GenHowTo: Learning to Generate Actions and State Transformations from Instructional Videos

We address the task of generating temporally consistent and physically plausible images of actions and object state transformations. Given an input image and a text prompt describing the targeted transformation, our generated im- ages preserve the environment and transform objects in the initial image. Our contributions are threefold. First, we leverage a large body of instructional videos and automati- cally mine a dataset of triplets of consecutive frames cor- responding to initial object states, actions, and resulting object transformations. Second, equipped with this data, we develop and train a conditioned diffusion model dubbed GenHowTo. Third, we evaluate GenHowTo on a variety of objects and actions and show superior performance com- pared to existing methods. In particular, we introduce a quantitative evaluation where GenHowTo achieves 88% and 74% on seen and unseen interaction categories, respec- tively, outperforming prior work by a large margin.