MCPInstall

threejs-textures

Three.js textures: image loading, color spaces, UV channels and transforms, texture types from canvas to HDR and KTX2, render targets, texture memory. Use when working with images on materials, UV coordinates, cubemaps, render targets or texture optimization; on WebGL or with WebGPU and TSL.

Installs
12.9K
Last 24h
+252
Trending rank
#1078
Security
Pass

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/cloudai-x/threejs-skills --skill threejs-textures
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads threejs-textures automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

Security audits

Independent scans of the skill's instructions and bundled files.

Gen Agent Trust Hub PassSAFE ·The skill is a technical guide for using textures in Three.js. It provides standard code examples for loading, configuring, and managing various texture types like cube maps and HDR environments. No malicious patterns or security risks were identified.
Socket PassNo alerts
Snyk PassLOW ·Risk: LOW · No issues
Runlayer PassNONE ·1 file scanned · No issues
ZeroLeaks PassNONE ·Score: 93/100 · 2 sections analyzed

SKILL.md

The instructions your agent reads when the skill is triggered.

Three.js Textures

Targets three.js r186. Examples use WebGLRenderer; the WebGPU and TSL section uses WebGPURenderer.

Quick Start

import * as THREE from "three";

const renderer = new THREE.WebGLRenderer({ antialias: true });
renderer.setPixelRatio(window.devicePixelRatio);
renderer.setSize(window.innerWidth, window.innerHeight);
document.body.appendChild(renderer.domElement);

const scene = new THREE.Scene();
const camera = new THREE.PerspectiveCamera(60, window.innerWidth / window.innerHeight, 0.1, 100);
camera.position.set(0, 0, 2);

// Color images must be tagged sRGB or they render too bright and washed out
const texture = new THREE.TextureLoader().load("diffuse.jpg");
texture.colorSpace = THREE.SRGBColorSpace;

const mesh = new THREE.Mesh(
  new THREE.PlaneGeometry(2, 2),
  new THREE.MeshBasicMaterial({ map: texture }),
);
scene.add(mesh);

renderer.setAnimationLoop(animate);

function animate() {
  mesh.rotation.z += 0.01;
  renderer.render(scene, camera);
}

Loading

// load() returns the texture at once; its image arrives asynchronously
const texture = new THREE.TextureLoader().load("diffuse.jpg");
texture.colorSpace = THREE.SRGBColorSpace;
mesh.material.map = texture; // before the first render; afterwards also set mesh.material.needsUpdate = true

// When the next step needs the loaded image, await it
const colorMap = await new THREE.TextureLoader().loadAsync("diffuse.jpg");
colorMap.colorSpace = THREE.SRGBColorSpace;

TextureLoader's onProgress parameter is unsupported and never fires; a LoadingManager reports progress as each file finishes:

const manager = new THREE.LoadingManager();
manager.onProgress = (url, loaded, total) => console.log(`${loaded}/${total}: ${url}`);
const loader = new THREE.TextureLoader(manager);

Color Space

Color management is on by default and every texture starts at NoColorSpace:

  • LDR color images get SRGBColorSpace: map, emissiveMap, specularMap, sheen color and specular color maps, LDR matcaps, and canvas or video used as color.
  • Scalar data keeps NoColorSpace: normal, bump, displacement, roughness, metalness, ambient-occlusion and alpha maps.
  • HDR environments are LinearSRGBColorSpace; the HDR loaders set that themselves. GLTFLoader sets the color spaces of the textures it loads.
const colorMap = new THREE.TextureLoader().load("color.jpg");
colorMap.colorSpace = THREE.SRGBColorSpace; // authored color image

const normalMap = new THREE.TextureLoader().load("normal.jpg");
// normalMap stays at NoColorSpace: it is data, not color

Transform: Wrap, Repeat, Offset, Rotation

const texture = new THREE.TextureLoader().load("diffuse.jpg");

// Tiling requires repeat wrapping; the default clamps edge pixels
texture.wrapS = THREE.RepeatWrapping;
texture.wrapT = THREE.RepeatWrapping;
texture.repeat.set(4, 4); // tile 4x4

texture.offset.set(0.5, 0.5); // shift in UV units
texture.center.set(0.5, 0.5); // pivot for rotation
texture.rotation = Math.PI / 4; // radians

MirroredRepeatWrapping tiles with a mirror flip on each repeat.

