Canvas Generative logo

Canvas Generative

Community
AThevon
canvas-generative

Algorithmic and generative art with Canvas 2D - particles, flow fields, noise, fractals, L-systems.

Overview

PublisherAThevon
Repositorygenjutsu
Skill namecanvas-generative
Stars
687
Forks
52
Bundled files
1
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by AThevon on GitHub. Read the source before you install it.

Installation

Install the Canvas Generative AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/AThevon/genjutsu/tree/main/skills/_jutsu/canvas-generative
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/AThevon/genjutsu.git /tmp/genjutsu
mkdir -p .claude/skills
cp -r /tmp/genjutsu/skills/_jutsu/canvas-generative .claude/skills/canvas-generative
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Canvas Generative in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Canvas Generative on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Canvas Generative is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Version-sensitive. Every API name, SDK gate and browser-support claim below was verified on 2026-09-08 against primary sources. What against, and when, is in _jutsu/VERSIONS.md. If that date is old, re-verify before acting on a version number.

Canvas Generative

Algorithmic and generative art with Canvas 2D. Concise rules here. Deep-dive and reference implementations in references/.


Canvas 2D Setup

DPR-Aware Sizing

Every canvas must be sharp on Retina/HiDPI displays. Set the buffer size to the physical pixel size, scale down with CSS.

js
function setupCanvas(canvas, width, height) {
  const dpr = window.devicePixelRatio || 1;
  canvas.width = width * dpr;
  canvas.height = height * dpr;
  canvas.style.width = `${width}px`;
  canvas.style.height = `${height}px`;
  const ctx = canvas.getContext('2d');
  ctx.scale(dpr, dpr);
  return ctx;
}

Resize Handler

js
function handleResize(canvas, ctx, draw) {
  const ro = new ResizeObserver(([entry]) => {
    const { width, height } = entry.contentRect;
    const dpr = window.devicePixelRatio || 1;
    canvas.width = width * dpr;
    canvas.height = height * dpr;
    ctx.scale(dpr, dpr);
    draw(); // re-render after resize
  });
  ro.observe(canvas.parentElement);
  return () => ro.disconnect();
}

Animation Loop (RAF)

js
let animId;
let prevTime = 0;

function loop(time) {
  const dt = Math.min((time - prevTime) / 1000, 0.1); // cap delta to avoid spiral of death
  prevTime = time;

  update(dt);
  render(ctx);

  animId = requestAnimationFrame(loop);
}

// Start
animId = requestAnimationFrame(loop);
// Stop
cancelAnimationFrame(animId);

Noise

TypeCharacteristicsBest For
PerlinSmooth, grid-aligned bias, cheaperTerrain, clouds, gentle organic textures
SimplexNo grid artifacts, better gradients, slightly costlierFlow fields, organic motion, seamless tiling
Worley (Cellular)Distance-to-nearest-point, cell-likeVoronoi patterns, caustics, cracks, cell textures

Usage rules:

  • Always scale input coordinates (divide by a noiseScale factor) -- raw pixel coords produce visual noise
  • Use octaves (fractal Brownian motion) for detail: sum multiple noise calls at increasing frequency and decreasing amplitude
  • Seed your noise for reproducibility
js
// fBm pattern
function fbm(x, y, octaves = 4, lacunarity = 2, gain = 0.5) {
  let value = 0, amplitude = 1, frequency = 1, maxAmp = 0;
  for (let i = 0; i < octaves; i++) {
    value += amplitude * noise2D(x * frequency, y * frequency);
    maxAmp += amplitude;
    amplitude *= gain;
    frequency *= lacunarity;
  }
  return value / maxAmp; // normalize to [-1, 1]
}

Particle Systems

Pool Pattern (No GC Pressure)

Pre-allocate a fixed array. Never new or splice at runtime.

js
const POOL_SIZE = 10000;
const particles = new Array(POOL_SIZE);
let aliveCount = 0;

// Init pool
for (let i = 0; i < POOL_SIZE; i++) {
  particles[i] = { x: 0, y: 0, vx: 0, vy: 0, life: 0, maxLife: 0, active: false };
}

function spawn(x, y) {
  if (aliveCount >= POOL_SIZE) return;
  const p = particles[aliveCount++];
  p.x = x; p.y = y;
  p.vx = (Math.random() - 0.5) * 2;
  p.vy = (Math.random() - 0.5) * 2;
  p.life = 0; p.maxLife = 60 + Math.random() * 60;
  p.active = true;
}

function update() {
  for (let i = aliveCount - 1; i >= 0; i--) {
    const p = particles[i];
    p.x += p.vx; p.y += p.vy;
    p.life++;
    if (p.life >= p.maxLife) {
      // Swap with last alive, shrink pool
      particles[i] = particles[--aliveCount];
      particles[aliveCount] = p;
      p.active = false;
    }
  }
}

Flow Fields

Grid of angle vectors that steer particles. The classic generative recipe.

