DiffusionBrowser: Interactive Diffusion Previews via Multi-Branch Decoders
This work addresses the need for faster and more transparent video generation for users, offering incremental improvements in interactivity and speed.
The paper tackled the problem of video diffusion models being slow and opaque during generation by proposing DiffusionBrowser, a model-agnostic decoder framework that enables interactive previews at any denoising step, achieving over 4× real-time speed (less than 1 second for a 4-second video) and providing new control capabilities.
Video diffusion models have revolutionized generative video synthesis, but they are imprecise, slow, and can be opaque during generation -- keeping users in the dark for a prolonged period. In this work, we propose DiffusionBrowser, a model-agnostic, lightweight decoder framework that allows users to interactively generate previews at any point (timestep or transformer block) during the denoising process. Our model can generate multi-modal preview representations that include RGB and scene intrinsics at more than 4$\times$ real-time speed (less than 1 second for a 4-second video) that convey consistent appearance and motion to the final video. With the trained decoder, we show that it is possible to interactively guide the generation at intermediate noise steps via stochasticity reinjection and modal steering, unlocking a new control capability. Moreover, we systematically probe the model using the learned decoders, revealing how scene, object, and other details are composed and assembled during the otherwise black-box denoising process.