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
import { AmbientLight, DirectionalLight, HemisphereLight } from 'three';
import { Preset } from './internal.js';
export function addLights(viewer) {
const { state, options } = viewer;
if (options.preset === Preset.ASSET_GENERATOR) {
const hemiLight = new HemisphereLight();
hemiLight.name = 'hemi_light';
viewer.scene.add(hemiLight);
viewer.lights.push(hemiLight);
return;
}
// Studio three-point rig. Ambient, key and fill ride the camera so the
// avatar stays evenly lit from the viewer's angle as it orbits; the rim/back
// light is anchored in world space behind-and-above the subject to keep a
// consistent edge highlight that separates dark avatars from dark cards.
const ambient = new AmbientLight(state.ambientColor, state.ambientIntensity);
ambient.name = 'ambient_light';
viewer.defaultCamera.add(ambient);
const key = new DirectionalLight(state.directColor, state.directIntensity);
key.position.set(0.5, 0.6, 0.866); // upper front-right, ~60º
key.name = 'main_light';
viewer.defaultCamera.add(key);
const fill = new DirectionalLight(
state.fillColor ?? '#DCE6FF',
state.directIntensity * (state.fillRatio ?? 0.4),
);
fill.position.set(-0.9, 0.1, 0.6); // opposite side, near eye level
fill.name = 'fill_light';
viewer.defaultCamera.add(fill);
const rim = new DirectionalLight(
state.rimColor ?? '#FFFFFF',
state.directIntensity * (state.rimRatio ?? 0.65),
);
rim.position.set(-1.2, 2.4, -2.6); // behind-above, throws a hair/shoulder edge
rim.name = 'rim_light';
viewer.scene.add(rim);
viewer.lights.push(ambient, key, fill, rim);
// World-anchored overhead light whose only real job is the soft contact
// shadow on the ground catcher (see viewer.js _updateShadowCatcher). It is
// deliberately NOT the camera-riding key: a camera-space caster would spin
// the shadow around the model as the user orbits. Intensity stays low so
// the calibrated studio rig above still defines the look. Skipped on
// low-power devices along with the shadow map itself.
if (!viewer._lowPower) {
const shadowSun = new DirectionalLight('#FFFFFF', 0.25);
shadowSun.position.set(1.5, 4, 2.5);
shadowSun.castShadow = true;
shadowSun.shadow.mapSize.set(1024, 1024);
shadowSun.shadow.bias = -0.002;
shadowSun.name = 'shadow_light';
viewer.scene.add(shadowSun);
viewer.lights.push(shadowSun);
}
}
export function removeLights(viewer) {
viewer.lights.forEach((light) => light.parent?.remove(light));
viewer.lights.length = 0;
}