Lilian Weng • 4/12/2024

Diffusion Models for Video Generation

This technical article delves into using diffusion models for video generation, a more complex task than image synthesis due to temporal consistency and data requirements. It reviews approaches for training video models from scratch, detailing parameterization, sampling basics, and the v-prediction method to avoid issues like color shift.

0 comments

#Machine Learning #generative ai #Deep Learning