Filtering, Mipmaps, Anisotropy

const texture = new THREE.TextureLoader().load("diffuse.jpg");

// Minification (many texture pixels per screen pixel)
texture.minFilter = THREE.LinearMipmapLinearFilter; // default: trilinear blend between mip levels
texture.minFilter = THREE.NearestFilter; // hard pixel look, samples only the base level

// Magnification (texture magnified on screen)
texture.magFilter = THREE.LinearFilter; // default
texture.magFilter = THREE.NearestFilter; // pixelated

// Mipmaps are generated whenever generateMipmaps is true, whatever the filters.
// Turn generation off to save memory (UI sprites, data textures):
texture.generateMipmaps = false;
texture.minFilter = THREE.LinearFilter; // and do not request a mipmap filter

// Sharper at grazing angles; clamp to what the hardware supports
texture.anisotropy = renderer.capabilities.getMaxAnisotropy();

Texture Types

Base class from any image element:

const image = new Image();
image.src = "sprite.png";

const imageTexture = new THREE.Texture(image);
imageTexture.colorSpace = THREE.SRGBColorSpace;
image.onload = () => (imageTexture.needsUpdate = true); // upload once the pixels exist

DataTexture from raw typed-array pixels. Its defaults differ from Texture: NearestFilter for both filters, flipY false, mipmaps off:

const size = 256;
const data = new Uint8Array(size * size * 4); // RGBA, one byte per channel
for (let i = 0; i < size; i++) {
  for (let j = 0; j < size; j++) {
    const index = (i * size + j) * 4;
    data[index] = i;
    data[index + 1] = j;
    data[index + 2] = 128;
    data[index + 3] = 255;
  }
}
const dataTexture = new THREE.DataTexture(data, size, size);
dataTexture.needsUpdate = true;

CanvasTexture for dynamically drawn textures (sets needsUpdate on construction):

const canvas = document.createElement("canvas");
canvas.width = 256;
canvas.height = 256;
const ctx = canvas.getContext("2d");
ctx.fillStyle = "#404040";
ctx.fillRect(0, 0, 256, 256);
ctx.fillStyle = "#ffffff";
ctx.font = "48px sans-serif";
ctx.fillText("Hello", 40, 150);

const canvasTexture = new THREE.CanvasTexture(canvas);
canvasTexture.colorSpace = THREE.SRGBColorSpace; // canvas colors are sRGB

// after each redraw of the canvas:
canvasTexture.needsUpdate = true;

VideoTexture for a playing HTMLVideoElement (re-uploads every frame, mipmaps off):

const video = document.createElement("video");
video.src = "video.mp4";
video.loop = true;
video.muted = true;
video.play();

const videoTexture = new THREE.VideoTexture(video);
videoTexture.colorSpace = THREE.SRGBColorSpace;

VideoFrameTexture for WebCodecs: each decoded frame is pushed by hand:

const frameTexture = new THREE.VideoFrameTexture();
frameTexture.colorSpace = THREE.SRGBColorSpace;
const decoder = new VideoDecoder({
  output(frame) {
    if (frameTexture.image instanceof VideoFrame) frameTexture.image.close(); // free the replaced frame, or decoding stalls
    frameTexture.setFrame(frame);
  },
  error: (e) => console.error(e),
});

HTMLTexture draws an HTML element through the browser's HTML-in-Canvas API: the renderer moves the element into its own canvas and re-uploads it on paint events; where the browser lacks that API the texture stays empty.

Compressed Textures (KTX2)

import { KTX2Loader } from "three/addons/loaders/KTX2Loader.js";

const ktx2Loader = new KTX2Loader();
// No setTranscoderPath call: the transcoder loads from the three package.
// Set one only when self-hosting the decoders, pinned to this release.
ktx2Loader.detectSupport(renderer); // required before loading; works with both renderers

const compressedMap = await ktx2Loader.loadAsync("diffuse.ktx2");
// colorSpace comes from the file's transfer function; nothing to set by hand

Basis Universal KTX2 transcodes to a compressed format the GPU supports: smaller uploads and less memory than PNG or JPEG, which upload uncompressed (where the GPU supports none of the targets, KTX2Loader also falls back to uncompressed RGBA). Compressed textures cannot be flipped or mipmapped at upload: flipY stays false, so author the file in the orientation your UVs expect, and mipmaps must be embedded in the file.

