AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

DeCoRL: Decomposed Consistency Reinforcement Learning for Multi-Image Composition

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

NeurIPS 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Multimodal LLM benchmarks · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes DeCoRL, a decomposed consistency reinforcement learning approach for multi-image composition (MICo) in image editing models, motivated by SFT's weak object-level correspondence supervision.

Problem

Multi-image composition in large-scale image editing models, where consistency degrades as the number of reference images grows; SFT gives only implicit supervision on the composed image and fails to learn reliable object-level correspondence across references.

Approach

Decomposes consistency into explicit object-level correspondence signals and optimizes them with reinforcement learning rather than relying on implicit image-level supervision from SFT. Abstract is truncated, so the exact reward design is not stated.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add DeCoRL object-level RL after SFT (ablation)Qwen-Image-Edit2511 + SFT · DeCo-Bench (overall)consistency score: 7.05 -> 8.16Improved
add DeCoRL after SFT, 6 reference images (ablation)Qwen-Image-Edit2511 + SFT · DeCo-Bench (6 images)consistency score: 6.15 -> 7.61Improved
add multi-image SFT before RL (ablation)Qwen-Image-Edit2511 · DeCo-Bench (overall)consistency score: 6.22 -> 7.05Improved
add multi-image SFT, 2 reference images (ablation)Qwen-Image-Edit2511 · DeCo-Bench (2 images)consistency score: 8.41 -> 8.34No clear effect
add DeCoRL after SFT (ablation)Qwen-Image-Edit2511 + SFT · MICo-Bench (overall)overall score: 22.90 -> 37.58 -> 54.03Improved
add DeCoRL after SFT (ablation)Qwen-Image-Edit2511 + SFT · MICo-Bench (Naturalness)Naturalness: 6.54 -> 8.01Improved
add DeCoRL after SFT (ablation)Qwen-Image-Edit2511 + SFT · MICo-Bench (Prompt Following)PF: 8.76 -> 8.69No clear effect
add DeCoRL after SFT (ablation)Qwen-Image-Edit2511 + SFT · MICo-Bench (Subject Resemblance)SR: 7.01 -> 7.64Improved
use a CoT-trained single-object reward VLM instead of the vanilla VLM (ablation)DeCoRL · DeCo-Bench (overall)consistency score: 7.21 -> 8.16Improved
use a vanilla VLM as the reward in place of no RL (ablation)Qwen-Image-Edit2511 + SFT · DeCo-Bench (4 images)consistency score: 6.80 -> 6.42Hurt
fine-tune the reward VLM with chain-of-thought supervision (ablation)Qwen3-VL-32B reward model · EditReward-Benchhuman-label agreement accuracy: 0.354 -> 0.721Improved
fine-tune the reward VLM with chain-of-thought supervision (ablation)DeCoRL training dynamics · multi-object consistency score during RLreward level: below 0.5 (vanilla) vs above 0.6 (trained)Improved

Experimental setup

Baselines: supervised fine-tuning

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy