# Enhancing Video Physical Consistency via Role-aware Joint Training and Modality-decoupled Denoising

arXiv: https://arxiv.org/abs/2607.04653 · HTML: https://mlsanity.com/paper/2607.04653/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Video generation, one of 14 directions in Diffusion models.
- Builds on 26 of its 34 references that are in the map.
- First posted on arXiv 6 Jul 2026.

## Placed in

- Diffusion models / [Video generation](https://mlsanity.com/diffusion/video-generation/)

## Builds on (most cited first, 26 of 26)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Wan: Open and Advanced Large-Scale Video Generative Models](https://mlsanity.com/paper/2503.20314/) (2025)
- [Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets](https://mlsanity.com/paper/2311.15127/) (2023)
- [RAFT: Recurrent All-Pairs Field Transforms for Optical Flow](https://mlsanity.com/paper/2003.12039/) (2020)
- [Video Diffusion Models](https://mlsanity.com/paper/2204.03458/) (2022)
- [CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer](https://mlsanity.com/paper/2408.06072/) (2024)
- [Imagen Video: High Definition Video Generation with Diffusion Models](https://mlsanity.com/paper/2210.02303/) (2022)
- [Make-A-Video: Text-to-Video Generation without Text-Video Data](https://mlsanity.com/paper/2209.14792/) (2022)
- [HunyuanVideo: A Systematic Framework For Large Video Generative Models](https://mlsanity.com/paper/2412.03603/) (2024)
- [Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2304.08818/) (2023)
- [AnimateDiff: Animate Your Personalized Text-to-Image Diffusion Models without Specific Tuning](https://mlsanity.com/paper/2307.04725/) (2023)
- [Wonder3D: Single Image to 3D using Cross-Domain Diffusion](https://mlsanity.com/paper/2310.15008/) (2023)
- [Latte: Latent Diffusion Transformer for Video Generation](https://mlsanity.com/paper/2401.03048/) (2024)
- [Guiding a Diffusion Model with a Bad Version of Itself](https://mlsanity.com/paper/2406.02507/) (2024)
- [Open-Sora Plan: Open-Source Large Video Generation Model](https://mlsanity.com/paper/2412.00131/) (2024)
- [Autoregressive Video Generation without Vector Quantization](https://mlsanity.com/paper/2412.14169/) (2024)
- [PhysGen: Rigid-Body Physics-Grounded Image-to-Video Generation](https://mlsanity.com/paper/2409.18964/) (2024)
- [VideoPhy-2: A Challenging Action-Centric Physical Commonsense Evaluation in Video Generation](https://mlsanity.com/paper/2503.06800/) (2025)
- [VideoREPA: Learning Physics for Video Generation through Relational Alignment with Foundation Models](https://mlsanity.com/paper/2505.23656/) (2025)
- [PhyT2V: LLM-Guided Iterative Self-Refinement for Physics-Grounded Text-to-Video Generation](https://mlsanity.com/paper/2412.00596/) (2024)
- [Goku: Flow Based Video Generative Foundation Models](https://mlsanity.com/paper/2502.04896/) (2025)
- [InfinityStar: Unified Spacetime AutoRegressive Modeling for Visual Generation](https://mlsanity.com/paper/2511.04675/) (2025)
- [4DNeX: Feed-Forward 4D Generative Modeling Made Easy](https://mlsanity.com/paper/2508.13154/) (2025)
- [OmniVDiff: Omni Controllable Video Diffusion for Generation and Understanding](https://mlsanity.com/paper/2504.10825/) (2025)
- [UnityVideo: Unified Multi-Modal Multi-Task Learning for Enhancing World-Aware Video Generation](https://mlsanity.com/paper/2512.07831/) (2025)

## Built on by (most cited first, 3 of 3)

- [Scale-Separated Conditioning for Style-Encoder-Free Diffusion Stylization](https://mlsanity.com/paper/2608.19719/) (2026)
- [FaultLens: Learning Compact Behavioral Test Suites for Generated Operational Programs](https://mlsanity.com/paper/2608.26746/) (2026)
- [CrossPortrait: Cross-Species Portrait Animation via Adversarial Motion Alignment](https://mlsanity.com/paper/2610.17532/) (2026)

## Abstract (arXiv)

While modern video diffusion models excel in visual fidelity, maintaining long-range physical consistency remains a formidable challenge. Conventional pixel-reconstruction objectives mainly focus on appearance details and often fail to capture the underlying dynamics of a scene. To mitigate this, recent efforts have integrated auxiliary modalities (e.g., optical flow) to introduce physics priors via joint training with video appearance. However, these methods have three main limitations: (1) they do not distinguish the different motion patterns of different entity types; (2) joint modeling of visual and auxiliary modalities can cause capacity conflicts and weaken the pretrained visual prior; and (3) auxiliary modalities may accumulate errors during inference. To address these issues, we propose VPT, a fine-tuning framework for improving physical consistency in video diffusion models. VPT introduces a role-aware signal that groups entities into agents, controlled objects, passive objects, and background, so that different physical roles can be modeled more clearly. We further propose a modality-decoupled denoising strategy, where the visual and auxiliary channels are assigned independent noise levels. Together with a loss-weight decay strategy, this design makes auxiliary modalities serve as soft constraints rather than strong dependencies, mitigating recursive prediction errors during inference. We also introduce cross-step auto-guidance to further strengthen physical dynamics. Experiments show that VPT improves physical consistency while preserving visual quality, achieving relative gains of 39.4% in SA and 17.9% in PC on VideoPhy benchmark over Wan2.1-T2V-1.3B, and consistent improvements on VideoPhy-2 benchmark. The project page is available at https://tom-zgt.github.io/VPT.