Solarwm Data Pipeline

Turn raw video into a camera-annotated, captioned, quality-scored world-model training corpus with the SolarWM data engine. Use when adding a raw source, running annotation workers, calibrating quality thresholds, verifying clip geometry, or building a training mixture. Triggers — "annotate my videos with camera poses", "add a new source", "keep-rate is zero", "are these poses right", "build a training set". Excludes model training, validation, inference and latent preencoding, which belong to the training side of the project.

Enderfga Updated

File contents

enderfga/solarwm-data commit e782b4c32f

Frequently asked questions

npx get-skill add enderfga/solarwm-data-pipeline