Eliminated.
Ship ready-made frames, not generated imagery. The model copies a folder and writes a few lines — it never spends image tokens drawing motion that already exists.
Slicer ready
Slice a short video into a drop-in WebP scroll-animation package a frontier model wires into any repo — with no token burn generating imagery. Ship ready-made frames, not pixels a model has to draw.
Generating hero imagery from scratch is slow and non-deterministic. Hand the model a finished package instead.
Ship ready-made frames, not generated imagery. The model copies a folder and writes a few lines — it never spends image tokens drawing motion that already exists.
WebP frame sequences are lazy-loadable and CLS-free — a fraction of the weight of shipping video or a heavy JS animation library. Engineered to help your vitals, not hurt them.
Copy the folder to public/, open index.html, and it animates now — no server, no build step, zero external requests.
Zero-clone with uvx, or clone and launch the full dark studio.
$ uvx --from <release-wheel-url> seo-video-slicer
Grab the wheel URL from the latest Release.
Needs ffmpeg and node on your PATH.
$ git clone https://github.com/ehukaimedia/seo-video-slicer \
&& cd seo-video-slicer \
&& ./start.command
Sets up the venv, builds the UI, and serves the slicer on
localhost:8000.