lucidrains/phenaki-pytorch
A PyTorch implementation of a text-to-video generation model capable of producing videos up to 2 minutes long.

Velocity · 7d
+0.6
★ / day
Trend
→steady
star history
Implementation of Phenaki Video in PyTorch, a text-guided video generation system that uses Mask GIT architecture with transformer encoders and token critic techniques for improved generation quality. The model encodes visual data into discrete tokens using a C-ViViT architecture and generates videos from textual prompts.