Environment Maps

Cube maps (the loader tags the result sRGB itself):

const cubeTexture = new THREE.CubeTextureLoader().load([
  "px.jpg", "nx.jpg", // +X, -X
  "py.jpg", "ny.jpg", // +Y, -Y
  "pz.jpg", "nz.jpg", // +Z, -Z
]);

scene.background = cubeTexture;
scene.environment = cubeTexture; // lights Standard and Physical materials (WebGLRenderer also Lambert and Phong)

HDR equirectangular files — set mapping, then assign; both renderers prefilter the environment themselves:

import { HDRLoader } from "three/addons/loaders/HDRLoader.js";

const env = await new HDRLoader().loadAsync("environment.hdr");
env.mapping = THREE.EquirectangularReflectionMapping; // without it the background draws as a flat screen quad
scene.environment = env;
scene.background = env;

EXRLoader (three/addons/loaders/EXRLoader.js) loads .exr files the same way and UltraHDRLoader loads Ultra HDR JPEGs; both set their own linear color space.

Scene-level controls:

scene.environmentIntensity = 0.5; // dim the image-based lighting
scene.environmentRotation.y = Math.PI / 4;

scene.backgroundBlurriness = 0.5; // 0-1
scene.backgroundIntensity = 1.0;
scene.backgroundRotation.y = Math.PI;

Both renderers convert an equirectangular environment into a cube map with a larger texture internally (better sampling, more memory); when memory matters, ship a lower-resolution equirect file.

PMREMGenerator is for generated environments — a scene such as RoomEnvironment captured with fromScene:

import { RoomEnvironment } from "three/addons/environments/RoomEnvironment.js";

const pmrem = new THREE.PMREMGenerator(renderer);
const room = new RoomEnvironment();
scene.environment = pmrem.fromScene(room, 0.04).texture;
room.dispose(); // frees the room's geometries and materials
pmrem.dispose();

Do not run a loaded equirect file through PMREMGenerator "for better reflections" — direct assignment with mapping set produces the same prefiltering.

Render Targets

const renderTarget = new THREE.WebGLRenderTarget(512, 512);
const rtScene = new THREE.Scene(); // must not contain the mesh that samples the target
rtScene.add(new THREE.Mesh(new THREE.BoxGeometry(), new THREE.MeshNormalMaterial())); // unlit, visible without lights

renderer.setRenderTarget(renderTarget);
renderer.render(rtScene, camera);
renderer.setRenderTarget(null); // back to the canvas

// the output is a texture like any other
mesh.material.map = renderTarget.texture;

Depth buffer as a texture:

const depthTarget = new THREE.WebGLRenderTarget(512, 512);
depthTarget.depthTexture = new THREE.DepthTexture(512, 512); // default type UnsignedIntType

MSAA and multiple attachments:

// MSAA: resolved into .texture automatically after rendering
const msaaTarget = new THREE.WebGLRenderTarget(512, 512, { samples: 4 });

// MRT: N color attachments, read back through .textures
const mrtTarget = new THREE.WebGLRenderTarget(512, 512, { count: 2 });
const firstAttachment = mrtTarget.textures[0];
const secondAttachment = mrtTarget.textures[1];

Dynamic Reflections

const cubeRenderTarget = new THREE.WebGLCubeRenderTarget(256, {
  generateMipmaps: true,
  minFilter: THREE.LinearMipmapLinearFilter,
});
const cubeCamera = new THREE.CubeCamera(0.1, 1000, cubeRenderTarget);
scene.add(cubeCamera);

mesh.material.envMap = cubeRenderTarget.texture;

// each frame: hide the reflective mesh so it does not see itself
mesh.visible = false;
cubeCamera.position.copy(mesh.position);
cubeCamera.update(renderer, scene);
mesh.visible = true;

