v0.1.0

First public release of splat-slim — training-free 3D Gaussian Splatting compression as a CLI + library.

Highlights

• ~74% mean size reduction across three scenes at modest PSNR cost — no retraining.
• Four-stage pipeline: adaptive opacity pruning → outlier cleaning → SH-degree reduction → mixed-precision quantization.
• Framework-agnostic: operates on the exported .ply, self-decodable sidecar metadata.
• ruff + pytest CI, MIT licensed.

See the README for the quickstart and per-scene results table.