A lightweight 2.5D character player for the web, with lip sync and webcam face tracking.
On npm: install with npm install @livliv-dev/player; the package is the player alone
LivLiv85932741b222
diff --git a/API.md b/API.mdindex d687060..8deee1d 100644--- a/API.md+++ b/API.md@@ -1,14 +1,14 @@-# @livliv/player API+# @livliv-dev/player APIA lightweight 2.5D character player for the web. `Player` animates a LivLiv model (a `.livliv` package, or its folder:`model.json` and `textures/`) on a `<canvas>` with WebGL2; VRM and PNGTuber characters play on the same API through-`@livliv/player/body`. The core has no dependencies (~75 KB minified, ~30 KB gzipped, the renderer and the idle motion included).+`@livliv-dev/player/body`. The core has no dependencies (~75 KB minified, ~30 KB gzipped, the renderer and the idle motion included).Lip sync and face tracking are separate imports, so you only ship them if you use them.-> Not on npm yet. Build it from this repository (see [Building](#building)) and import `dist/index.js`.+> Install it with `npm install @livliv-dev/player`, or build it from this repository (see [Building](#building)).```js-import { Player } from '@livliv/player'+import { Player } from '@livliv-dev/player'const player = new Player(document.querySelector('canvas'))await player.load('/avatars/sena.livliv') // the package as downloaded, or its folder unzipped: '/avatars/sena/'@@ -111,8 +111,8 @@ try {}```-**Errors.** What a load rejects with for a reason of its own is a `LivLivError` (from `@livliv/player` and-`@livliv/player/body`): `code` says what went wrong and `status` the HTTP status. The message is for people and can change.+**Errors.** What a load rejects with for a reason of its own is a `LivLivError` (from `@livliv-dev/player` and+`@livliv-dev/player/body`): `code` says what went wrong and `status` the HTTP status. The message is for people and can change.`new Player()` throws the same class. A camera or microphone keeps the browser's own errors.| `code` | |@@ -252,13 +252,13 @@ the way livliv.ai shows it, so screenshots and recordings of the canvas carry itThe mark is part of the package's licence (`meta.license`), not a lock: the player draws what the package asks for andhas no switch to hide it. A package without it is exported from a paid plan.-## Lip sync: `@livliv/player/voice`+## Lip sync: `@livliv-dev/player/voice`The mouth follows a voice: its loudness, and the vowel's shape where the model has round / wide mouth frames. Withoutface tracking the voice also moves the head, brows and eyes a little, the way people move while they talk.```js-import { createSpeech, startLipSync } from '@livliv/player/voice'+import { createSpeech, startLipSync } from '@livliv-dev/player/voice'// speech clips (TTS replies): played back to back, the mouth following each one exactlyconst speech = createSpeech(player)@@ -277,7 +277,7 @@ syllables, default false). `startLipSync` returns `{ level, gain, stop() }`. A syours: `stop()` leaves it playing. Browsers let an `<audio>` element be routed through Web Audio only once, so after`startLipSync(player, el)` that element always plays through the player's audio context (it still plays normally).-## Face tracking: `@livliv/player/tracking`+## Face tracking: `@livliv-dev/player/tracking`The character follows a face in the webcam, with MediaPipe Face Landmarker running in the browser: camera images stayon the device. Needs the `@mediapipe/tasks-vision` package (1.1.1) and downloads ~8 MB the first time (the wasm runtime@@ -285,7 +285,7 @@ and the face model; cached after that). MediaPipe itself, as Google ships it, al(`odml.pa.googleapis.com`) while it runs.```js-import { preloadFaceModel, startFaceTracking } from '@livliv/player/tracking'+import { preloadFaceModel, startFaceTracking } from '@livliv-dev/player/tracking'cameraButton.onpointerenter = () => preloadFaceModel() // optional: start the download earlyconst tracking = await startFaceTracking(player, {@@ -367,16 +367,16 @@ remove()`move` (`{ x, y, s }`: move / scale the picture this frame). `order` sorts drivers: face tracking -10, yours 0,lip sync 10. `dispose()` is called when a driver is removed.