STR-Match: Matching SpatioTemporal Relevance Score for Training-Free Video Editing
This addresses video editing challenges for AI and creative applications, but it is incremental as it builds on existing diffusion models with a new optimization method.
The paper tackled the problem of temporal inconsistency and limited domain transformation in text-guided video editing by proposing STR-Match, a training-free algorithm that uses a novel spatiotemporal relevance score for latent optimization, resulting in improved visual quality and consistency in experiments.
Previous text-guided video editing methods often suffer from temporal inconsistency, motion distortion, and-most notably-limited domain transformation. We attribute these limitations to insufficient modeling of spatiotemporal pixel relevance during the editing process. To address this, we propose STR-Match, a training-free video editing algorithm that produces visually appealing and spatiotemporally coherent videos through latent optimization guided by our novel STR score. The score captures spatiotemporal pixel relevance across adjacent frames by leveraging 2D spatial attention and 1D temporal modules in text-to-video (T2V) diffusion models, without the overhead of computationally expensive 3D attention mechanisms. Integrated into a latent optimization framework with a latent mask, STR-Match generates temporally consistent and visually faithful videos, maintaining strong performance even under significant domain transformations while preserving key visual attributes of the source. Extensive experiments demonstrate that STR-Match consistently outperforms existing methods in both visual quality and spatiotemporal consistency.