What problem does it solve? Building performant generative art in the browser requires solving recurring problems: blurry canvases on Retina displays, garbage-collection stutter from particle allocation, and slow GPU readbacks. This Skill provides copy-paste-ready patterns and complete reference implementations for Canvas 2D generative art. ## Core Features & Use Cases - Canvas Setup & Performance Rules: DPR-aware sizing, resize handling, delta-capped animation loops, double buffering, and explicit anti-patterns (no clearRect for trails, no getImageData in loops, no allocation in hot loops). - Complete Algorithm Implementations: Seeded Simplex noise (2D/3D), zero-allocation particle pool with struct-of-arrays layout, noise-driven flow fields, L-system renderer with presets (plant, Koch, Sierpinski, dragon, tree), and a Lorenz attractor with RK4 integration. - Use Case: A developer wants an animated flow-field background for a landing page. Load the Skill, copy the FlowField class from references/algorithms.md, seed the noise, and render with trail fading for an organic line pattern at 60fps. ## Quick Start Use the canvas-generative skill to build a flow field animation with 2000 particles on a full-screen canvas.