-## Characters that aren't LivLiv models: `@livliv/player/body`+## Characters that aren't LivLiv models: `@livliv-dev/player/body`A `BodyPlayer` plays a character a `Body` draws, with Player's API where it applies: `load()`, `params`, the idlemotion, `setExpression()`, `act()`, `lookAt()`, `blink()`, `use()` and the loop. Lip sync and face tracking work on itas on a Player. Two bodies come with the package:```js-import { BodyPlayer } from '@livliv/player/body'-import { PngBody } from '@livliv/player/png'-import { VrmBody } from '@livliv/player/vrm' // needs three and @pixiv/three-vrm+import { BodyPlayer } from '@livliv-dev/player/body'+import { PngBody } from '@livliv-dev/player/png'+import { VrmBody } from '@livliv-dev/player/vrm' // needs three and @pixiv/three-vrmconst png = new BodyPlayer(canvasA, new PngBody())await png.load({ closed: 'me-closed.png', open: 'me-open.png', blink: 'me-blink.png' }) // name -> URL or Blob@@ -385,11 +385,11 @@ const vrm = new BodyPlayer(canvasB, new VrmBody())await vrm.load('/avatars/me.vrm') // a URL, a File, or files picked```-- **PNGTuber** (`@livliv/player/png`): pictures named `closed` and `open` (the mouth), optionally `blink` and+- **PNGTuber** (`@livliv-dev/player/png`): pictures named `closed` and `open` (the mouth), optionally `blink` and`blink-open` (any image the browser reads, 8192 px a side at most). One is shown at a time: the open one while thevoice is loud, the blink ones when the eyes close. The picture sways, tilts and breathes with the head and body andhops with the voice; there are no expressions but the smile param, and no turn beyond that sway. `name.txt` names it.-- **VRM** (`@livliv/player/vrm`): VRM 0.x and 1.0, drawn with [three.js](https://threejs.org) and+- **VRM** (`@livliv-dev/player/vrm`): VRM 0.x and 1.0, drawn with [three.js](https://threejs.org) and[three-vrm](https://github.com/pixiv/three-vrm), which you install (`npm install three @pixiv/three-vrm`; three.js 0.170or newer, played here on 0.170, 0.186 and 0.187 with three-vrm 3.5.5; optional peer dependencies, so the rest of thepackage does not need them). Files from VRoid Studio and UniVRM play. So does a real file from the VRM add-on for Blender; a fresh export from Blender hasn't been tested, and [issue #1](https://github.com/livliv-dev/player/issues/1) is still open. The head and neck turn, the spine sways, the eyes@@ -435,7 +435,7 @@ an error and the model before stays. Before anything reaches the GPU:- A name, an expression's label and the mark's text are the author's text: show them with `textContent`, not as HTML.These limits are `LIMITS` in `src/runtime/format.ts`; the largest model LivLiv has built uses a tenth of them or less.-A VRM is looked through before three.js's glTF loader parses it (`VRM_LIMITS` in `@livliv/player/vrm`): a binary glTF, with+A VRM is looked through before three.js's glTF loader parses it (`VRM_LIMITS` in `@livliv-dev/player/vrm`): a binary glTF, withnothing outside itself (a file naming an outside URL is refused, so it cannot make the page fetch anything), 32 MB ofJSON, 16 777 216 vertices, pictures of 8192 px a side and 536 870 912 pixels in all, read from their headers before they aredecoded. Security reports: [SECURITY.md](SECURITY.md).diff --git a/CHANGELOG.md b/CHANGELOG.mdindex e3f1534..e7e6086 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -2,12 +2,13 @@## 0.3.0-The interface settled before the package goes on npm: these change what a page written for 0.2 does.+The first version on npm, with the interface settled first: these change what a page written for 0.2 does.