "../../../../src/lib/types";
export const SIZES: Record<string, [number, number]> = {
street: [2752, 1536],
@@ -177,3 +177,22 @@ export function appState(patch: Partial<State> = {}): State {
...patch,
};
}
+
+/** The app with Settings open on the LoRA (beta) tab (Settings also reads the service's version). */
+export const settingsState = (patch: Partial<State> = {}): State =>
+ appState({ overlay: "settings", settingsTab: "lora", loras: [LORA], status: { server: { text_generation: true }, service: { version: "0.10.0" } } as unknown as State["status"], ...patch });
+
+/** A LoRA the user has loaded in Settings → LoRA (beta): a stand-in name, no real file. */
+export const LORA: LoraItem = {
+ id: "film-look",
+ name: "Film look",
+ source: String.raw`C:\Users\you\Downloads\film-look.safetensors`,
+ weight: 0.7,
+ triggerWords: "analog film photo",
+ onByDefault: false,
+ size: 84_200_000,
+ warning: null,
+ notes: [],
+ addedAt: "2026-10-20T09:00:00Z",
+ available: true,
+};
diff --git a/marketing/video/src/app/motion.tsx b/marketing/video/src/app/motion.tsx
index 468a41e..d69083f 100644
--- a/marketing/video/src/app/motion.tsx
+++ b/marketing/video/src/app/motion.tsx
@@ -179,13 +179,13 @@ export function Pointer({ path, clicks = [], hideBefore = -1, hideAfter = 1e9 }:
const captionBase: CSSProperties = { position: "absolute", fontFamily: '"Inter Variable", Inter, sans-serif', color: "#f2f2f2", pointerEvents: "none" };
/** Big title + small line, words rising in one after another; fades out at `end`. */
-export function Caption({ title, sub, start, end, x = 96, y = 820, align = "left", size = 64 }: { title: string; sub?: string; start: number; end: number; x?: number; y?: number; align?: "left" | "center"; size?: number }) {
+export function Caption({ title, sub, start, end, x = 96, y = 820, align = "left", size = 64, maxWidth }: { title: string; sub?: string; start: number; end: number; x?: number; y?: number; align?: "left" | "center"; size?: number; maxWidth?: number }) {
const f = useCurrentFrame();
const gone = prog(f, end - 10, 10, inOut);
if (f < start || f > end) return null;
const words = title.split(" ");
return (
- <div style={{ ...captionBase, left: align === "center" ? 0 : x, right: align === "center" ? 0 : undefined, top: y, textAlign: align, opacity: 1 - gone, transform: `translateY(${-gone * 16}px)` }}>
+ <div style={{ ...captionBase, left: align === "center" ? 0 : x, right: align === "center" ? 0 : undefined, top: y, maxWidth, textAlign: align, opacity: 1 - gone, transform: `translateY(${-gone * 16}px)` }}>
<div style={{ fontSize: size, fontWeight: 650, letterSpacing: "-0.035em", lineHeight: 1.05, textShadow: "0 4px 30px rgba(0,0,0,0.55)" }}>
{words.map((w, i) => {
const p = prog(f, start + i * 3, 16);
@@ -197,7 +197,7 @@ export function Caption({ title, sub, start, end, x = 96, y = 820, align = "left
})}
</div>
{sub && (
- <div style={{ marginTop: 14, fontSize: size * 0.36, fontWeight: 450, color: "rgba(240,240,240,0.72)", letterSpacing: "-0.01em", opacity: prog(f, start + words.length * 3 + 4, 16), textShadow: "0 2px 18px rgba(0,0,0,0.6)" }}>
+ <div style={{ marginTop: 14, lineHeight: maxWidth ? 1.35 : undefined, fontSize: size * 0.36, fontWeight: 450, color: "rgba(240,240,240,0.72)", letterSpacing: "-0.01em", opacity: prog(f, start + words.length * 3 + 4, 16), textShadow: "0 2px 18px rgba(0,0,0,0.6)" }}>
{sub}
</div>
)}
diff --git a/marketing/video/src/app/sound.tsx b/marketing/video/src/app/sound.tsx
index 9ef1cd1..8dd71ba 100644
--- a/marketing/video/src/app/sound.tsx
+++ b/marketing/video/src/app/sound.tsx
@@ -3,7 +3,7 @@
// marketing/audio/make_sfx.py), placed on the frames where things happen. Every level is here,
// so the mix can be tuned in one place.
import { Html5Audio, Sequence, staticFile } from "remotion";
-import { CUES, cueFrames, cueSpan, FPS, type Cue } from "./timeline";
+import { CUES, cueFrames, cueSpan, FPS, LAUNCH_VOICE, type Cue } from "./timeline";
const LEVEL = {
music: 0.62,
@@ -55,7 +55,7 @@ function Say({ c }: { c: Cue }) {
return (
<Sequence from={c.at} durationInFrames={n} layout="none" name={`voice ${c.line}.${c.from}`}>
<Html5Audio
- src={staticFile(`voice/${String(c.line).padStart(2, "0")}.wav`)}
+ src={staticFile(`${(c.set ?? LAUNCH_VOICE).dir}/${String(c.line).padStart(2, "0")}.wav`)}
trimBefore={Math.round(a * FPS)}
volume={(f) => LEVEL.voice * Math.min(1, f / 2, (n - f) / 2)}
/>
@@ -63,18 +63,18 @@ function Say({ c }: { c: Cue }) {
);
}
-export const Voice = () => (
+export const Voice = ({ cues = CUES }: { cues?: Cue[] }) => (
<>
- {CUES.map((c) => (
+ {cues.map((c) => (
<Say key={`${c.line}.${c.from}`} c={c} />
))}
</>
);
/** 1 with no voice, DUCK.level while the voice speaks, with smooth ramps around each cue. */
-function duck(f: number): number {
+function duck(f: number, cues: Cue[]): number {
let d = 0;
- for (const c of CUES) {
+ for (const c of cues) {
const a = c.at;
const b = c.at + cueFrames(c);
const v = f < a ? 1 - (a - f) / DUCK.attack : f > b ? 1 - (f - b) / DUCK.release : 1;
@@ -83,7 +83,8 @@ function duck(f: number): number {
return 1 - (1 - DUCK.level) * d;
}
-/** Music for the whole film, under the voice, fading out over the last `fadeOut` frames. */
-export function Music({ frames, fadeOut = 45 }: { frames: number; fadeOut?: number }) {
- return <Html5Audio src={staticFile("music.m4a")} volume={(f) => LEVEL.music * duck(f) * Math.min(1, Math.max(0, (frames - f) / fadeOut))} />;
+/** Music for the whole film, under the voice (`cues`), fading out over the last `fadeOut` frames.