  1. Build grid: Divide canvas into cells, compute an angle per cell (from noise)
  2. Lookup: Particle position maps to grid cell, retrieve angle
  3. Steer: Apply angle as velocity, accumulate over frames
js
const cols = Math.ceil(width / cellSize);
const rows = Math.ceil(height / cellSize);
const field = new Float32Array(cols * rows);

// Fill with noise-based angles
for (let y = 0; y < rows; y++) {
  for (let x = 0; x < cols; x++) {
    field[y * cols + x] = noise2D(x * 0.05, y * 0.05) * Math.PI * 2;
  }
}

// Particle follow
function followField(p) {
  const col = Math.floor(p.x / cellSize);
  const row = Math.floor(p.y / cellSize);
  if (col >= 0 && col < cols && row >= 0 && row < rows) {
    const angle = field[row * cols + col];
    p.vx += Math.cos(angle) * force;
    p.vy += Math.sin(angle) * force;
  }
  // Damping to prevent runaway velocity
  p.vx *= 0.98;
  p.vy *= 0.98;
}

Fractals & L-Systems

An L-system encodes recursive structure as string rewriting + turtle graphics.

ComponentRole
AxiomStarting string (e.g. "F")
RulesProduction rules (e.g. "F" -> "F[+F]F[-F]F")
AngleTurtle turn angle per +/-
IterationsHow many times to apply rules
js
function lsystem(axiom, rules, iterations) {
  let current = axiom;
  for (let i = 0; i < iterations; i++) {
    current = current.split('').map(c => rules[c] || c).join('');
  }
  return current;
}

function drawLSystem(ctx, commands, len, angle) {
  const stack = [];
  for (const c of commands) {
    switch (c) {
      case 'F': ctx.lineTo(ctx._x += Math.cos(ctx._a) * len, ctx._y += Math.sin(ctx._a) * len); break;
      case '+': ctx._a += angle; break;
      case '-': ctx._a -= angle; break;
      case '[': stack.push({ x: ctx._x, y: ctx._y, a: ctx._a }); break;
      case ']': { const s = stack.pop(); ctx._x = s.x; ctx._y = s.y; ctx._a = s.a; ctx.moveTo(s.x, s.y); } break;
    }
  }
}

Double Buffer Pattern

Render to an offscreen canvas, then blit to the visible one. Eliminates flicker and enables trail effects.

js
const offscreen = document.createElement('canvas');
offscreen.width = canvas.width;
offscreen.height = canvas.height;
const offCtx = offscreen.getContext('2d');

function render() {
  // Draw to offscreen
  offCtx.fillStyle = 'rgba(0, 0, 0, 0.05)'; // trail fade
  offCtx.fillRect(0, 0, offscreen.width, offscreen.height);
  drawParticles(offCtx);

  // Blit to screen
  ctx.drawImage(offscreen, 0, 0);
}

Do Not

1. Never clearRect every frame for trail effects

Clearing destroys the trail. Use a semi-transparent fill instead.

js
// BAD -- kills trails
ctx.clearRect(0, 0, w, h);

// GOOD -- fades previous frame
ctx.fillStyle = 'rgba(0, 0, 0, 0.02)';
ctx.fillRect(0, 0, w, h);

2. Never getImageData in the animation loop

getImageData reads back from GPU -- extremely slow. Avoid unless absolutely required (e.g. one-time color sampling).

js
// BAD -- 60fps GPU readback
function loop() {
  const data = ctx.getImageData(0, 0, w, h); // blocks rendering pipeline
  processPixels(data);
  requestAnimationFrame(loop);
}

// GOOD -- sample once, cache
const colorMap = ctx.getImageData(0, 0, w, h);
function getColor(x, y) {
  const i = (y * w + x) * 4;
  return [colorMap.data[i], colorMap.data[i+1], colorMap.data[i+2]];
}

3. Always respect DPR for sharpness

A canvas without DPR scaling looks blurry on Retina displays. See setup section above.

js
// BAD
canvas.width = 800;
canvas.height = 600;

// GOOD
const dpr = window.devicePixelRatio || 1;
canvas.width = 800 * dpr;
canvas.height = 600 * dpr;
canvas.style.width = '800px';
canvas.style.height = '600px';
ctx.scale(dpr, dpr);

4. Never allocate in the hot loop

No new, no object spread, no array creation inside update() or render(). Pre-allocate everything.

js
// BAD
function update() {
  particles.forEach(p => {
    const force = { x: Math.cos(a), y: Math.sin(a) }; // new object every frame per particle
    p.vx += force.x;
  });
}

// GOOD
let fx = 0, fy = 0; // reuse
function update() {
  for (let i = 0; i < aliveCount; i++) {
    fx = Math.cos(a); fy = Math.sin(a);
    particles[i].vx += fx;
  }
}

Quick Reference: Loading Sub-skills

NeedLoad
Noise implementations, particle pool, flow field, L-system, attractorsreferences/algorithms.md
Timing and easing for animated transitions../motion-principles/SKILL.md
3D generative (shaders, GPU particles)../threejs-r3f/SKILL.md

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Canvas Generative AI skill do?

Algorithmic and generative art with Canvas 2D - particles, flow fields, noise, fractals, L-systems.

Why use Canvas Generative on TypingMind?

Because you install it once and use it with any model. Canvas Generative is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Canvas Generative in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/AThevon/genjutsu/tree/main/skills/_jutsu/canvas-generative. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Canvas Generative?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Canvas Generative?

As many as you like. As long as a model supports skills, you can use Canvas Generative with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Canvas Generative AI skill free?

It is published on GitHub by AThevon. Check the repository for licensing terms. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