PhyT2V: LLM-Guided Iterative Self-Refinement for Physics-Grounded Text-to-Video Generation
This addresses the issue of physical inaccuracies in AI-generated videos for applications requiring realistic content, though it appears incremental as it builds on existing T2V models.
The paper tackles the problem of text-to-video generation lacking adherence to real-world physical rules by introducing PhyT2V, a data-independent technique that uses LLM-guided iterative self-refinement to improve physical realism. The result is a 2.3x improvement in adherence to physical rules and a 35% gain over existing prompt enhancers.
Text-to-video (T2V) generation has been recently enabled by transformer-based diffusion models, but current T2V models lack capabilities in adhering to the real-world common knowledge and physical rules, due to their limited understanding of physical realism and deficiency in temporal modeling. Existing solutions are either data-driven or require extra model inputs, but cannot be generalizable to out-of-distribution domains. In this paper, we present PhyT2V, a new data-independent T2V technique that expands the current T2V model's capability of video generation to out-of-distribution domains, by enabling chain-of-thought and step-back reasoning in T2V prompting. Our experiments show that PhyT2V improves existing T2V models' adherence to real-world physical rules by 2.3x, and achieves 35% improvement compared to T2V prompt enhancers. The source codes are available at: https://github.com/pittisl/PhyT2V.