+ * `src` and `from` (seconds into the file) pick the track and where it starts. */
+export function Music({ frames, fadeOut = 45, cues = CUES, src = "music.m4a", from = 0 }: { frames: number; fadeOut?: number; cues?: Cue[]; src?: string; from?: number }) {
+ return <Html5Audio src={staticFile(src)} trimBefore={Math.round(from * FPS)} volume={(f) => LEVEL.music * duck(f, cues) * Math.min(1, Math.max(0, (frames - f) / fadeOut))} />;
}
diff --git a/marketing/video/src/app/theme.css b/marketing/video/src/app/theme.css
index 463004d..2793930 100644
--- a/marketing/video/src/app/theme.css
+++ b/marketing/video/src/app/theme.css
@@ -10,3 +10,5 @@ body { background: #000; }
.kokon-window { position: relative; overflow: hidden; border-radius: 12px; background: var(--ground); color: var(--text); font-family: var(--font);
box-shadow: 0 0 0 1px rgba(255,255,255,0.08), 0 40px 120px rgba(0,0,0,0.6), 0 12px 40px rgba(0,0,0,0.5); display: flex; flex-direction: column; }
.kokon-window .app { flex: 1; min-height: 0; height: auto; }
+/* The silk thread round the composer shows while the prompt has focus, which a video can't give it. */
+.kokon-window.silk .composer::before { opacity: 1; }
diff --git a/marketing/video/src/app/timeline.ts b/marketing/video/src/app/timeline.ts
index 3f6eae7..e953257 100644
--- a/marketing/video/src/app/timeline.ts
+++ b/marketing/video/src/app/timeline.ts
@@ -52,28 +52,36 @@ export const beatFrame = (k: number) => Math.round((BEAT0 + k * BEAT) * FPS);
// Voice-over: public/voice/NN.wav from marketing/audio/gen_voice.py; voice.json holds where each
// sentence starts and ends in its file (seconds).
// ---------------------------------------------------------------------------------------------
-type Lines = Record<string, { dur: number; sentences: [number, number][] }>;
-const LINES = voice as unknown as Lines;
+export type Lines = Record<string, { dur: number; sentences: [number, number][] }>;
+/** A voice-over: the folder of its files in public/ (NN.wav per line) and where each sentence is. */
+export interface VoiceSet {
+ dir: string;
+ lines: Lines;
+}
+export const LAUNCH_VOICE: VoiceSet = { dir: "voice", lines: voice as unknown as Lines };
-/** A piece of one line: sentences `from`..`to` (inclusive), starting at main-timeline frame `at`. */
+/** A piece of one line: sentences `from`..`to` (inclusive), starting at main-timeline frame `at`.
+ * `set`: the voice-over it comes from (the launch film's by default). */
export interface Cue {
line: number;
from: number;
to: number;
at: number;
+ set?: VoiceSet;
}
+const linesOf = (c: Cue) => (c.set ?? LAUNCH_VOICE).lines[String(c.line)].sentences;
/** Seconds into the line's file where the cue's sentences start and end. */
export function cueSpan(c: Cue): [number, number] {
- const s = LINES[String(c.line)].sentences;
+ const s = linesOf(c);
return [s[c.from][0], s[c.to][1]];
}
export const cueFrames = (c: Cue) => Math.ceil((cueSpan(c)[1] - cueSpan(c)[0]) * FPS);
/** Frame (main timeline) where sentence `i` of the cue's line starts, when played from this cue. */
-export const sentenceAt = (c: Cue, i: number) => c.at + Math.round((LINES[String(c.line)].sentences[i][0] - cueSpan(c)[0]) * FPS);
+export const sentenceAt = (c: Cue, i: number) => c.at + Math.round((linesOf(c)[i][0] - cueSpan(c)[0]) * FPS);
const cue = (scene: SceneName, line: number, local: number, from = 0, to = from): Cue => ({ line, from, to, at: START[scene] + local });
-const all = (scene: SceneName, line: number, local: number): Cue => cue(scene, line, local, 0, LINES[String(line)].sentences.length - 1);
+const all = (scene: SceneName, line: number, local: number): Cue => cue(scene, line, local, 0, LAUNCH_VOICE.lines[String(line)].sentences.length - 1);
// Scene-local frames where things happen that the voice names. Scenes import these too.
// Create with Prompt+: a few words typed, Prompt+ on, Send on beat 10 and a cut to the card; it
diff --git a/marketing/video/src/app/timeline010.ts b/marketing/video/src/app/timeline010.ts
new file mode 100644
index 0000000..7410b3c
--- /dev/null
+++ b/marketing/video/src/app/timeline010.ts
@@ -0,0 +1,82 @@
+// The clock of the "Kokon 0.10" update video (composition Update010): the music's beat grid,
+// the shots (cut on beats), the voice-over cues (voiceover-0.10.md) and the frames of the
+// actions the voice names. Frames at 30 fps from the start of the video.
+import voice from "./voice-0.10.json";
+import { FPS, type Cue, type Lines, sentenceAt, type VoiceSet } from "./timeline";
+
+// Music: `src` from `from` seconds; its beats fall every BEAT seconds from BEAT0 (seconds into
+// the video). Placeholder until the new take: the launch track's tail (beat 76 of its grid at
+// 47.09 s), so its drop (60.3 s) lands on beat 26, the hatch.
+export const MUSIC = { src: "music.m4a", from: 8.45 + 76 * (60 / 118) };
+const BEAT = 60 / 118;
+const BEAT0 = 0;
+/** Frame of beat `k` of the video. */
+export const beat = (k: number) => Math.round((BEAT0 + k * BEAT) * FPS);
+
+/** Shot starts (frames), each on a beat. */
+export const SHOT = {
+ title: 0,
+ composer: beat(6),
+ card: beat(10),
+ making: beat(18),
+ logo: beat(24),
+ hatch: beat(26), // the drop
+ model: beat(30),
+ settings: beat(42),
+ /** On the beat before "Turn it on with one click." (set below, from the voice). */
+ pill: 0,
+ look: beat(56),
+ bar: beat(59),
+ silk: beat(62),
+ end: beat(66),
+};
+export const UPDATE_FRAMES = beat(80);
+
+// Voice-over: public/voice-0.10/NN.wav and voice-0.10.json from `gen_voice.py build`. Until the
+// takes exist the json holds estimated spans ("_provisional"): the cues still time the
+// captions and the music's ducking, but no voice plays.
