
Ep#21 TesserAct: Learning 4D Embodied World Models
RoboPapers
00:00
Advancements in Video Diffusion Models
This chapter explores the progress in video diffusion models, focusing on the enhanced capabilities of CogVideoX in producing high-quality 3D outputs. It covers the integration of normals for depth maps, the significance of surface geometry, and the tools used for annotating 3D videos to improve scene reconstruction.
Transcript
Play full episode