Relic Video World Model

Compressed historical latents with camera poses in KV cache (4× compression), extended teacher training (20-second sequences), and replayed back-propagation (block-wise differentiation) enabling real-time interactive video generation with long-range spatial consistency.

adu2021 Updated 7 repo stars

File contents

adu2021/skillxiv commit a3cfbc0272

Frequently asked questions

npx get-skill add adu2021/relic-video-world-model