UV Channels

Every UV-mapped material texture samples the UV attribute chosen by texture.channel: 0 reads uv (the default), 1 reads uv1, 2 reads uv2, 3 reads uv3. So aoMap and lightMap work on the default uv set; a separately unwrapped set goes in uv1 with the texture pointed at it:

const aoTexture = new THREE.TextureLoader().load("ao.jpg");

const geometry = new THREE.BoxGeometry(1, 1, 1);
const uvAttribute = geometry.attributes.uv;
const aoUvs = new Float32Array(uvAttribute.count * 2); // fill with the second unwrap
geometry.setAttribute("uv1", new THREE.BufferAttribute(aoUvs, 2));

const material = new THREE.MeshStandardMaterial({ aoMap: aoTexture });
material.aoMap.channel = 1; // sample "uv1" instead of "uv"
const aoMesh = new THREE.Mesh(geometry, material);
scene.add(aoMesh);

Texture Atlas

One image, many sprites, each selected with offset and repeat:

const atlas = await new THREE.TextureLoader().loadAsync("atlas.png");
atlas.colorSpace = THREE.SRGBColorSpace;
// filtering and mipmaps sample across sprite borders: leave padding around each sprite

// one clone per sprite: clones share the image source, so WebGLRenderer uploads it once
function spriteTexture(row, col, gridSize = 2) {
  const sprite = atlas.clone();
  sprite.offset.set(col / gridSize, 1 - (row + 1) / gridSize);
  sprite.repeat.set(1 / gridSize, 1 / gridSize);
  return sprite;
}

mesh.material.map = spriteTexture(0, 0); // top-left sprite of a 2x2 sheet

Standard Material Slots

Channel conventions worth knowing: aoMap reads red, roughnessMap reads green, metalnessMap reads blue (so one packed occlusion-roughness-metalness image works); alphaMap reads green.

const colorMap = new THREE.TextureLoader().load("color.jpg");
const emissiveMap = new THREE.TextureLoader().load("emissive.jpg");
const normalMap = new THREE.TextureLoader().load("normal.jpg");
colorMap.colorSpace = THREE.SRGBColorSpace;
emissiveMap.colorSpace = THREE.SRGBColorSpace;
// normalMap keeps NoColorSpace, as do roughness, metalness, ao, alpha and displacement maps

const material = new THREE.MeshStandardMaterial({
  map: colorMap,
  normalMap: normalMap,
  normalScale: new THREE.Vector2(1, 1),
  emissiveMap: emissiveMap,
  emissive: 0xffffff, // non-black, so the emissive map contributes
  roughness: 1, // scalar multipliers for the matching maps
  metalness: 0,
});

material.aoMap = new THREE.TextureLoader().load("ao.jpg");
material.aoMapIntensity = 1;

alphaMap takes effect with transparent: true (blended) or alphaTest (cut out); displacementMap scales with displacementScale and displacementBias and wants a subdivided mesh; normalMapType defaults to TangentSpaceNormalMap, set ObjectSpaceNormalMap for object-space normal maps.

Disposal and Memory

texture.dispose() frees the GPU upload once no other undisposed texture shares it (the atlas clones above share one); the image object remains. Dispose only textures that no other material still uses:

function disposeMaterial(material) {
  for (const key of Object.keys(material)) {
    const value = material[key];
    if (value && value.isTexture) {
      value.dispose();
    }
  }
  material.dispose();
}
console.log(renderer.info.memory.textures); // texture count currently on the GPU
console.log(renderer.capabilities.maxTextureSize); // largest supported dimension

Share one texture instance across materials rather than loading per mesh. Non-power-of-two sizes are fine: WebGL 2 supports mipmaps and repeat wrapping on them.

WebGPU and TSL

import * as THREE from "three/webgpu";
import { texture, uv } from "three/tsl";

const renderer = new THREE.WebGPURenderer({ antialias: true });
await renderer.init(); // before any other use of the renderer
renderer.setPixelRatio(window.devicePixelRatio);
renderer.setSize(window.innerWidth, window.innerHeight);
document.body.appendChild(renderer.domElement);

const scene = new THREE.Scene();
scene.background = new THREE.Color(0x111111); // the WebGPU canvas starts transparent

const camera = new THREE.PerspectiveCamera(60, window.innerWidth / window.innerHeight, 0.1, 100);
camera.position.set(0, 0, 2);

const map = new THREE.TextureLoader().load("diffuse.jpg");
map.colorSpace = THREE.SRGBColorSpace;
map.anisotropy = renderer.getMaxAnisotropy(); // no capabilities object under WebGPU

const mesh = new THREE.Mesh(new THREE.PlaneGeometry(2, 2), new THREE.MeshBasicNodeMaterial());
mesh.material.colorNode = texture(map, uv().mul(0.5).add(0.25)); // custom UV transform in TSL: the center half, zoomed 2x
scene.add(mesh);

renderer.setAnimationLoop(animate);

function animate() {
  mesh.rotation.z += 0.01;
  renderer.render(scene, camera);
}

