Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.
JavaScript 3d-generationai-agentsanimationavatardigital-humangenerative-aiglbgltfimage-to-3dllmmcpmcp-servermodel-context-protocolsolanatext-to-3dthreejsweb-componentswebglwebxrx402
/**
* Tests for the Avatar Studio export optimizer.
*
* The save flow's central contract is: optimization is best-effort and must
* NEVER make the save fail. These tests pin that guarantee — on invalid input
* the optimizer falls back to the original blob instead of throwing — and that
* a real, structurally-valid GLB round-trips smaller through the pipeline.
*/
import { describe, it, expect } from 'vitest';
import { optimizeAndValidateGlb } from '../src/avatar-studio-optimize.js';
// Minimal valid GLB: a 12-byte header + JSON chunk describing an empty asset.
// The glTF-Transform reader accepts this; it has no meshes, so the compression
// passes are no-ops and the output is not smaller — exercising the
// "not smaller → keep original" branch.
function emptyGlb() {
const json = JSON.stringify({ asset: { version: '2.0' } });
const enc = new TextEncoder();
let jsonBytes = enc.encode(json);
// JSON chunk must be 4-byte aligned, padded with spaces.
const pad = (4 - (jsonBytes.length % 4)) % 4;
if (pad) {
const padded = new Uint8Array(jsonBytes.length + pad).fill(0x20);
padded.set(jsonBytes);
jsonBytes = padded;
}
const total = 12 + 8 + jsonBytes.length;
const buf = new ArrayBuffer(total);
const dv = new DataView(buf);
dv.setUint32(0, 0x46546c67, true); // 'glTF'
dv.setUint32(4, 2, true); // version
dv.setUint32(8, total, true); // total length
dv.setUint32(12, jsonBytes.length, true); // chunk length
dv.setUint32(16, 0x4e4f534a, true); // 'JSON'
new Uint8Array(buf, 20).set(jsonBytes);
return new Blob([buf], { type: 'model/gltf-binary' });
}
describe('optimizeAndValidateGlb', () => {
it('falls back to the original blob on non-GLB input without throwing', async () => {
const garbage = new Blob([new Uint8Array([1, 2, 3, 4, 5, 6, 7, 8])], {
type: 'model/gltf-binary',
});
const result = await optimizeAndValidateGlb(garbage);
expect(result.optimized).toBe(false);
expect(result.blob).toBe(garbage); // same reference — original returned untouched
expect(result.sourceBytes).toBe(garbage.size);
expect(result.outputBytes).toBe(garbage.size);
});
it('keeps the original when the optimized output is not smaller', async () => {
const glb = emptyGlb();
const result = await optimizeAndValidateGlb(glb);
// An empty asset has nothing to compress; the meshopt re-encode never
// beats the tiny source, so we keep the original rather than regress.
expect(result.optimized).toBe(false);
expect(result.blob).toBe(glb);
});
it('always resolves to a Blob and consistent size fields', async () => {
const glb = emptyGlb();
const result = await optimizeAndValidateGlb(glb);
expect(result.blob).toBeInstanceOf(Blob);
expect(result.outputBytes).toBe(result.blob.size);
expect(typeof result.sourceBytes).toBe('number');
});
});