Flexible Diffusion Modeling of Long Videos
This work addresses video generation for applications like autonomous driving simulation, though it appears incremental as it builds on existing diffusion models with architectural and scheduling optimizations.
The authors tackled the problem of generating long-duration videos by developing a diffusion-based framework that allows flexible conditioning on arbitrary frame subsets, resulting in temporally coherent videos over 25 minutes long with improved performance on multiple datasets.
We present a framework for video modeling based on denoising diffusion probabilistic models that produces long-duration video completions in a variety of realistic environments. We introduce a generative model that can at test-time sample any arbitrary subset of video frames conditioned on any other subset and present an architecture adapted for this purpose. Doing so allows us to efficiently compare and optimize a variety of schedules for the order in which frames in a long video are sampled and use selective sparse and long-range conditioning on previously sampled frames. We demonstrate improved video modeling over prior work on a number of datasets and sample temporally coherent videos over 25 minutes in length. We additionally release a new video modeling dataset and semantically meaningful metrics based on videos generated in the CARLA autonomous driving simulator.