PhyEditBench: A Real-World Multi-Stage Benchmark for Physics-Aware Image Editing

• URL: http://arxiv.org/abs/2606.26551v2
• Date: Fri, 26 Jun 2026 07:00:33 GMT
• Title: PhyEditBench: A Real-World Multi-Stage Benchmark for Physics-Aware Image Editing
• Abstract summary: We introduce PhyEditBench, a benchmark designed to assess the physical understanding of editing models.<n>It comprises 238 high-quality, high-resolution, real-world instances meticulously extracted from videos.<n>Our empirical analysis of current SOTA editing methods exposes substantial limitations in their physics-based reasoning.
• Score: 14.070238906847488
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: While instruction-based image editing, enabled by multi-modal generative models, has advanced significantly, existing benchmarks lack a comprehensive evaluation of physics-based reasoning, a critical capability for handling real-world scenarios. To address this, we introduce PhyEditBench, a benchmark designed to assess the physical understanding of editing models. Guided by a hierarchical taxonomy, we establish 4 primary classes and 12 subclasses. It comprises 238 high-quality, high-resolution, real-world instances meticulously extracted from videos to capture authentic physical dynamics, alongside 35 synthetic Anti-Physics instances. Our empirical analysis of current SOTA editing methods exposes substantial limitations in their physics-based reasoning. We further propose a training-free baseline named PhyWorld that uses test-time scaling and a latent reduction strategy. PhyWorld outperforms comparable models and suggests that the video generation process can effectively serve as a reasoning mechanism for image editing. The project page is available at https://github.com/Previsior/PhyEditBench.

Related papers

• PhyCo: Learning Controllable Physical Priors for Generative Motion [55.59209981836171]
  We present PhyCo, a framework that introduces continuous, interpretable, and physically grounded control into video generation.<n>Our approach integrates three key components: (i) a large-scale dataset of over 100K photorealistic simulation videos where friction, restitution, deformation, and force are systematically varied across diverse scenarios; (ii) physics-supervised fine-tuning of a pretrained diffusion model using a ControlNet conditioned on pixel-aligned physical property maps; and (iii) VLM-guided reward optimization, where a fine-tuned vision-language model evaluates generated videos with targeted physics queries and provides differentiable feedback.
  arXiv  Detail & Related papers  (2026-04-30T17:53:03Z)
• PhysRVG: Physics-Aware Unified Reinforcement Learning for Video Generative Models [100.65199317765608]
  Physical principles are fundamental to realistic visual simulation, but remain a significant oversight in transformer-based video generation.<n>We introduce a physics-aware reinforcement learning paradigm for video generation models that enforces physical collision rules directly in high-dimensional spaces.<n>We extend this paradigm to a unified framework, termed Mimicry-Discovery Cycle (MDcycle), which allows substantial fine-tuning.
  arXiv  Detail & Related papers  (2026-01-16T08:40:10Z)
• PhysWorld: From Real Videos to World Models of Deformable Objects via Physics-Aware Demonstration Synthesis [52.905353023326306]
  We propose PhysWorld, a framework that synthesizes physically plausible and diverse demonstrations to learn efficient world models.<n>Experiments show that PhysWorld has competitive performance while enabling inference speeds 47 times faster than the recent state-of-the-art method, i.e., PhysTwin.
  arXiv  Detail & Related papers  (2025-10-24T13:25:39Z)
• PICABench: How Far Are We from Physically Realistic Image Editing? [71.82009431774311]
  We introduce PICABench, which systematically evaluates physical realism across eight sub-dimensions.<n>We propose the PICAEval, a reliable evaluation protocol that uses VLM-as-a-judge with per-case, region-level human annotations.<n>We also explore effective solutions by learning physics from videos and construct a training dataset PICA-100K.
  arXiv  Detail & Related papers  (2025-10-20T15:53:57Z)
• LikePhys: Evaluating Intuitive Physics Understanding in Video Diffusion Models via Likelihood Preference [57.086932851733145]
  We introduce LikePhys, a training-free method that evaluates intuitive physics in video diffusion models.<n>We benchmark intuitive physics understanding in current video diffusion models.<n> Empirical results show that, despite current models struggling with complex and chaotic dynamics, there is a clear trend of improvement in physics understanding as model capacity and inference settings scale.
  arXiv  Detail & Related papers  (2025-10-13T15:19:07Z)
• "PhyWorldBench": A Comprehensive Evaluation of Physical Realism in Text-to-Video Models [38.14213802594432]
  PhyWorldBench is a benchmark designed to evaluate video generation models based on their adherence to the laws of physics.<n>We introduce a novel ""Anti-Physics" category, where prompts intentionally violate real-world physics.<n>We evaluate 12 state-of-the-art text-to-video generation models, including five open-source and five proprietary models.
  arXiv  Detail & Related papers  (2025-07-17T17:54:09Z)
• Towards World Simulator: Crafting Physical Commonsense-Based Benchmark for Video Generation [51.750634349748736]
  Text-to-video (T2V) models have made significant strides in visualizing complex prompts. However, the capacity of these models to accurately represent intuitive physics remains largely unexplored. We introduce PhyGenBench to evaluate physical commonsense correctness in T2V generation.
  arXiv  Detail & Related papers  (2024-10-07T17:56:04Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.