+const json = voice as unknown as Lines & { _provisional?: boolean };
+export const VOICE_READY = !json._provisional;
+export const UPDATE_VOICE: VoiceSet = { dir: "voice-0.10", lines: json };
+
+const cue = (line: number, at: number, from = 0, to = from): Cue => ({ line, from, to, at, set: UPDATE_VOICE });
+const all = (line: number, at: number): Cue => cue(line, at, 0, UPDATE_VOICE.lines[String(line)].sentences.length - 1);
+const span = (line: number, i: number) => UPDATE_VOICE.lines[String(line)].sentences[i];
+
+/** Prompt+ in the composer: typing on "Type a few words.", the prompt written while the voice
+ * says "Prompt Plus writes the full prompt." (its sentence starts at `write`). */
+const WRITE = SHOT.card + 7;
+export const CUES_010: Cue[] = [
+ all(1, 10),
+ all(2, WRITE - Math.round(span(2, 1)[0] * FPS)),
+ all(3, SHOT.hatch + 10),
+ all(4, SHOT.settings + 6),
+ all(5, SHOT.look + 6),
+ all(6, SHOT.end + 12),
+];
+export const LINE = (n: number) => CUES_010[n - 1];
+export const said = (n: number, i: number) => sentenceAt(LINE(n), i);
+/** Frame where sentence `i` of line `n` ends. */
+export const saidEnd = (n: number, i: number) => LINE(n).at + Math.round((span(n, i)[1] - span(n, 0)[0]) * FPS);
+/** The last beat at or before frame `f`. */
+const beatBefore = (f: number) => beat(Math.floor((f / FPS - BEAT0) / BEAT));
+SHOT.pill = Math.min(beatBefore(said(4, 2) + 2), SHOT.look - 45);
+
+/** LoRA: "Load a LoRA…" clicked as the voice says "Load one in Settings.", the pill on "one
+ * click" (but before the cut to the new look). */
+export const LORA_CLICK = { load: said(4, 1) + 6, pill: Math.min(saidEnd(4, 2) - 8, SHOT.look - 14) };
+
+/** The Prompt+ shots (frames of the video), in the shape Create uses (scenes/AppScenes.tsx). */
+export const PROMPT = {
+ ask: SHOT.composer + 14,
+ typeCps: 48,
+ pill: SHOT.card - 14,
+ send: SHOT.card - 1,
+ job: SHOT.card,
+ write: WRITE,
+ written: SHOT.making - 14,
+ leave: SHOT.making - 6,
+ form: SHOT.logo - 1,
+ done: SHOT.hatch - 6,
+ hatch: SHOT.hatch,
+};
diff --git a/marketing/video/src/app/voice-0.10.json b/marketing/video/src/app/voice-0.10.json
new file mode 100644
index 0000000..1ac0378
--- /dev/null
+++ b/marketing/video/src/app/voice-0.10.json
@@ -0,0 +1,85 @@
+{
+ "_provisional": true,
+ "1": {
+ "dur": 2.298,
+ "sentences": [
+ [
+ 0,
+ 2.298
+ ]
+ ]
+ },
+ "2": {
+ "dur": 6.669,
+ "sentences": [
+ [
+ 0,
+ 1.615
+ ],
+ [
+ 2.015,
+ 4.313
+ ],
+ [
+ 4.713,
+ 6.669
+ ]
+ ]
+ },
+ "3": {
+ "dur": 8.034,
+ "sentences": [
+ [
+ 0,
+ 1.615
+ ],
+ [
+ 2.015,
+ 5.678
+ ],
+ [
+ 6.078,
+ 8.034
+ ]
+ ]
+ },
+ "4": {
+ "dur": 6.669,
+ "sentences": [
+ [
+ 0,
+ 1.956
+ ],
+ [
+ 2.356,
+ 3.971
+ ],
+ [
+ 4.371,
+ 6.669
+ ]
+ ]
+ },
+ "5": {
+ "dur": 2.98,
+ "sentences": [
+ [
+ 0,
+ 2.98
+ ]
+ ]
+ },
+ "6": {
+ "dur": 4.996,
+ "sentences": [
+ [
+ 0,
+ 2.298
+ ],
+ [
+ 2.698,
+ 4.996
+ ]
+ ]
+ }
+}
\ No newline at end of file
diff --git a/marketing/video/src/mocks/api.ts b/marketing/video/src/mocks/api.ts
index c581bbb..33fc959 100644
--- a/marketing/video/src/mocks/api.ts
+++ b/marketing/video/src/mocks/api.ts
@@ -1,5 +1,6 @@
// The service API, offline: image URLs point at the video's static files
-// (public/img/<id>.jpg and public/thumb/<id>.jpg); requests never resolve.
+// (public/img/<id>.jpg and public/thumb/<id>.jpg); requests never resolve, except the settings
+// Settings needs before it shows (a stand-in config; only the LoRA tab is filmed).
import { staticFile } from "remotion";
import type { AppErrorBody } from "../../../../src/lib/types";
@@ -12,7 +13,8 @@ export const setPrivateInputs = (_on: boolean) => {};
export const configure = (_b: string, _t: string) => {};
export const isConfigured = () => true;
const never = () => new Promise<never>(() => {});
-export const api: Record<string, (...a: unknown[]) => Promise<never>> = new Proxy({}, { get: () => never });
+const ANSWERS: Record<string, unknown> = { config: { outputDir: "C:/Users/you/Pictures/Kokon", upscalerModel: "superscale-sp", gpuBackend: "cuda", activeProfile: "official", profiles: [{ id: "official" }], ui: {} } };
+export const api: Record<string, (...a: unknown[]) => Promise<never>> = new Proxy({}, { get: (_t, name: string) => (name in ANSWERS ? () => Promise.resolve(ANSWERS[name]) : never) });
export const urls = {
input: (id: string) => staticFile(`img/${id}.jpg`),
inputThumb: (id: string) => staticFile(`thumb/${id}.jpg`),
diff --git a/marketing/video/src/scenes/AppScenes.tsx b/marketing/video/src/scenes/AppScenes.tsx
index e340407..228760f 100644
--- a/marketing/video/src/scenes/AppScenes.tsx
+++ b/marketing/video/src/scenes/AppScenes.tsx
@@ -25,19 +25,19 @@ const inpaintReq = (prompt: string, from: string, action: "remove" | "add") => (
inpaint_action: action,
});
// Made with Prompt+ from a few words: the record keeps the written prompt and the user's own.
-const STREET = gen("street", "generate", PROMPTS.enhanced, undefined, { original_prompt: PROMPTS.idea, request: { prompt: PROMPTS.idea, mode: "generate", images: [], quality: "quality", resolution: "auto", advanced: {}, enhance: true } });
-const REMOVED = gen("street-removed", "inpaint", PROMPTS.remove, "street", { request: inpaintReq(PROMPTS.remove, "street", "remove") });
-const CAT = gen("street-cat", "inpaint", PROMPTS.add, "street-removed", { request: inpaintReq(PROMPTS.add, "street-removed", "add") });
+export const STREET = gen("street", "generate", PROMPTS.enhanced, undefined, { original_prompt: PROMPTS.idea, request: { prompt: PROMPTS.idea, mode: "generate", images: [], quality: "quality", resolution: "auto", advanced: {}, enhance: true } });
+export const REMOVED = gen("street-removed", "inpaint", PROMPTS.remove, "street", { request: inpaintReq(PROMPTS.remove, "street", "remove") });
+export const CAT = gen("street-cat", "inpaint", PROMPTS.add, "street-removed", { request: inpaintReq(PROMPTS.add, "street-removed", "add") });
const STREET_PROJECT = { id: "current", title: "Italian street café", updatedAt: Date.now() - 20_000, cover: "gen:street", count: 1 };
const withProject = (count: number, cover: string): State["projects"] => [{ ...STREET_PROJECT, count, cover }, ...PROJECTS];
const proj = (n: number, cover: string) => ({ projects: withProject(n, cover), projectId: "current" });
-const SEND: [number, number] = [1232, 857];
+export const SEND: [number, number] = [1232, 857];
/** The composer, close: a macro shot past the window's bottom edge, sharp on the composer only. */
const COMPOSER_CAM: Cam = { x: 850, y: 630, s: 1.8 };
const COMPOSER_FOCUS: [number, number] = [690, 1010];
/** The empty project's composer is shorter and lower: centre it. */
-const PROMPT_CAM: Cam = { x: 850, y: 615, s: 1.7 };
+export const PROMPT_CAM: Cam = { x: 850, y: 615, s: 1.7 };
const PROMPT_FOCUS: [number, number] = [800, 1050];
