Warped Diffusion: Solving Video Inverse Problems with Image Diffusion Models
This addresses temporal consistency issues in video processing for applications like restoration and enhancement, representing an incremental improvement by adapting image models to video tasks.
The paper tackles the problem of flickering and temporal inconsistency when using image diffusion models for video inverse tasks by viewing frames as continuous functions and videos as sequences of warping transformations, enabling the use of image-trained models for video inpainting and 8x super-resolution with improved performance over existing methods.
Using image models naively for solving inverse video problems often suffers from flickering, texture-sticking, and temporal inconsistency in generated videos. To tackle these problems, in this paper, we view frames as continuous functions in the 2D space, and videos as a sequence of continuous warping transformations between different frames. This perspective allows us to train function space diffusion models only on images and utilize them to solve temporally correlated inverse problems. The function space diffusion models need to be equivariant with respect to the underlying spatial transformations. To ensure temporal consistency, we introduce a simple post-hoc test-time guidance towards (self)-equivariant solutions. Our method allows us to deploy state-of-the-art latent diffusion models such as Stable Diffusion XL to solve video inverse problems. We demonstrate the effectiveness of our method for video inpainting and $8\times$ video super-resolution, outperforming existing techniques based on noise transformations. We provide generated video results: https://giannisdaras.github.io/warped_diffusion.github.io/.