TSL sampling: texture(map) uses uv() plus the texture's own offset/repeat/rotation; an explicit node takes over the coordinates entirely. The classic slots (map, normalMap, roughnessMap, aoMap) work unchanged on node materials; colorNode is for custom sampling.

import * as THREE from "three/webgpu";
import { texture, uv } from "three/tsl";
const map = new THREE.TextureLoader().load("diffuse.jpg");
map.colorSpace = THREE.SRGBColorSpace;
mesh.material.colorNode = texture(map); // uv() plus the texture's offset/repeat/rotation
mesh.material.colorNode = texture(map, uv(1)); // explicit node: samples the "uv1" attribute

KTX2Loader.detectSupport(renderer) queries device features, so call it after await renderer.init(). For dynamic reflections use CubeRenderTarget (not WebGLCubeRenderTarget) with the same hide-update-show CubeCamera pattern.

import * as THREE from "three/webgpu";

const cubeRenderTarget = new THREE.CubeRenderTarget(256); // replaces WebGLCubeRenderTarget
const cubeCamera = new THREE.CubeCamera(0.1, 1000, cubeRenderTarget);
scene.add(cubeCamera);
mesh.material.envMap = cubeRenderTarget.texture;
mesh.visible = false; // hide so the mesh does not reflect itself
cubeCamera.position.copy(mesh.position);
cubeCamera.update(renderer, scene);
mesh.visible = true;

Common Mistakes

  • geometry.setAttribute("uv2", geometry.attributes.uv) → material.aoMap = aoTexture;: aoMap samples the attribute picked by texture.channel (default 0, uv), so it needs no extra attribute and uv2 is read only by textures whose channel is 2; only a separate baked unwrap goes in uv1, with aoTexture.channel = 1.
  • new RGBELoader() → new HDRLoader(): renamed from RGBELoader to HDRLoader in r180; the old class still runs but logs a deprecation warning.
  • import { EXRLoader } from "three/examples/jsm/loaders/EXRLoader.js" → import { EXRLoader } from "three/addons/loaders/EXRLoader.js": the browser import maps map three/addons/ to examples/jsm/ and define no three/examples/ key, so the examples path fails to resolve there.
  • texture.encoding = THREE.sRGBEncoding → texture.colorSpace = THREE.SRGBColorSpace: encoding and the sRGBEncoding constant were removed; colorSpace replaced them.
  • const texture = loader.load("color.jpg") → const texture = loader.load("color.jpg"); texture.colorSpace = THREE.SRGBColorSpace;: an LDR color image left at NoColorSpace renders too bright and washed out.
  • texture.minFilter = THREE.LinearFilter; // no mipmaps → texture.generateMipmaps = false;: mipmaps are generated whenever generateMipmaps is true, whatever the filter, so LinearFilter alone still uploads them.
  • texture.generateMipmaps = false; texture.wrapS = THREE.ClampToEdgeWrapping; // non-power-of-two size → texture.wrapS = THREE.RepeatWrapping;: WebGL 2 supports mipmaps and repeat wrapping on non-power-of-two textures, so nothing has to be disabled for them.
  • texture.anisotropy = renderer.capabilities.getMaxAnisotropy() → texture.anisotropy = renderer.getMaxAnisotropy(): WebGPURenderer has no capabilities object; this is the call under WebGPU.
  • material.envMapIntensity = 0.5 → scene.environmentIntensity = 0.5: envMapIntensity scales only the material's own envMap, not scene.environment.
  • scene.environment = pmrem.fromEquirectangular(hdr).texture → hdr.mapping = THREE.EquirectangularReflectionMapping; scene.environment = hdr;: the renderer prefilters an equirectangular environment itself, so manual PMREM of a loaded file adds nothing.
  • scene.background = hdr → hdr.mapping = THREE.EquirectangularReflectionMapping; scene.background = hdr;: without mapping, an equirectangular texture draws as a flat screen-space quad instead of a surround.
  • ktx2Loader.detectSupportAsync(renderer) → await renderer.init(); ktx2Loader.detectSupport(renderer);: the async variant was deprecated in r181, and support detection needs the initialized device.
  • ktx2Loader.setTranscoderPath("libs/basis/") → const ktx2Loader = new KTX2Loader(); ktx2Loader.detectSupport(renderer);: the transcoder loads from the package, so a path is only needed when self-hosting, pinned to three@0.186.0.
  • new THREE.WebGLMultipleRenderTargets(width, height, 2) → new THREE.WebGLRenderTarget(width, height, { count: 2 }): removed in r162; the attachments are renderTarget.textures[0..count-1].
  • new THREE.WebGLCubeRenderTarget(256) → new THREE.CubeRenderTarget(256): under WebGPURenderer use CubeRenderTarget imported from three/webgpu; WebGLCubeRenderTarget is WebGL-only.
  • loader.load(url, onLoad, onProgress, onError) → manager.onProgress = (url, loaded, total) => console.log(url, loaded, total): TextureLoader's onProgress parameter is unsupported and never fires.
  • mesh.material.map = texture; // after the mesh has rendered → mesh.material.map = texture; mesh.material.needsUpdate = true;: WebGLRenderer builds a material's shader for the maps it has at its first render, so a map added later stays invisible until the shader is rebuilt, which needsUpdate forces.
  • normalMap.colorSpace = THREE.LinearSRGBColorSpace → normalMap.colorSpace = THREE.NoColorSpace: normal, roughness, metalness, ao, alpha and displacement maps hold data, not color, so they keep the default NoColorSpace.
  • const texture = await new Promise((resolve, reject) => loader.load(url, resolve, undefined, reject)) → const texture = await loader.loadAsync(url): loadAsync is that promise, built into the Loader base class.
  • cubeCamera.update(renderer, scene) → mesh.visible = false; cubeCamera.update(renderer, scene); mesh.visible = true;: while visible, the reflective mesh is drawn into the cube map that its own envMap reads.

