PhysLoop, 물리 인식 LLM으로 비디오 생성 모델을 자가개선
PhysLoop: Harnessing LLMs' Physics Awareness for Self-Improving Video Generation Models
- 매체
- Hugging Face
- 게시
- 중요도
- 4/5
- 화제 지수
- 18
한국어 요약
비디오 생성 모델이 물리적 역학과 인과 관계 일관성에서 어려움을 겪는 문제를 해결하기 위해, LLM의 물리 지식을 활용해 비디오를 생성·선별·보정하는 PhysLoop 프레임워크를 제안했다. 제어 가능한 학습 데이터 구성과 추론 시간 비디오 복구를 지원하며, 100K개의 다양한 실내·외 장면 비디오와 평가 데이터셋 두 가지를 공개했다.
원문 발췌 · English
Video generation models produce visually realistic videos but still struggle with physical dynamics and causal consistency. Learning from generated videos requires fine-grained physical supervision, yet reliable quality judgments, correction pairs, and controlled variations are difficult to obtain. Large language models (LLMs) possess physical knowledge, but struggle to reliably guide video generation models in producing their training videos. We present PhysLoop, a unified harness designed to support physically grounded self-improvement in video generation models. The harness guides and constrains LLMs throughout video generation, selection, and correction, distilling their physical knowledge into physically consistent videos. The framework supports controllable training-data construction and inference-time video repair. By identifying and correcting failures, PhysLoop marks a key milestone toward recursive self-improvement (RSI) in video generation models with limited human intervention. We release 100K videos spanning diverse, realistic indoor and outdoor scenes, along with generation tools and two evaluation datasets: PhysLoop-Gen and PhysLoop-Repair. We evaluate controllable generation and localized repair on these datasets, respectively, and assess physical behavior on the external Physics-IQ Verified benchmark.