Photorealistic Video Generation with Diffusion Models

278
citations
#28
in ECCV 2024
of 2387 papers
9
Top Authors
5
Data Points

Abstract

We present W.A.L.T, a transformer-based approach for photorealistic video generation via diffusion modeling. Our approach has two key design decisions. First, we use a causal encoder to jointly compress images and videos within a unified latent space, enabling training and generation across modalities. Second, for memory and training efficiency, we use a window attention architecture tailored for joint spatial and spatiotemporal generative modeling. Taken together these design decisions enable us to achieve state-of-the-art performance on established video (UCF-101 and Kinetics-600) and image (ImageNet) generation benchmarks without using classifier free guidance. Finally, we also train a cascade of three models for the task of text-to-video generation consisting of a base latent video diffusion model, and two video super-resolution diffusion models to generate videos of $512 \times 896$ resolution at $8$ frames per second.

Citation History

Jan 26, 2026
264
Feb 1, 2026
270+6
Feb 6, 2026
275+5
Feb 13, 2026
278+3
Feb 13, 2026
278