/** Where the stage image sits once a result is selected (window px), to fly into it. */
const STAGE_IMG = { x: 850, y: 319, w: 820 };
@@ -46,7 +46,7 @@ const STAGE_IMG = { x: 850, y: 319, w: 820 };
const stepAt = (f: number, start: number, every: number) => Math.max(0, Math.min(8, Math.floor((f - start) / every) + 1));
/** Big centred words that rise in with the voice and leave with the next ones. */
-function Words({ items, end }: { items: { text: string; at: number }[]; end: number }) {
+export function Words({ items, end }: { items: { text: string; at: number }[]; end: number }) {
const f = useCurrentFrame();
return (
<>
@@ -80,50 +80,64 @@ export const CREATE_FRAMES = LENGTH.create;
/** The card in two still shots: the whole card while Prompt+ writes (a slow push), then, from
* the percentage on, closer, with the logo where the full-frame logo the image hatches from
* sits (same centre and size, measured), so the cut between them reads as one logo. */
-const CARD_CAM: Cam = { x: 852, y: 360, s: 1.55 };
+export const CARD_CAM: Cam = { x: 852, y: 360, s: 1.55 };
const CARD_CAM_END = 1.6;
-const LOGO_CAM: Cam = { x: 848, y: 364, s: 1.756 };
-const PILL: [number, number] = [840, 853];
+export const LOGO_CAM: Cam = { x: 848, y: 364, s: 1.756 };
+export const PILL: [number, number] = [840, 853];
+
+/** When things happen in a Prompt+ shot (scene frames): typing, the Prompt+ pill, Send, the
+ * prompt being written and leaving, the logo forming, the hatch. Create uses CREATE; the 0.10
+ * update video its own. */
+export type PromptPlusTimes = Pick<typeof CREATE, "ask" | "pill" | "send" | "job" | "write" | "written" | "leave" | "form" | "done" | "hatch"> & { typeCps?: number };
/** The job's percentage at scene frame `f`: the image starts while Prompt+ is still on screen. */
-const createPct = (f: number) => kf(f, [[CREATE.write, 0], [CREATE.form, 92], [CREATE.done, 100]]);
+const pctFor = (C: PromptPlusTimes) => (f: number) => kf(f, [[C.write, 0], [C.form, 92], [C.done, 100]]);
-export function Create() {
- const f = useCurrentFrame();
- const base = appState();
- const cues = sceneCues("create");
- const C = CREATE;
- const T0 = C.ask + 2;
- const T1 = typingDone(PROMPTS.idea, T0, 32);
- // Prompt+ writes the whole prompt in about two seconds on screen (sped up, like the image).
+/** The job card's clock at scene frame `f`. */
+export function promptPlusCard(f: number, C: PromptPlusTimes): CardMotion {
+ const pct = pctFor(C);
+ // Prompt+ writes the whole prompt between `write` and `written` (sped up, like the image).
const cps = (PROMPTS.enhanced.length * FPS) / (C.written - C.write);
- const card: CardMotion = {
+ return {
t: Math.max(0, f - C.job) / FPS,
- pctAt: (t) => createPct(C.job + t * FPS),
+ pctAt: (t) => pct(C.job + t * FPS),
chars: Math.max(0, f - C.write) * (cps / FPS),
cps,
leave: prog(f, C.leave, 21, (x) => x),
hatchMs: f >= C.hatch ? ((f - C.hatch) * 1000) / FPS : null,
};
+}
+
+/** The window at scene frame `t`: the words typed, the job with Prompt+ writing, then the image. */
+export function promptPlusView(t: number, C: PromptPlusTimes, card: CardMotion, size: { width: number; height: number } = { width: 1440, height: 900 }) {
+ const base = appState();
+ const cpsType = C.typeCps ?? 32;
const request = { prompt: PROMPTS.idea, mode: "generate" as const, images: [], quality: "quality" as const, resolution: "auto" as const, advanced: {}, enhance: true };
+ let state: State;
+ const composer = { ...base.composer, enhance: t >= C.pill };
+ if (t < C.job) {
+ state = appState({ composer: { ...composer, prompt: typed(PROMPTS.idea, t, C.ask + 2, cpsType) } });
+ } else if (t < C.hatch) {
+ const pct = pctFor(C)(t);
+ const j = jobItem("j-street", "generate", PROMPTS.idea, "street", undefined, Math.min(8, Math.floor((pct / 100) * 8)));
+ const writing = t < C.write + 40;
+ // The prompt is written while the image is being made; it stays on screen until it leaves.
+ const job = { ...j.job, status: writing ? ("enhancing_prompt" as const) : j.job.status, step: writing ? null : j.job.step, enhanced_prompt: t < C.leave + 21 ? PROMPTS.enhanced : null };
+ state = appState({ session: [{ ...j.item, request }], selected: "j-street", jobs: { "j-street": job }, composer, ...proj(1, "gen:street") });
+ } else {
+ state = appState({ session: [genItem(STREET)], selected: "street", composer, ...proj(1, "gen:street") });
+ }
+ return <AppWindow state={state} {...size} card={card} />;
+}
- const view = (t: number) => {
- let state: State;
- const composer = { ...base.composer, enhance: t >= C.pill };
- if (t < C.job) {
- state = appState({ composer: { ...composer, prompt: typed(PROMPTS.idea, t, T0, 32) } });
- } else if (t < C.hatch) {
- const pct = createPct(t);
- const j = jobItem("j-street", "generate", PROMPTS.idea, "street", undefined, Math.min(8, Math.floor((pct / 100) * 8)));
- const writing = t < C.write + 40;
- // The prompt is written while the image is being made; it stays on screen until it leaves.
- const job = { ...j.job, status: writing ? ("enhancing_prompt" as const) : j.job.status, step: writing ? null : j.job.step, enhanced_prompt: t < C.leave + 21 ? PROMPTS.enhanced : null };
- state = appState({ session: [{ ...j.item, request }], selected: "j-street", jobs: { "j-street": job }, composer, ...proj(1, "gen:street") });
- } else {
- state = appState({ session: [genItem(STREET)], selected: "street", composer, ...proj(1, "gen:street") });
- }
- return <AppWindow state={state} width={1440} height={900} card={card} />;
- };
+export function Create() {
+ const f = useCurrentFrame();
+ const cues = sceneCues("create");
+ const C = CREATE;
+ const T0 = C.ask + 2;
+ const T1 = typingDone(PROMPTS.idea, T0, 32);
+ const card = promptPlusCard(f, C);
+ const view = (t: number) => promptPlusView(t, C, card);
const enter = prog(f, 0, 34);
const typingCam = camAt(f, [
diff --git a/marketing/video/src/scenes/Layout.tsx b/marketing/video/src/scenes/Layout.tsx
index 44361c0..7bd9858 100644
--- a/marketing/video/src/scenes/Layout.tsx
+++ b/marketing/video/src/scenes/Layout.tsx
@@ -1,7 +1,7 @@
// Key app states at the video's window size, rendered at 1:1 to measure click targets.
import { AbsoluteFill, useCurrentFrame } from "remotion";
import { AppWindow } from "../app/AppWindow";
-import { appState, gen, genItem, jobItem, PROMPTS, ref } from "../app/fixtures";
+import { appState, gen, genItem, jobItem, LORA, PROMPTS, ref, settingsState } from "../app/fixtures";
import { Preload } from "../app/Preload";
export function Layout() {
@@ -17,6 +17,10 @@ export function Layout() {
const j = jobItem("j1", "inpaint", PROMPTS.remove, "street-removed", "street", 3, "remove");
return appState({ session: [genItem(a), j.item], selected: "j1", jobs: { j1: j.job } });
})(),
+ // 5-7: the 0.10 update video: the LoRA pill (off, then on), Settings → LoRA (beta).