### What to change| 0.2 | 0.3 ||---|---|+| `import ... from '@livliv/player'` and its subpaths | `'@livliv-dev/player'`, the package's name on npm. || `player.act('away')` | `player.setAway(true)` and `setAway(false)`. `act('away')` now throws: it was a switch among one-off moves. || `player.puppet` | No longer in the types. `player.poke()` and `player.canvasToModel()` are what pages used it for. The field still exists and can change in any version. || `player.load(a); player.load(b)` played `a`, then `b` | The new load replaces the one running, which rejects with an `AbortError`. If you catch errors, let this one through: `if (e.name === 'AbortError') return`. |@@ -17,7 +18,7 @@ The interface settled before the package goes on npm: these change what a page w### New- `load(source, { signal, onProgress, poster })`: call a load off, follow how far it has come, show a picture until the first frame.-- `LivLivError`, with a `code` for each reason, from `@livliv/player` and `@livliv/player/body`.+- `LivLivError`, with a `code` for each reason, from `@livliv-dev/player` and `@livliv-dev/player/body`.- `player.contextLost` and `player.onContextLost`: the page hears when the WebGL context is gone. See [API.md](API.md#when-the-context-is-lost).- `dispose({ release: true })`: gives the WebGL context up, for a canvas leaving the page.- `player.setAway()`, `player.poke()`, `player.canvasToModel()`.diff --git a/FORMAT.md b/FORMAT.mdindex 4de4043..98806d7 100644--- a/FORMAT.md+++ b/FORMAT.md@@ -2,7 +2,7 @@A LivLiv model is a 2D character cut into layers, with the anchors and rig settings a player needs to move it: headturns, blinks, gaze, brows, lip sync, breathing, hair that swings, and expressions. This document describes the package-that [livliv.ai](https://livliv.ai) exports and that LivLiv Player (`@livliv/player`) and LivLiv Stage read.+that [livliv.ai](https://livliv.ai) exports and that LivLiv Player (`@livliv-dev/player`) and LivLiv Stage read.- JSON Schema: [`schema/livliv-model-v4.json`](schema/livliv-model-v4.json) (`model.json`) and[`schema/livliv-keyforms-v1.json`](schema/livliv-keyforms-v1.json) (`keyforms.json`)diff --git a/README.ja.md b/README.ja.mdindex b335610..a3f5912 100644--- a/README.ja.md+++ b/README.ja.md@@ -29,8 +29,12 @@## 自分のページで使う+```sh+npm install @livliv-dev/player+```+```js-import { Player } from '@livliv/player'+import { Player } from '@livliv-dev/player'const player = new Player(document.querySelector('canvas'))await player.load('characters/k01/') // model.json と textures/ のあるフォルダ、または .livliv ファイル@@ -43,8 +47,8 @@ player.params.mouthOpen = 0.6 // 0..1リップシンクとフェイストラッキング:```js-import { createSpeech } from '@livliv/player/voice'-import { startFaceTracking } from '@livliv/player/tracking' // @mediapipe/tasks-vision が必要+import { createSpeech } from '@livliv-dev/player/voice'+import { startFaceTracking } from '@livliv-dev/player/tracking' // @mediapipe/tasks-vision が必要const speech = createSpeech(player)button.onclick = () => { speech.unlock(); speech.say('/reply.mp3') }@@ -55,9 +59,9 @@ await startFaceTracking(player)VRM や PNGTuber も同じ API で:```js-import { BodyPlayer } from '@livliv/player/body'-import { VrmBody } from '@livliv/player/vrm' // three と @pixiv/three-vrm が必要-import { PngBody } from '@livliv/player/png'+import { BodyPlayer } from '@livliv-dev/player/body'+import { VrmBody } from '@livliv-dev/player/vrm' // three と @pixiv/three-vrm が必要+import { PngBody } from '@livliv-dev/player/png'const vrm = new BodyPlayer(canvasA, new VrmBody())await vrm.load('/avatars/me.vrm')@@ -67,7 +71,7 @@ await png.load({ closed: 'me-closed.png', open: 'me-open.png' })API の全体は [API.md](API.md)(英語)にあります:オプション、パラメーター、表情とアクション、ドライバー、VRM と PNGTuber、信頼できないファイルの上限、モデルパッケージ、フォーマットのバージョン。LivLiv のパッケージ形式(`.livliv`、`model.json` v4)は [FORMAT.md](FORMAT.md)、JSON Schema は [`schema/`](schema/) にあります。v4 より前に書き出したパッケージは [FORMAT-v3.md](FORMAT-v3.md) のとおりで、そのまま再生できます。-> まだ npm には公開していません。このリポジトリからビルドして `dist/index.js` をインポートしてください。+> npm パッケージに入っているのはプレイヤーだけです。厳選キャラクターはこのリポジトリの [`characters/`](characters/) にあります。