arXiv:2605.28230v1 Announce Type: new 
Abstract: Modern video generative models produce visually impressive results, yet frequently violate basic physical principles. We propose Proprio, a training-free framework that enables a frozen video generator to assess and improve the physical plausibility of its own outputs. Inspired by proprioception, the biological sense of one's own movement, Proprio treats the model's flow residual under controlled latent perturbations as a self-scoring signal. Samples that are better explained by the generator's learned dynamics induce smaller and more stable residuals. We aggregate this signal across timesteps and perturbations, focus it on motion-relevant regions with a dynamic spatiotemporal mask, and use it for best-of-N search, gradient-based self-refinement, or both. Across text-to-video and image-to-video benchmarks, Proprio consistently improves physical plausibility, outperforming VLM-based scoring, and external world-model baselines in several settings. With TurboWan2.2, Proprio improves Physics-IQ from 32.2 to 37.5 (+16.5%) and VideoPhy2-hard physical commonsense from 45.6 to 55.0 (+20.6%). Human evaluation further shows that raters prefer Proprio-selected or refined videos for physical plausibility in roughly two-thirds of comparisons. These results suggest that frozen video generators contain actionable internal signals for evaluating and improving the physical plausibility of their own outputs.

تم تقديم بروبريو كإطار عمل بدون تدريب يعزز من المصداقية الفيزيائية لتوليد الفيديو من خلال السماح لمولد الفيديو المجمد بتقييم مخرجاته ذاتيًا. تستلهم هذه الابتكار من مفهوم الإحساس الذاتي، مما يمكّن النموذج من تقييم بقايا التدفق الخاصة به تحت اضطرابات محكومة، وبالتالي تحسين واقعية الفيديوهات المولدة.

Proprio se ha presentado como un marco sin entrenamiento que mejora la plausibilidad física de la generación de video al permitir que un generador de video congelado autoevalúe sus salidas. Esta innovación se inspira en el concepto de propriocepción, permitiendo que el modelo evalúe sus residuos de flujo bajo perturbaciones controladas, mejorando así el realismo de los videos generados.

Proprio a été introduit comme un cadre sans entraînement qui améliore la plausibilité physique de la génération vidéo en permettant à un générateur vidéo figé d'auto-évaluer ses sorties. Cette innovation s'inspire du concept de proprioception, permettant au modèle d'évaluer ses résidus de flux sous des perturbations contrôlées, améliorant ainsi le réalisme des vidéos générées.

Proprio has been introduced as a training-free framework that enhances the physical plausibility of video generation by allowing a frozen video generator to self-assess its outputs. This innovation leverages the concept of proprioception, enabling the model to evaluate its flow residuals under controlled perturbations, thereby improving the realism of generated videos.