← ClaudeAtlas

generative-art-svg-enginelisted

Deterministic generative-art engine on the Python stdlib that turns a seed into layered, palette-driven SVG compositions - LCG pseudo-noise, seeded palettes, shape primitives (circles, paths, grids, waves), tint/weight rules - and exports browser-ready self-contained SVG. Reproducible per seed with zero design tools and zero MCP.
RMAzrim/opencode-artes · ★ 2 · AI & Automation · score 76
Install: claude install-skill RMAzrim/opencode-artes
# Generative Art SVG Engine Turn any integer seed into original SVG artwork using nothing but the Python standard library. A deterministic Linear Congruential Generator (LCG) drives every random decision — palette, composition, density, jitter — so a seed always reproduces the same image, perfect for NFTs-of-the-future, CI-generated cover art, favicons, and README headers. Output is a standalone `<svg>` you can open in any browser. ## 1. System Architecture & Prerequisites - **Runtime**: CPython 3.9+ (`math`, `json`, `argparse`, `random` replaced by a manual LCG). No Pillow, no svgwrite, no MCP. - **Engine core**: - `LCG` — Schrage-style 64-bit LCG with period 2^32, `next_float()` ∈ [0,1); seeding from any int or string hash. - `palette()` — pick `bg` (dark) + 4 accent colors from a baked 8-color wheel, by seed rotation with complementary/triad rules. - `compose()` — layers: grid-of-circles with radius jitter, random walk "waves" (smooth path), low-poly rings, and per-layer opacity/weight. - `render_svg()` — emits full SVG with `<defs>`, gradients, viewBox `0 0 800 800`. - **Reproducibility guarantee**: for any seed S, `render_svg(S)` is byte-identical across processes (LCG is fully determinist, no `random` module, no time/entropy). ## 2. Input/Output Data Contracts ``` python svg_art.py --seed 1234 --size 800 --out art.svg python svg_art.py --seed random --count 6 --out-dir gallery/ python svg_art.py --digest --seed 42 ``` - `--seed` int (or `rand