使うキャラクターをライセンスと一緒にページのそばにコピーしてください。## 厳選キャラクターdiff --git a/README.md b/README.mdindex 835771c..8f959ed 100644--- a/README.md+++ b/README.md@@ -30,8 +30,12 @@## Use it in your page+```sh+npm install @livliv-dev/player+```+```js-import { Player } from '@livliv/player'+import { Player } from '@livliv-dev/player'const player = new Player(document.querySelector('canvas'))await player.load('characters/k01/') // a folder with model.json and textures/, or a .livliv file@@ -44,8 +48,8 @@ player.params.mouthOpen = 0.6 // 0..1Lip sync and face tracking:```js-import { createSpeech } from '@livliv/player/voice'-import { startFaceTracking } from '@livliv/player/tracking' // needs @mediapipe/tasks-vision+import { createSpeech } from '@livliv-dev/player/voice'+import { startFaceTracking } from '@livliv-dev/player/tracking' // needs @mediapipe/tasks-visionconst speech = createSpeech(player)button.onclick = () => { speech.unlock(); speech.say('/reply.mp3') }@@ -56,9 +60,9 @@ await startFaceTracking(player)A VRM or a PNGTuber, on the same API:```js-import { BodyPlayer } from '@livliv/player/body'-import { VrmBody } from '@livliv/player/vrm' // needs three and @pixiv/three-vrm-import { PngBody } from '@livliv/player/png'+import { BodyPlayer } from '@livliv-dev/player/body'+import { VrmBody } from '@livliv-dev/player/vrm' // needs three and @pixiv/three-vrm+import { PngBody } from '@livliv-dev/player/png'const vrm = new BodyPlayer(canvasA, new VrmBody())await vrm.load('/avatars/me.vrm')@@ -71,7 +75,8 @@ the limits on untrusted files, the model package and the format version. The Liv`model.json` v4) is in [FORMAT.md](FORMAT.md), with a JSON Schema in [`schema/`](schema/); packages exported before v4are described in [FORMAT-v3.md](FORMAT-v3.md) and still play.-> Not on npm yet: build it from this repository and import `dist/index.js`.+> The npm package is the player alone. The featured characters are in this repository's [`characters/`](characters/): copy+> the ones you use next to your page, with their licence.## Featured charactersdiff --git a/README.zh-CN.md b/README.zh-CN.mdindex 80ddeaa..10f9c64 100644--- a/README.zh-CN.md+++ b/README.zh-CN.md@@ -31,8 +31,12 @@## 在自己的页面里用+```sh+npm install @livliv-dev/player+```+```js-import { Player } from '@livliv/player'+import { Player } from '@livliv-dev/player'const player = new Player(document.querySelector('canvas'))await player.load('characters/k01/') // 含 model.json 和 textures/ 的文件夹,或一个 .livliv 文件@@ -45,8 +49,8 @@ player.params.mouthOpen = 0.6 // 0..1口型同步和面捕:```js-import { createSpeech } from '@livliv/player/voice'-import { startFaceTracking } from '@livliv/player/tracking' // 需要 @mediapipe/tasks-vision+import { createSpeech } from '@livliv-dev/player/voice'+import { startFaceTracking } from '@livliv-dev/player/tracking' // 需要 @mediapipe/tasks-visionconst speech = createSpeech(player)button.onclick = () => { speech.unlock(); speech.say('/reply.mp3') }@@ -57,9 +61,9 @@ await startFaceTracking(player)VRM 和 PNGTuber 用同一套接口:```js-import { BodyPlayer } from '@livliv/player/body'-import { VrmBody } from '@livliv/player/vrm' // 需要 three 和 @pixiv/three-vrm-import { PngBody } from '@livliv/player/png'+import { BodyPlayer } from '@livliv-dev/player/body'+import { VrmBody } from '@livliv-dev/player/vrm' // 需要 three 和 @pixiv/three-vrm+import { PngBody } from '@livliv-dev/player/png'const vrm = new BodyPlayer(canvasA, new VrmBody())await vrm.load('/avatars/me.vrm')@@ -69,7 +73,7 @@ await png.load({ closed: 'me-closed.png', open: 'me-open.png' })完整的 API 见 [API.md](API.md)(英文):选项、参数、表情和动作、驱动、VRM 和 PNGTuber、不可信文件的上限、模型包和格式版本。LivLiv 模型包格式(`.livliv`、`model.json` v4)见 [FORMAT.md](FORMAT.md),JSON Schema 在 [`schema/`](schema/);v4 之前导出的模型包见 [FORMAT-v3.md](FORMAT-v3.md),照样能播放。-> 还没有发布到 npm:请从这个仓库构建,引入 `dist/index.js`。+> npm 包里只有播放器。精选角色在这个仓库的 [`characters/`](characters/) 里,把要用的角色连同许可一起复制到你的页面旁边。