+ appState({ loras: [LORA], session: [genItem(a)], selected: "street", composer: { ...base.composer, prompt: PROMPTS.idea } }),
+ appState({ loras: [LORA], session: [genItem(a)], selected: "street", composer: { ...base.composer, prompt: PROMPTS.idea, loras: [LORA.id] } }),
+ settingsState({ session: [genItem(a)], selected: "street" }),
];
return (
<AbsoluteFill style={{ background: "#000" }}>
diff --git a/marketing/video/src/scenes/Update010.tsx b/marketing/video/src/scenes/Update010.tsx
new file mode 100644
index 0000000..7535168
--- /dev/null
+++ b/marketing/video/src/scenes/Update010.tsx
@@ -0,0 +1,429 @@
+// "Kokon 0.10": a 40-second update video from the launch film's pieces (Reddit, YouTube, the
+// site). Prompt+ first (a few words, the prompt written, the image hatching on the drop), then
+// why nothing new is needed (one model, two jobs), the user's own LoRAs (beta), the new look,
+// and the end card. Shots hold still or push in slowly and change by cutting on the beat
+// (app/timeline010.ts). The on-screen words carry every message, since Reddit plays muted.
+import type { CSSProperties, ReactNode } from "react";
+import { AbsoluteFill, interpolate, spring, useCurrentFrame, useVideoConfig } from "remotion";
+import { siGithub } from "simple-icons";
+import { AppWindow } from "../app/AppWindow";
+import { appState, genItem, jobItem, LORA, PROJECTS, PROMPTS, settingsState } from "../app/fixtures";
+import { FullImage } from "../app/FullImage";
+import { Mark, MARK_PATHS } from "../app/Mark";
+import { Backdrop, type Cam, Camera, Caption, DepthOfField, Dissolve, inOut, prog, Pointer, typingDone, Vignette } from "../app/motion";
+import { Music, Send, Sfx, Voice } from "../app/sound";
+import { FPS } from "../app/timeline";
+import { beat, CUES_010, LORA_CLICK, MUSIC, PROMPT, said, SHOT, UPDATE_FRAMES, VOICE_READY } from "../app/timeline010";
+import { CardMotionContext, type CardMotion } from "../mocks/card";
+import { Hatch } from "../mocks/Cocoon";
+import type { State } from "../mocks/store";
+import { CARD_CAM, CAT, LOGO_CAM, PILL, PROMPT_CAM, promptPlusCard, promptPlusView, REMOVED, SEND, STREET, Words } from "./AppScenes";
+
+export { UPDATE_FRAMES };
+const font = '"Inter Variable", Inter, sans-serif';
+const PAPER = "#F7F6F2";
+
+/** Shows `children` from frame `from` (inclusive) to `to` (exclusive): one shot. */
+const Shot = ({ from, to, children }: { from: number; to: number; children: ReactNode }) => {
+ const f = useCurrentFrame();
+ return f >= from && f < to ? <>{children}</> : null;
+};
+
+/** A slow push from scale `a` to `b` over the shot. */
+const push = (f: number, from: number, to: number, a: number, b: number) => a + (b - a) * prog(f, from, to - from, (x) => x);
+
+// ---------------------------------------------------------------------------------------------
+// Title: the mark assembles, "Kokon", and the release.
+// ---------------------------------------------------------------------------------------------
+function Title() {
+ const f = useCurrentFrame();
+ const { fps } = useVideoConfig();
+ const sp = (delay: number, damping = 14) => spring({ frame: f - delay, fps, config: { damping, mass: 0.8, stiffness: 110 } });
+ const stem = sp(2, 18);
+ const upper = sp(10);
+ const lower = sp(16);
+ const shift = prog(f, 26, 22, inOut);
+ const word = prog(f, 30, 22);
+ const badge = spring({ frame: f - beat(3), fps, config: { damping: 12, mass: 0.6, stiffness: 160 } });
+ const tag = prog(f, beat(4), 18);
+ const size = 150;
+ return (
+ <AbsoluteFill style={{ background: "#050505", alignItems: "center", justifyContent: "center", fontFamily: font }}>
+ <div style={{ position: "absolute", inset: 0, background: `radial-gradient(circle at 50% 46%, rgba(247,246,242,${0.07 * stem}) 0%, transparent 45%)` }} />
+ <div style={{ display: "flex", flexDirection: "column", alignItems: "center", gap: 38, transform: `scale(${1 + prog(f, 0, SHOT.composer, (x) => x) * 0.04})` }}>
+ <div style={{ display: "flex", alignItems: "center", gap: interpolate(shift, [0, 1], [0, 40]) }}>
+ <svg width={size} height={size} viewBox="0 0 1024 1024" style={{ overflow: "visible" }}>
+ <path d={MARK_PATHS.stem} fill={PAPER} style={{ transformOrigin: "175px 1000px", transform: `translateY(${(1 - stem) * 160}px) scaleY(${0.5 + stem * 0.5})`, opacity: stem }} />
+ <path d={MARK_PATHS.upper} fill={PAPER} style={{ transformOrigin: "390px 549px", transform: `rotate(${(1 - upper) * -70}deg) scale(${upper})`, opacity: Math.min(1, upper * 2) }} />
+ <path d={MARK_PATHS.lower} fill={PAPER} style={{ transformOrigin: "390px 587px", transform: `rotate(${(1 - lower) * 60}deg) scale(${lower})`, opacity: Math.min(1, lower * 2) }} />
+ </svg>
+ <div style={{ overflow: "hidden", width: interpolate(shift, [0, 1], [0, 720]), height: size + 20, display: "flex", alignItems: "center", gap: 30 }}>
+ <span style={{ fontSize: 156, fontWeight: 620, color: PAPER, letterSpacing: "-0.045em", lineHeight: 1, transform: `translateX(${(1 - word) * -60}px)`, opacity: word, whiteSpace: "nowrap" }}>Kokon</span>
+ <span
+ style={{
+ padding: "10px 26px 14px",
+ borderRadius: 999,
+ border: `3px solid ${PAPER}`,
+ color: PAPER,
+ fontSize: 74,
+ fontWeight: 600,
+ letterSpacing: "-0.03em",
+ lineHeight: 1,
+ opacity: Math.min(1, badge * 1.5),
+ transform: `scale(${0.6 + 0.4 * badge})`,
+ whiteSpace: "nowrap",
+ }}
+ >
+ 0.10
+ </span>
+ </div>
+ </div>
+ <div style={{ fontSize: 40, fontWeight: 450, color: "rgba(247,246,242,0.78)", letterSpacing: "-0.02em", opacity: tag, transform: `translateY(${(1 - tag) * 18}px)`, filter: `blur(${(1 - tag) * 6}px)` }}>
+ Prompt+ · your own LoRAs · a new look
+ </div>
+ </div>
+ </AbsoluteFill>
+ );
+}
+
+// ---------------------------------------------------------------------------------------------
+// Prompt+: a few words typed, Prompt+ on, Send. The card: the prompt is written, the percentage,
+// the logo forms; on the drop the image hatches full frame. Then "One model. Two jobs."
