Space-Time Diffusion Features for Zero-Shot Text-Driven Motion Transfer
This addresses the problem of motion transfer for video synthesis across diverse objects, which is incremental as it builds on prior methods limited to similar categories.
The paper tackles the problem of text-driven motion transfer across drastically different object categories, such as translating a jumping dog into a dolphin, by introducing a space-time feature loss from a pre-trained diffusion model to preserve motion while complying with text prompts, achieving results in this challenging setting.
We present a new method for text-driven motion transfer - synthesizing a video that complies with an input text prompt describing the target objects and scene while maintaining an input video's motion and scene layout. Prior methods are confined to transferring motion across two subjects within the same or closely related object categories and are applicable for limited domains (e.g., humans). In this work, we consider a significantly more challenging setting in which the target and source objects differ drastically in shape and fine-grained motion characteristics (e.g., translating a jumping dog into a dolphin). To this end, we leverage a pre-trained and fixed text-to-video diffusion model, which provides us with generative and motion priors. The pillar of our method is a new space-time feature loss derived directly from the model. This loss guides the generation process to preserve the overall motion of the input video while complying with the target object in terms of shape and fine-grained motion traits.