## 精选角色diff --git a/package.json b/package.jsonindex ec0626b..1067b0a 100644--- a/package.json+++ b/package.json@@ -1,10 +1,9 @@{- "name": "@livliv/player",+ "name": "@livliv-dev/player","version": "0.3.0","description": "LivLiv Player: a lightweight 2.5D character player for the web, with lip sync and webcam face tracking. Plays LivLiv, VRM and PNGTuber characters; WebGL2, no dependencies in the core.",- "private": true,"license": "Apache-2.0",- "homepage": "https://livliv.ai/developers",+ "homepage": "https://github.com/livliv-dev/player#readme","repository": {"type": "git","url": "git+https://github.com/livliv-dev/player.git"@@ -91,12 +90,17 @@"typescript": "~6.0.3","vite": "^8.3.4"},+ "publishConfig": {+ "access": "public"+ },"scripts": {"build": "rm -rf dist && vite build && tsc -p tsconfig.build.json && node scripts/dts-extensions.mjs && node scripts/pack-characters.mjs","test:types": "node test/types.mjs","check": "tsc -p tsconfig.json","test": "node --disable-warning=ExperimentalWarning --import ./test/ts.mjs --test \"test/*.test.ts\"","test:browser": "node test/smoke.mjs",- "test:engines": "node test/smoke.mjs firefox && node test/smoke.mjs webkit"+ "test:engines": "node test/smoke.mjs firefox && node test/smoke.mjs webkit",+ "prepublishOnly": "npm run check && npm test",+ "prepack": "npm run build"}}diff --git a/src/base.ts b/src/base.tsindex 5d03db2..5598937 100644--- a/src/base.ts+++ b/src/base.ts@@ -59,7 +59,7 @@ export interface Driver {dispose?(): void}-/** what a driver, the lip sync and the face tracking need of a player: a Player, or a BodyPlayer (@livliv/player/body) */+/** what a driver, the lip sync and the face tracking need of a player: a Player, or a BodyPlayer (@livliv-dev/player/body) */export interface Host {/** adds a driver; returns a function that removes it */use(driver: Driver): () => voiddiff --git a/src/body.ts b/src/body.tsindex 24acc69..fa19509 100644--- a/src/body.ts+++ b/src/body.ts@@ -1,7 +1,7 @@-// @livliv/player/body: characters that aren't LivLiv models, on the surface of Player. A Body draws one (a PNGTuber's-// pictures: @livliv/player/png; a VRM: @livliv/player/vrm; or your own); the BodyPlayer gives it what a LivLiv model gets-// from Player: the idle motion, expressions, actions and drivers, so lip sync (@livliv/player/voice) and face tracking-// (@livliv/player/tracking) work for it the same way. Each frame the pose is put together as Player does it (your params,+// @livliv-dev/player/body: characters that aren't LivLiv models, on the surface of Player. A Body draws one (a PNGTuber's+// pictures: @livliv-dev/player/png; a VRM: @livliv-dev/player/vrm; or your own); the BodyPlayer gives it what a LivLiv model gets+// from Player: the idle motion, expressions, actions and drivers, so lip sync (@livliv-dev/player/voice) and face tracking+// (@livliv-dev/player/tracking) work for it the same way. Each frame the pose is put together as Player does it (your params,// the idle motion, the drivers, actions, expressions); the Body turns it into its own picture.import { pickLabel, type Labels } from './runtime/format'import { defaultParams, type Params } from './runtime/params'diff --git a/src/index.ts b/src/index.tsindex 63476b8..bd80b8d 100644--- a/src/index.ts+++ b/src/index.ts@@ -1,7 +1,7 @@-// @livliv/player: plays a LivLiv model (an exported model package: model.json and textures/) on a canvas, outside+// @livliv-dev/player: plays a LivLiv model (an exported model package: model.json and textures/) on a canvas, outside// livliv.ai. The renderer is livliv.ai's own (src/runtime/puppet.ts); this wraps it in a small API: load, a frame// loop with the idle motion, params, expressions, actions, the corner mark the package carries, and drivers. Face-// tracking and lip sync are separate imports ('@livliv/player/tracking', '@livliv/player/voice'), so the core stays small.