Videoshop: Localized Semantic Video Editing with Noise-Extrapolated Diffusion Inversion
This addresses the problem of fine-grained video editing for users who need precise control over object locations and appearance, offering a more flexible alternative to text-based methods.
The paper tackles the problem of localized semantic video editing by introducing Videoshop, a training-free algorithm that allows users to modify the first frame with any editing software and automatically propagates those changes to remaining frames with semantic, spatial, and temporal consistency. The method produces higher quality edits against 6 baselines on 2 editing benchmarks using 10 evaluation metrics.
We introduce Videoshop, a training-free video editing algorithm for localized semantic edits. Videoshop allows users to use any editing software, including Photoshop and generative inpainting, to modify the first frame; it automatically propagates those changes, with semantic, spatial, and temporally consistent motion, to the remaining frames. Unlike existing methods that enable edits only through imprecise textual instructions, Videoshop allows users to add or remove objects, semantically change objects, insert stock photos into videos, etc. with fine-grained control over locations and appearance. We achieve this through image-based video editing by inverting latents with noise extrapolation, from which we generate videos conditioned on the edited image. Videoshop produces higher quality edits against 6 baselines on 2 editing benchmarks using 10 evaluation metrics.