PICABench: How Far Are We from Physically Realistic Image Editing?

TL;DR

This paper introduces PICABench, a comprehensive benchmark for evaluating physical realism in image editing, highlighting current limitations and proposing solutions to improve physical consistency in generated images.

Contribution

The paper presents PICABench for systematic evaluation of physical realism in image editing and introduces PICAEval, a new evaluation protocol using vision-language models and human annotations.

Findings

Physical realism in image editing remains a significant challenge.

Most mainstream models do not adequately capture physical effects.

Learning physics from videos can improve physical realism.

Abstract

Image editing has achieved remarkable progress recently. Modern editing models could already follow complex instructions to manipulate the original content. However, beyond completing the editing instructions, the accompanying physical effects are the key to the generation realism. For example, removing an object should also remove its shadow, reflections, and interactions with nearby objects. Unfortunately, existing models and benchmarks mainly focus on instruction completion but overlook these physical effects. So, at this moment, how far are we from physically realistic image editing? To answer this, we introduce PICABench, which systematically evaluates physical realism across eight sub-dimension (spanning optics, mechanics, and state transitions) for most of the common editing operations (add, remove, attribute change, etc.). We further propose the PICAEval, a reliable evaluation protocol that uses VLM-as-a-judge with per-case, region-level human annotations and questions. Beyond benchmarking, we also explore effective solutions by learning physics from videos and construct a training dataset PICA-100K. After evaluating most of the mainstream models, we observe that physical realism remains a challenging problem with large rooms to explore. We hope that our benchmark and proposed solutions can serve as a foundation for future work moving from naive content editing toward physically consistent realism.