Generative Art Starter Kit: 10 Python Generators for Pen Plotters
10 Python generators for pen plotters. 30 pre-plotted example SVGs, a 5,700-word guide with measured stroke-travel data and 30 named recipes, plus an AI scoring rubric distilled from 105+ autoresearch experiments.
Most generative art tutorials give you a script that makes pretty patterns on screen. They don't tell you how to make those patterns work on a pen plotter, where stroke ordering matters, ink coverage needs to be in a specific range, and tiny details get lost in the mechanical precision of the pen.
This kit comes from 105+ experiments running an autoresearch loop across 75+ generators, scoring thousands of outputs to find what actually plots well.
What you get:
10 working generators (all ship as standalone Python files, all share a common plotter_utils.py module):
- Flow Field, Perlin noise particle traces, stdlib only
- L-System Tree, 5 presets (oak, willow, fern, bush, sierpinski)
- Voronoi Cells, Lloyd's relaxation (requires numpy + scipy)
- Spiral Patterns, Archimedean, logarithmic, hypotrochoid
- Reaction-Diffusion, Gray-Scott (spots, worms, coral, maze, zebra presets, requires numpy)
- Subdivision Grid, Vera Molnar-inspired recursive cuts
- Subject-Driven Hatching, NO background hatching, bring your own image or use the built-in mask
- Space-Filling Curves, Hilbert, Heighway dragon, Gosper flowsnake
- Topographic Contours, multi-octave noise with ridged-multifractal mode (requires numpy)
- Cellular Automata, 1-D elementary rules (30, 90, 110, 184, etc.)
30 pre-plotted example SVGs (three seeds per generator) in content/examples/, each with its seed, parameters and plot time recorded in a MANIFEST.md alongside PNG previews for every file.
5,700-word plotter-focused guide (content/guide.md) with:
- Measured stroke-optimization before/after numbers (17-19x air-travel reduction on real output)
- Four mermaid diagrams (scoring pipeline, stroke-optimization comparison, L-system expansion, generator-selection tree)
- Failure-mode gallery (feed rate, wet ink, scoring rejects, re-plot cost)
- 30 named parameter recipes (three per generator, copy-paste CLI ready)
- AxiDraw + iDraw setup walkthroughs with pen pressure, feed rate, paper pairing
- An FAQ and a "common gotchas" section from nine months of autoresearch
Production features:
- SVG output optimized for plotters (single-stroke, no fills)
- Shared nearest-neighbor stroke ordering with measured 17-19x air-travel reduction
- Built-in plot-time estimator (+/- 15% vs real wall clock)
- Configurable paper sizes (A5, A4, A3, Letter) with margin-aware layout
- Metadata embedded in every SVG (seed, full parameter dump, path count, draw travel, air travel, plot time, pen)
- AI scoring script with Gemini and OpenAI providers, plus a real scoring log from a 10-piece gallery run (average 7.2 / 10)
Scoring rubric from 105+ experiments:
- 6 dimensions: composition, complexity, coherence, ink coverage, plottability, uniqueness
- Quality gates: 8+ = print immediately, 6-7.9 = worth plotting, < 4 = rework
- LLM scoring prompt included, run it on your outputs
Who this is for:
- Developers with pen plotters (AxiDraw, iDraw, etc.) who want algorithmic art
- Creative technologists exploring the intersection of code and physical art
- Anyone curious about generative art with a focus on real output, not just screen renders
Who this is NOT for:
- If you want Processing/p5.js sketches, this is Python-only
- If you don't care about physical output, you won't use half the features
All generators are Python 3.10+. Core generators have zero dependencies. Advanced generators need numpy/scipy.