+// ---------------------------------------------------------------------------------------------
+/** The card while the image is made: the whole card, pulled back from the writing shot. */
+const MAKING_CAM: Cam = { x: 852, y: 430, s: 1.4 };
+const PROMPT_FOCUS: [number, number] = [800, 1050];
+
+function PromptPlus() {
+ const f = useCurrentFrame();
+ const P = PROMPT;
+ const card = promptPlusCard(f, P);
+ const onCard = f >= SHOT.card;
+ const cam: Cam = !onCard
+ ? { ...PROMPT_CAM, s: push(f, SHOT.composer, SHOT.card, 1.7, 1.84) }
+ : f < SHOT.making
+ ? { ...CARD_CAM, s: push(f, SHOT.card, SHOT.making, 1.5, 1.6) }
+ : f < SHOT.logo
+ ? { ...MAKING_CAM, s: push(f, SHOT.making, SHOT.logo, 1.4, 1.46) }
+ : LOGO_CAM;
+ const app = (
+ <Camera cam={cam} free={!onCard}>
+ {promptPlusView(f, P, card)}
+ {!onCard && (
+ <Pointer
+ path={[
+ [SHOT.composer + 30, 980, 960],
+ [P.pill - 2, PILL[0], PILL[1]],
+ [P.send - 10, 1120, 880],
+ [P.send - 1, SEND[0], SEND[1]],
+ ]}
+ clicks={[P.pill, P.send]}
+ hideBefore={SHOT.composer + 30}
+ />
+ )}
+ </Camera>
+ );
+ return (
+ <AbsoluteFill>
+ <Backdrop />
+ {!onCard ? <DepthOfField focus={PROMPT_FOCUS} blur={7}>{app}</DepthOfField> : app}
+ <Vignette strength={onCard ? 0.75 : 0} />
+ <Caption title="Type a few words." start={SHOT.composer + 6} end={SHOT.card - 2} y={150} size={52} />
+ <Caption title="Prompt+ writes the full prompt." sub="Off by default · one click in the composer · any language" start={P.write} end={SHOT.making - 2} y={890} size={52} />
+ <Caption title="Then Kokon makes the image." sub="Qwen-Image 2.1 Turbo on your own GPU (sped up here)" start={SHOT.making + 4} end={SHOT.logo} y={890} size={52} />
+ </AbsoluteFill>
+ );
+}
+
+/** The image hatching full frame from the card's logo, held under "One model. Two jobs." */
+function Hatched() {
+ const f = useCurrentFrame();
+ const card = promptPlusCard(f, PROMPT);
+ return (
+ <AbsoluteFill>
+ <div style={{ position: "absolute", inset: 0, isolation: "isolate" }}>
+ <CardMotionContext.Provider value={card}>
+ <Hatch aspect={1920 / 1080} fill>
+ <FullImage before="street" push={push(f, SHOT.hatch, SHOT.model, 1, 1.04)} dim={0.42 * prog(f, said(3, 0) - 8, 12)} />
+ </Hatch>
+ </CardMotionContext.Provider>
+ </div>
+ <Words
+ items={[
+ { text: "One model.", at: said(3, 0) },
+ { text: "Two jobs.", at: said(3, 0) + 22 },
+ ]}
+ end={SHOT.model + 6}
+ />
+ </AbsoluteFill>
+ );
+}
+
+// ---------------------------------------------------------------------------------------------
+// One model, two jobs: Qwen3-VL 8B is already loaded as Qwen-Image 2.1's text encoder; Prompt+
+// asks it to write. Silk threads run from your words into it and out to its two jobs.
+// ---------------------------------------------------------------------------------------------
+/** A thread of silk from (x0, y0) to (x1, y1): a few strands, drawn in over `p` (0–1). */
+function Thread({ x0, y0, x1, y1, p, glow = 0 }: { x0: number; y0: number; x1: number; y1: number; p: number; glow?: number }) {
+ const strands = [0, 1, 2, 3];
+ const mx = (x0 + x1) / 2;
+ return (
+ <g style={{ filter: glow > 0 ? `drop-shadow(0 0 ${10 * glow}px rgba(247,246,242,${0.8 * glow}))` : undefined }}>
+ {strands.map((i) => {
+ const w = (i - 1.5) * 7;
+ const d = `M ${x0} ${y0} C ${mx} ${y0 + w}, ${mx} ${y1 - w}, ${x1} ${y1}`;
+ return <path key={i} d={d} pathLength={1} fill="none" stroke={PAPER} strokeOpacity={i === 1 ? 0.95 : 0.35} strokeWidth={i === 1 ? 2.4 : 1.2} strokeDasharray="1 1" strokeDashoffset={1 - p} strokeLinecap="round" />;
+ })}
+ </g>
+ );
+}
+
+const glass: CSSProperties = {
+ position: "absolute",
+ borderRadius: 26,
+ background: "linear-gradient(180deg, rgba(255,255,255,0.08), rgba(255,255,255,0.025))",
+ boxShadow: "inset 0 0 0 1px rgba(255,255,255,0.1), 0 40px 100px rgba(0,0,0,0.5)",
+ color: PAPER,
+ fontFamily: font,
+};
+
+function OneModel() {
+ const f = useCurrentFrame();
+ const t0 = SHOT.model;
+ const show = (at: number, dur = 18) => prog(f, at, dur);
+ const rise = (p: number): CSSProperties => ({ opacity: p, transform: `translateY(${(1 - p) * 24}px)`, filter: `blur(${(1 - p) * 8}px)` });
+ const reads = said(3, 1);
+ const writes = reads + 30;
+ const none = said(3, 2);
+ const a = show(t0 + 2);
+ const b = show(t0 + 8);
+ const thIn = prog(f, t0 + 6, 20, inOut);
+ const thRead = prog(f, reads, 18, inOut);
+ const thWrite = prog(f, writes, 18, inOut);
+ const newGlow = Math.min(prog(f, writes + 12, 10), 1 - 0.6 * prog(f, writes + 30, 30));
+ const s = push(f, t0, SHOT.settings, 1, 1.035);
+ return (
+ <AbsoluteFill style={{ fontFamily: font }}>
+ <Backdrop glow={0.9} />
+ <div style={{ position: "absolute", inset: 0, transform: `scale(${s})` }}>
+ <div style={{ position: "absolute", top: 96, left: 0, right: 0, textAlign: "center", fontSize: 76, fontWeight: 660, letterSpacing: "-0.045em", color: "#f4f4f2", ...rise(show(t0, 14)) }}>One model. Two jobs.</div>
+ <svg width={1920} height={1080} style={{ position: "absolute", inset: 0, overflow: "visible" }}>
+ <Thread x0={460} y0={540} x1={640} y1={540} p={thIn} />
+ <Thread x0={1120} y0={500} x1={1250} y1={400} p={thRead} />
+ <Thread x0={1120} y0={580} x1={1250} y1={690} p={thWrite} glow={newGlow} />
+ </svg>
+ {/* your words */}
+ <div style={{ ...glass, left: 90, top: 480, width: 370, padding: "22px 26px", ...rise(a) }}>
+ <div style={{ fontSize: 18, color: "rgba(247,246,242,0.55)", marginBottom: 8 }}>Your words</div>
+ <div style={{ fontSize: 26, fontWeight: 560, letterSpacing: "-0.02em" }}>sunny italian street, café, red scooter</div>
