Generative Art

Lorenz / neural / flow-field SVG art seeded deterministically from GitHub metrics.

Modules

scripts/generative.py — top-level orchestrator
scripts/art/shared.py — seed_hash(), _hex_slice(), Noise2D
scripts/banner_patterns.py — PatternType enum (shared with banner)

How it works

fetch_metrics fetches live GitHub data → metrics.json
seed_hash(metrics) produces a stable SHA-256 hex string from the metrics dict
The seed initializes numpy's default_rng — output is fully deterministic for a given metrics snapshot
Four SVG files are generated: activity (light/dark) + community (light/dark)

Entry points

from scripts.generative import generate_community_art, generate_activity_art

generate_activity_art(metrics=metrics_dict)
generate_community_art(metrics=metrics_dict)

CLI

# Fetch metrics first
uv run python -m scripts.fetch_metrics \
  --owner wyattowalsh \
  --repo wyattowalsh \
  --output /tmp/metrics.json

# Generate art from metrics
uv run python -m scripts.cli generate generative \
  --metrics-path /tmp/metrics.json

Output files

File	Description
`.github/assets/img/generative-activity.svg`	Activity-seeded attractor (light)
`.github/assets/img/generative-activity-dark.svg`	Activity-seeded attractor (dark)
`.github/assets/img/generative-community.svg`	Community-seeded attractor (light)
`.github/assets/img/generative-community-dark.svg`	Community-seeded attractor (dark)

PatternType selection

The active pattern is chosen based on metrics hash characteristics — callers don't select it manually. All five live PatternType values (LORENZ, NEURAL, FLOW, MICRO, AIZAWA) may appear.