See Also

  • threejs-materials — material properties, blending, and physical materials beyond texture slots.
  • threejs-loaders — loading managers, glTF models, and the loader family generally.
  • threejs-shaders — custom GLSL texture sampling under WebGLRenderer.
  • threejs-lighting — environment lighting and light units.
  • threejs-asset-pipeline — encoding a model's textures to KTX2, WebP or AVIF with gltf-transform.

FAQ

What does the threejs-textures skill do?

Three.js textures: image loading, color spaces, UV channels and transforms, texture types from canvas to HDR and KTX2, render targets, texture memory. Use when working with images on materials, UV coordinates, cubemaps, render targets or texture optimization; on WebGL or with WebGPU and TSL.

How do I install threejs-textures?

Run npx skills add https://github.com/cloudai-x/threejs-skills --skill threejs-textures in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/threejs-textures/ (personal) or .claude/skills/threejs-textures/ (project).

Which AI agents can use threejs-textures?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is threejs-textures safe to install?

Security partners have reviewed it: Gen Agent Trust Hub passed; Socket passed; Snyk passed; Runlayer passed; ZeroLeaks passed. As with any skill, read the SKILL.md and bundled scripts before installing.

Who maintains threejs-textures?

It is published by cloudai-x in cloudai-x/threejs-skills. It has 12,870 recorded installs.

More from cloudai-x/threejs-skills

View all →

threejs-animation

cloudai-x/threejs-skills

Keyframe, skeletal and morph-target animation, procedural motion and clip blending, on WebGL or with WebGPU and TSL. Use when animating objects, playing glTF animations, posing bones, driving morph targets, or mixing and crossfading clips.

Frontend & Design20.7K installs

threejs-fundamentals

cloudai-x/threejs-skills

Three.js scene setup: scenes and fog, cameras, the renderer and tone mapping, the animation loop, object hierarchies, transforms, core math and cleanup, on WebGL or with WebGPU and TSL. Use when creating a scene from scratch, configuring a camera or renderer, handling resize, or arranging objects.

Frontend & Design17.7K installs

threejs-shaders

cloudai-x/threejs-skills

Custom GLSL vertex and fragment shaders: uniforms, varyings, shader chunks, extending built-in materials, instanced attributes, plus the same effects as node materials with TSL on WebGPU. Use when writing a shader, a custom material or effect, moving vertices in a shader, or porting GLSL to TSL.