+ </div>
+ {/* the model */}
+ <div style={{ ...glass, left: 640, top: 400, width: 480, height: 280, padding: "34px 36px", display: "flex", flexDirection: "column", justifyContent: "center", ...rise(b) }}>
+ <div style={{ display: "flex", alignItems: "center", gap: 18 }}>
+ <Mark size={44} color={PAPER} />
+ <span style={{ fontSize: 46, fontWeight: 660, letterSpacing: "-0.035em" }}>Qwen3-VL 8B</span>
+ </div>
+ <div style={{ marginTop: 16, fontSize: 23, lineHeight: 1.35, color: "rgba(247,246,242,0.7)" }}>The text encoder Qwen-Image 2.1 already loads. Same engine, same memory.</div>
+ </div>
+ {/* its two jobs */}
+ <div style={{ ...glass, left: 1250, top: 340, width: 580, padding: "20px 26px", ...rise(show(reads + 6)) }}>
+ <div style={{ fontSize: 30, fontWeight: 620, letterSpacing: "-0.02em" }}>Reads your prompt</div>
+ <div style={{ fontSize: 20, color: "rgba(247,246,242,0.6)", marginTop: 6 }}>so the image follows it, as always</div>
+ </div>
+ <div style={{ ...glass, left: 1250, top: 630, width: 580, padding: "20px 26px", boxShadow: `inset 0 0 0 ${1 + newGlow}px rgba(247,246,242,${0.12 + 0.6 * newGlow}), 0 40px 100px rgba(0,0,0,0.5), 0 0 ${60 * newGlow}px rgba(247,246,242,${0.25 * newGlow})`, ...rise(show(writes + 6)) }}>
+ <div style={{ display: "flex", alignItems: "center", gap: 12 }}>
+ <span style={{ fontSize: 30, fontWeight: 620, letterSpacing: "-0.02em" }}>Writes your prompt</span>
+ <span style={{ fontSize: 17, fontWeight: 650, padding: "4px 11px", borderRadius: 999, background: PAPER, color: "#0d0d0d" }}>NEW · Prompt+</span>
+ </div>
+ <div style={{ fontSize: 20, color: "rgba(247,246,242,0.6)", marginTop: 6 }}>a full prompt from your few words</div>
+ </div>
+ <div style={{ position: "absolute", top: 880, left: 0, right: 0, textAlign: "center", fontSize: 50, fontWeight: 640, letterSpacing: "-0.03em", color: "#f4f4f2", ...rise(show(none - 2, 16)) }}>
+ No second model to download.
+ </div>
+ </div>
+ </AbsoluteFill>
+ );
+}
+
+// ---------------------------------------------------------------------------------------------
+// LoRA (beta): load one in Settings, then one click on the LoRA pill turns it on.
+// ---------------------------------------------------------------------------------------------
+/** "Load a LoRA…" in Settings, and the LoRA pill in a project's composer (window px, measured
+ * with the Layout composition at 1:1). */
+const LOAD_BTN: [number, number] = [988, 271];
+const LORA_PILL: [number, number] = [932, 853];
+const SETTINGS_CAM = { x: 640, y: 450 };
+const PILL_CAM: Cam = { x: 880, y: 770, s: 1.9 };
+const PILL_FOCUS: [number, number] = [560, 860];
+const clickLoad = LORA_CLICK.load;
+const loaded = clickLoad + 8;
+const clickPill = LORA_CLICK.pill;
+
+function Lora() {
+ const f = useCurrentFrame();
+ const project = { session: [genItem(STREET)], selected: "street", projects: [{ id: "current", title: "Italian street café", updatedAt: Date.now() - 20_000, cover: "gen:street", count: 1 }, ...PROJECTS], projectId: "current" };
+ const settings = (t: number) => {
+ const state = settingsState({ ...project, loras: t >= loaded ? [LORA] : [] });
+ return <AppWindow state={state} width={1440} height={900} />;
+ };
+ const inSettings = f < SHOT.pill;
+ const base = appState();
+ const pillState: State = appState({ ...project, loras: [LORA], composer: { ...base.composer, prompt: PROMPTS.idea, loras: f >= clickPill ? [LORA.id] : null } });
+ return (
+ <AbsoluteFill>
+ <Backdrop />
+ {inSettings ? (
+ <Camera cam={{ ...SETTINGS_CAM, s: push(f, SHOT.settings, SHOT.pill, 1.48, 1.54) }}>
+ <Dissolve frame={f} cuts={[loaded]} view={settings} dur={8} />
+ <Pointer path={[[SHOT.settings + 4, 1180, 560], [clickLoad - 2, LOAD_BTN[0], LOAD_BTN[1]], [clickLoad + 20, LOAD_BTN[0] + 40, LOAD_BTN[1] + 140]]} clicks={[clickLoad]} hideBefore={SHOT.settings + 4} hideAfter={clickLoad + 22} />
+ </Camera>
+ ) : (
+ <DepthOfField focus={PILL_FOCUS} blur={6}>
+ <Camera cam={{ ...PILL_CAM, s: push(f, SHOT.pill, SHOT.look, 1.9, 2.02) }} free>
+ <AppWindow state={pillState} width={1440} height={900} />
+ <Pointer path={[[SHOT.pill + 4, 1080, 960], [clickPill - 2, LORA_PILL[0], LORA_PILL[1]], [SHOT.look, LORA_PILL[0] + 14, LORA_PILL[1] + 30]]} clicks={[clickPill]} hideBefore={SHOT.pill + 4} />
+ </Camera>
+ </DepthOfField>
+ )}
+ <Vignette side="left" strength={inSettings ? 0.75 : 0} reach={0.7} />
+ {!inSettings && <Vignette side="top" strength={0.6} reach={0.6} />}
+ <Caption title="Your own LoRAs." sub="Beta · load a Qwen-Image 2.1 LoRA once, in Settings" start={SHOT.settings + 4} end={SHOT.pill} x={60} y={430} size={50} maxWidth={420} />
+ <Caption title="One click turns it on." sub="Per project · tested on Windows · no catalog yet" start={SHOT.pill + 4} end={SHOT.look} x={96} y={130} size={52} />
+ </AbsoluteFill>
+ );
+}
+
+// ---------------------------------------------------------------------------------------------
+// The new look: the window whole, its own title bar, the silk where things move.
+// ---------------------------------------------------------------------------------------------
+const LOOK_PROJECT = { projects: [{ id: "current", title: "Italian street café", updatedAt: Date.now() - 20_000, cover: "gen:street-cat", count: 3 }, ...PROJECTS], projectId: "current" };
+
+function NewLook() {
+ const f = useCurrentFrame();
+ const base = appState();
+ const idle = appState({ session: [genItem(STREET), genItem(REMOVED), genItem(CAT)], selected: "street-cat", ...LOOK_PROJECT });
+ // A fourth version being made: the status capsule spins its silk, a thread runs round it.
