Whole-Body Conditioned Egocentric Video Prediction

Researchers at UC Berkeley introduce PEVA, a model for whole-body conditioned egocentric video prediction using a diffusion transformer, enabling realistic future frame generation from first-person video inputs.

One continuously updated timeline instead of dozens of separate articles. New developments are appended as the story evolves.

1. AnnouncementJul 1, 2025, 09:00 AM 84%

   UC Berkeley researchers unveil PEVA, a diffusion-transformer model for whole-body conditioned egocentric video prediction

   Researchers at UC Berkeley introduce PEVA, a model for whole-body conditioned egocentric video prediction using a diffusion transformer, enabling realistic future frame generation from first-person video inputs.

   Read the full story →