Docs & Writing15.9K installs

threejs-materials

cloudai-x/threejs-skills

Choosing and configuring three.js materials - unlit, diffuse, specular, PBR, toon, point and line materials - shared properties like blending and stencil, environment response, and the WebGPU node-material and TSL equivalents. Use when styling meshes or tuning material behavior.

Frontend & Design14.9K installs

threejs-geometry

cloudai-x/threejs-skills

Creating and shaping three.js geometry: built-in shapes, 3D text, vertex buffers and attributes, merging, instancing, batching, point clouds and lines, on WebGL or with WebGPU and TSL. Use when building 3D shapes or text, custom vertex data, custom meshes, instanced meshes, or point clouds.

Media & Creative14.9K installs

threejs-interaction

cloudai-x/threejs-skills

Picking with raycasting, camera controls, transform gizmos, drag, hover and box selection for three.js, identical under WebGL and WebGPU. Use when handling mouse, touch or keyboard input, detecting clicks on 3D objects, adding camera controls, or making scene objects interactive.

Media & Creative14.6K installs

Popular media & creative skills

View category →

hyperframes

heygen-com/hyperframes

Read first to make, edit, animate or render any video with HyperFrames (HTML rendered to video), including launch films, explainers, music videos, charts, maps, vertical clips, title cards, captions and overlays. Holds the render contract, the work loop and what each kind of video needs.

Media & Creative1.2M installs

media-use

heygen-com/hyperframes

Get or make any media for a HyperFrames video - music, sound effects, images, icons, logos, voiceover, avatar and generated video, transcripts, cutouts, colour grades and looks - with HeyGen or free local tools, and mix placed audio. Also for footage that looks dark, flat or dull, or should be stylized or hidden.

Media & Creative1.1M installs

hyperframes-registry

heygen-com/hyperframes

About 450 ready-made HyperFrames blocks and components installed as editable source - transitions, overlays, titles, kinetic type, captions, charts, map overlays, 3D, camera moves. Use it before hand-building any named visual (CRT scanlines, glitch, film grain, a chart); also for hyperframes add and catalog.

Media & Creative1M installs

hyperframes-creative

heygen-com/hyperframes

Non-animation creative direction for HyperFrames videos. Use for design spec (frame.md / design.md) handling, palettes, typography, narration, beat planning, audio-reactive visuals, composition patterns, and brand / style decisions. For atomic motion patterns and scene blueprints, use hyperframes-animation.

Media & Creative916.5K installs

ai-video-generation

101-skills/superpowers

Generate AI videos with Google Veo, Seedance 2.0, HappyHorse, Wan, Grok and 40+ models via inference.sh CLI. Models: Veo 3.1, Veo 3, Seedance 2.0, HappyHorse 1.0, Wan 2.5, Grok Imagine Video, OmniHuman, Fabric, HunyuanVideo. Capabilities: text-to-video, image-to-video, reference-to-video, video editing, lipsync, avatar animation, video upscaling, foley sound. Use for: social media videos, marketing content, explainer videos, product demos, AI avatars. Triggers: video generation, ai video, text to video, image to video, veo, animate image, video from image, ai animation, video generator, generate video, t2v, i2v, ai video maker, create video with ai, runway alternative, pika alternative, sora alternative, kling alternative, seedance, happyhorse

Media & Creative881.5K installs

ai-image-generation

101-skills/superpowers

Generate AI images with GPT-Image-2, FLUX, Gemini, Grok, Seedream, Reve and 50+ models via inference.sh CLI. Models: GPT-Image-2, FLUX Dev LoRA, FLUX.2 Klein LoRA, Gemini 3 Pro Image, Grok Imagine, Seedream 4.5, Reve, ImagineArt. Capabilities: text-to-image, image-to-image, inpainting, LoRA, image editing, upscaling, text rendering. Use for: AI art, product mockups, concept art, social media graphics, marketing visuals, illustrations. Triggers: flux, image generation, ai image, text to image, stable diffusion, generate image, ai art, midjourney alternative, dall-e alternative, text2img, t2i, image generator, ai picture, create image with ai, generative ai, ai illustration, grok image, gemini image, gpt image, openai image, chatgpt image

Media & Creative880.3K installs