nolangz/pixel2motion ? reverse-engineered prompt

Build me a Python based tool that takes a raster logo image like PNG, JPG, WebP, or even a screenshot, and turns it into a clean motion ready SVG, then an animated logo demo in HTML. I want the workflow to feel like a single guided pass, where it can fit the logo, check the vector quality, generate motion on semantic SVG parts, and export useful outputs like the SVG, an HTML preview, motion notes, and image evidence for QA.

Please make it easy to review the result visually, with a simple local preview and deterministic frame captures so I can compare the motion against the original logo. If it helps, look up current docs online for any browser or Playwright setup details. Keep the motion subtle, smooth, and brand friendly, not flashy. I also want the repo to include clear instructions for running the scripts and regenerating the showcase from a source logo all the way through to the final preview.