+ const j = jobItem("j-look", "edit", "Make it golden hour", "street-cat", "street-cat", 3);
+ const making = appState({ session: [genItem(STREET), genItem(REMOVED), genItem(CAT), j.item], selected: "street-cat", jobs: { "j-look": j.job }, ...LOOK_PROJECT, composer: { ...base.composer } });
+ const card: CardMotion = { t: (f - SHOT.bar) / FPS + 4, pctAt: (t) => Math.min(96, 20 + t * 7) };
+ const words = prog(f, SHOT.look + 2, 16);
+ let shot: ReactNode;
+ if (f < SHOT.bar) {
+ shot = (
+ <Camera cam={{ x: 720, y: 400, s: push(f, SHOT.look, SHOT.bar, 0.9, 0.93) }}>
+ <AppWindow state={idle} width={1440} height={900} className="silk" />
+ </Camera>
+ );
+ } else if (f < SHOT.silk) {
+ shot = (
+ <Camera cam={{ x: 1010, y: 150, s: push(f, SHOT.bar, SHOT.silk, 2.0, 2.08) }} free>
+ <AppWindow state={making} width={1440} height={900} card={card} />
+ </Camera>
+ );
+ } else {
+ shot = (
+ <Camera cam={{ x: 848, y: 600, s: push(f, SHOT.silk, SHOT.end, 2.3, 2.4) }} free>
+ <AppWindow state={making} width={1440} height={900} card={card} />
+ </Camera>
+ );
+ }
+ return (
+ <AbsoluteFill>
+ <Backdrop />
+ {shot}
+ {f < SHOT.bar && (
+ <div style={{ position: "absolute", top: 40, left: 0, right: 0, textAlign: "center", fontFamily: font, fontSize: 76, fontWeight: 660, letterSpacing: "-0.045em", color: "#f4f4f2", opacity: words, transform: `translateY(${(1 - words) * 24}px)`, filter: `blur(${(1 - words) * 8}px)` }}>
+ A new look.
+ </div>
+ )}
+ {f >= SHOT.bar && <Vignette strength={0.8} />}
+ <Caption title="Its own title bar." sub="Calmer, deeper, in light and dark" start={SHOT.bar + 2} end={SHOT.silk} y={900} size={52} />
+ <Caption title="Spun from silk." sub="Threads show what's being made" start={SHOT.silk + 2} end={SHOT.end} y={900} size={52} />
+ </AbsoluteFill>
+ );
+}
+
+// ---------------------------------------------------------------------------------------------
+// End card.
+// ---------------------------------------------------------------------------------------------
+function End() {
+ const f = useCurrentFrame();
+ const t = SHOT.end;
+ const a = prog(f, t, 28);
+ const b = prog(f, t + 16, 22);
+ const c = prog(f, said(6, 1) - 4, 20);
+ const fade = prog(f, UPDATE_FRAMES - 20, 20, inOut);
+ return (
+ <AbsoluteFill style={{ fontFamily: font, background: "#050505", alignItems: "center", justifyContent: "center", opacity: 1 - fade }}>
+ <div style={{ position: "absolute", inset: 0, background: "radial-gradient(circle at 50% 42%, rgba(247,246,242,0.06) 0%, transparent 45%)" }} />
+ <div style={{ display: "flex", flexDirection: "column", alignItems: "center", gap: 30, marginTop: -40 }}>
+ <div style={{ display: "flex", alignItems: "center", gap: 34, opacity: a, transform: `scale(${0.94 + a * 0.06})`, filter: `blur(${(1 - a) * 8}px)` }}>
+ <Mark size={120} color={PAPER} />
+ <span style={{ fontSize: 128, fontWeight: 620, color: PAPER, letterSpacing: "-0.045em", lineHeight: 1 }}>Kokon</span>
+ <span style={{ padding: "8px 22px 12px", borderRadius: 999, border: `3px solid ${PAPER}`, color: PAPER, fontSize: 60, fontWeight: 600, letterSpacing: "-0.03em", lineHeight: 1 }}>0.10</span>
+ </div>
+ <div style={{ fontSize: 40, color: "rgba(247,246,242,0.82)", letterSpacing: "-0.02em", opacity: b, transform: `translateY(${(1 - b) * 14}px)` }}>Free and open source · Windows · macOS · Linux</div>
+ <div style={{ marginTop: 18, padding: "18px 40px", borderRadius: 999, background: PAPER, color: "#0d0d0d", fontSize: 34, fontWeight: 650, letterSpacing: "-0.02em", opacity: c, transform: `translateY(${(1 - c) * 14}px)` }}>kokon.app</div>
+ <div style={{ display: "flex", alignItems: "center", gap: 10, fontSize: 24, color: "rgba(247,246,242,0.62)", opacity: c, transform: `translateY(${(1 - c) * 14}px)` }}>
+ <svg width={22} height={22} viewBox="0 0 24 24" aria-hidden>
+ <path d={siGithub.path} fill="rgba(247,246,242,0.62)" />
+ </svg>
+ github.com/GabrieleLozupone/kokon
+ </div>
+ </div>
+ </AbsoluteFill>
+ );
+}
+
+/** Every effect, on the main timeline (the shots unmount when they end, a sound must not). */
+function Sounds() {
+ const T0 = PROMPT.ask + 2;
+ const T1 = typingDone(PROMPTS.idea, T0, PROMPT.typeCps);
+ return (
+ <>
+ <Sfx name="hit" at={beat(3)} gain={0.55} />
+ <Sfx name="typing" at={T0} dur={T1 - T0} />
+ <Sfx name="click" at={PROMPT.pill} />
+ <Send at={PROMPT.send} />
+ <Sfx name="riser" at={SHOT.hatch - 40} gain={0.8} />
+ <Sfx name="reveal" at={SHOT.hatch + 6} variant={0} />
+ <Sfx name="whoosh" at={SHOT.model} gain={0.7} />
+ <Sfx name="click" at={clickLoad} />
+ <Sfx name="reveal" at={loaded + 2} variant={1} gain={0.6} />
+ <Sfx name="click" at={clickPill} />
+ <Sfx name="whoosh" at={SHOT.look} gain={0.5} />
+ <Sfx name="hit" at={SHOT.end} gain={0.8} />
+ </>
+ );
+}
+
+export function Update010() {
+ return (
+ <AbsoluteFill style={{ background: "#050505" }}>
+ <Shot from={SHOT.title} to={SHOT.composer}>
+ <Title />
+ </Shot>
+ <Shot from={SHOT.composer} to={SHOT.hatch}>
+ <PromptPlus />
+ </Shot>
+ <Shot from={SHOT.hatch} to={SHOT.model}>
+ <Hatched />
+ </Shot>
+ <Shot from={SHOT.model} to={SHOT.settings}>
+ <OneModel />
+ </Shot>
+ <Shot from={SHOT.settings} to={SHOT.look}>
+ <Lora />
+ </Shot>
+ <Shot from={SHOT.look} to={SHOT.end}>
+ <NewLook />
+ </Shot>
+ <Shot from={SHOT.end} to={UPDATE_FRAMES}>
+ <End />
+ </Shot>
+ <Sounds />
+ {VOICE_READY && <Voice cues={CUES_010} />}
+ <Music frames={UPDATE_FRAMES} cues={VOICE_READY ? CUES_010 : []} src={MUSIC.src} from={MUSIC.from} fadeOut={30} />
+ </AbsoluteFill>
+ );
+}