+// tracking and lip sync are separate imports ('@livliv-dev/player/tracking', '@livliv-dev/player/voice'), so the core stays small.import { langOf, pickLabel, readModel, type Labels, type ModelMeta } from './runtime/format'import { FORMAT, FORMAT_VERSION, LEGACY_FORMAT, PARAM_RANGES, Puppet, defaultParams, loadBlob, type KeyformFile, type ModelDef, type Params } from './runtime/puppet'import { LivLivError, PlayerBase, aborted, contextGone, readBody, type LivLivErrorCode, type LoadJob, type LoadOptions, type PlayerOptions } from './base'@@ -183,7 +183,7 @@ function explain(e: unknown, where: string): unknown {}/** a file a Player cannot take but a BodyPlayer can: the words that say so */-const vrmHere = () => new LivLivError('not-a-model', 'LivLiv: this is a VRM or glTF file, and a Player plays LivLiv models: use a BodyPlayer with VrmBody (@livliv/player/vrm)')+const vrmHere = () => new LivLivError('not-a-model', 'LivLiv: this is a VRM or glTF file, and a Player plays LivLiv models: use a BodyPlayer with VrmBody (@livliv-dev/player/vrm)')const isGlb = async (b: Blob) => { const d = new DataView(await b.slice(0, 4).arrayBuffer()); return d.byteLength === 4 && d.getUint32(0, true) === 0x46546c67 }/** a .livliv package from a URL, as a Blob. One file is one connection to lose where a folder has dozens, and a dropped onediff --git a/src/png.ts b/src/png.tsindex 7982877..790e9cf 100644--- a/src/png.ts+++ b/src/png.ts@@ -1,4 +1,4 @@-// @livliv/player/png: a PNGTuber on a BodyPlayer (@livliv/player/body). A few pictures of the character, one shown at a+// @livliv-dev/player/png: a PNGTuber on a BodyPlayer (@livliv-dev/player/body). A few pictures of the character, one shown at a// time: the mouth closed or open (the voice, the microphone), the eyes closed when it blinks (optional, and with the// mouth open too). The picture sways with the head, tilts, breathes and hops with the voice, standing on the bottom of// the canvas. Files by name (any image the browser decodes): closed, open, blink, blink-open, and name.txt.diff --git a/src/tracking.ts b/src/tracking.tsindex 3a20b15..2d5c337 100644--- a/src/tracking.ts+++ b/src/tracking.ts@@ -1,4 +1,4 @@-// @livliv/player/tracking: the character follows a face in the webcam (MediaPipe Face Landmarker, in the browser; camera+// @livliv-dev/player/tracking: the character follows a face in the webcam (MediaPipe Face Landmarker, in the browser; camera// images stay on the device, though MediaPipe sends Google usage statistics while it runs). A separate import: it needs// the @mediapipe/tasks-vision package and downloads ~8 MB of model the first time (the wasm runtime and the face model;// cached by the browser after that).diff --git a/src/voice.ts b/src/voice.tsindex 19d4888..dab0735 100644--- a/src/voice.ts+++ b/src/voice.ts@@ -1,4 +1,4 @@-// @livliv/player/voice: lip sync. The mouth follows a voice (loudness, and the vowel's shape where the model has round /+// @livliv-dev/player/voice: lip sync. The mouth follows a voice (loudness, and the vowel's shape where the model has round /// wide mouths), and without face tracking the voice also moves the head, brows and eyes a little, the way people move// while they talk. Two ways in:// - startLipSync: live audio (the microphone, a call's MediaStream, or an <audio> / <video> element the page plays)diff --git a/src/vrm.ts b/src/vrm.tsindex e42f878..b899f3c 100644--- a/src/vrm.ts+++ b/src/vrm.ts@@ -1,4 +1,4 @@-// @livliv/player/vrm: a VRM (the open 3D avatar format: VRoid and others) on a BodyPlayer (@livliv/player/body), drawn with+// @livliv-dev/player/vrm: a VRM (the open 3D avatar format: VRoid and others) on a BodyPlayer (@livliv-dev/player/body), drawn with// three.js and @pixiv/three-vrm, which the page installs (optional peer dependencies) and which load only here. The// shared params become its pose: the head and neck turn and tilt (angleX / Y / Z), the spine sways (bodyX / Z), the// eyes blink (eyeLOpen / ROpen) and look (eyeBallX / Y), the mouth takes the visemes (mouthOpen with its vowel: aa open,