Apple (Gene Chou, Guandao Yang, Fujun Luan, Lars Mescheder, Wei Dong, Stephan R. Richter, Vladlen Koltun) · Interactive Video Diffusion

FastFrames-L

FastFrames-L is a 14B-parameter autoregressive, two-step video diffusion model that generates 480p video, released alongside the 1.3B FastFrames-S. The README reports a 35.4 ms total latency per call for FastFrames-S on one GPU at 480p, and says an optimised LingBot 1.3B runs 16x faster than the official model with camera-conditioned control. The model card lists the licence as apple-amlr, while the README says the code uses the Apple Sample Code License and the LingBot parts follow CC BY-NC-SA 4.0.

Open the source ↗

Similar Video & Animation releases

Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.