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.

通过扩散建模，我们提出了一种基于转换器的 W.A.L.T 方法，用于逼真视频生成。我们采用了两个关键设计决策：一是使用因果编码器在统一的潜在空间中联合压缩图像和视频，实现跨模态的训练和生成；二是为了记忆和训练效率，我们使用了一个针对联合空间和时空生成建模的窗口注意力架构。这些设计决策使我们能够在已知的视频 (UCF-101 和 Kinetics-600) 和图像 (ImageNet) 生成基准上达到最先进的性能，而不使用分类器引导。最后，我们还训练了三个模型的串联，用于文本到视频生成任务，包括基本的潜在视频扩散模型和两个视频超分辨率扩散模型，以每秒 8 帧生成 512 x 896 分辨率的视频。