# DeCoRL: Decomposed Consistency Reinforcement Learning for Multi-Image Composition

DevFeed: [DeCoRL: Decomposed Consistency Reinforcement Learning for Multi-Image Composition](<https://devfeed.tech/articles/decorl-decomposed-consistency-reinforcement-learning-for-multi-image-composition-108168.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.39265>)

Author: Zhiqiang Wu, Shuang Sun, Jiale Zhang, Jing Li, Yibo Chen, Xu Tang, Yao Hu

Published: 2026-10-30T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [image editing](<https://devfeed.tech/topics/image-editing.md>), [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [post-training](<https://devfeed.tech/topics/post-training.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [Image](<https://devfeed.tech/topics/image.md>)

Tags: [benchmark](<https://devfeed.tech/tags/benchmark.md>), [composition](<https://devfeed.tech/tags/composition.md>), [consistency](<https://devfeed.tech/tags/consistency.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [framework](<https://devfeed.tech/tags/framework.md>), [image](<https://devfeed.tech/tags/image.md>), [image-editing](<https://devfeed.tech/tags/image-editing.md>), [learning](<https://devfeed.tech/tags/learning.md>), [post-training](<https://devfeed.tech/tags/post-training.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [reward](<https://devfeed.tech/tags/reward.md>), [rl](<https://devfeed.tech/tags/rl.md>), [vlm](<https://devfeed.tech/tags/vlm.md>)

## AI overview

DeCoRL aims to improve consistency in multi-image composition, where performance can degrade as reference images accumulate. It identifies instruction-relevant objects across references, locates their corresponding regions in the generated image, and uses a vision-language reward model trained on single-object consistency annotations to produce object-level feedback. That feedback is aggregated into a multi-object reward for reinforcement-learning post-training. The authors introduce DeCo-Bench for compositions using 2-6 reference images and report consistent consistency improvements on it and MICo-Bench. The supplied text does not provide quantitative results or experimental details.

## Source excerpt

arXiv:2610.39265v1 Announce Type: cross Abstract: Multi-image composition (MICo) is a critical yet challenging capability for large-scale image editing models, where consistency degrades rapidly as the number of reference images increases. While supervised fine-tuning (SFT) can equip models with basic MICo ability, it provides only implicit supervision on the composed image and often fails to learn reliable object-level correspondence across multiple references. We propose DeCoRL, a reinforcement-learning (RL) framework that improves MICo consistency by decomposing holistic MICo consistency into object-level consistency. DeCoRL first establishes instruction-relevant object correspondences across reference images and localizes the corresponding regions in the generated image. It then uses a VLM-based reward model trained on single-object consistency annotations to produce fine-grained, object-wise feedback, which is aggregated into a multi-object consistency reward to guide RL post-training. We further introduce DeCo-Bench, a benchmark covering compositions from 2 to 6 reference images. Experiments on DeCo-Bench and the public MICo-Bench demonstrate consistent improvements in MICo consistency.