AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

MultiCrafter: High-Fidelity Multi-Subject Generation via Disentangled Attention and Identity-Aware Preference Alignment

Tao Wu, Yibo Jiang, Yehao Lu, Zhizhong Wang, Zeyi Huang, Zequn Qin, Xi Li

CVPR 2026 · Image generation › Text-to-image diffusion · approach: Preference optimization for diffusion

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add identity-disentangled attention regularization L_attn (single LoRA) (ablation)UNO baseline on FLUX · Multi-human benchmarkFace-Sim: 0.1474 -> 0.4983Improved
add L_attn with single LoRA (ablation)UNO baseline on FLUX · Multi-human benchmarkCLIP-T: 0.2645 -> 0.2637Hurt
add L_attn with single LoRA (ablation)UNO baseline on FLUX · Multi-human benchmarkAES (HPSv2): 0.2954 -> 0.2653Hurt
add MoE-LoRA on top of L_attn (ablation)MultiCrafter pre-training · Multi-human benchmarkFace-Sim: 0.4983 -> 0.5154Improved
add MoE-LoRA on top of L_attn (ablation)MultiCrafter pre-training · Multi-human benchmarkCLIP-T: 0.2637 -> 0.2674Improved
add IPPO post-training (online RL with Multi-ID reward) (ablation)MultiCrafter · Multi-human benchmarkAES (HPSv2): 0.2661 -> 0.2915Improved
add IPPO post-training (ablation)MultiCrafter · Multi-human benchmarkCLIP-T: 0.2674 -> 0.2753Improved
add IPPO post-training (ablation)MultiCrafter · Multi-human benchmarkFace-Sim: 0.5154 -> 0.5284Improved
retrain UNO on the authors' video-derived multi-human data (ablation)UNO · Multi-human benchmarkAES (HPSv2): 0.2954 -> 0.2656Hurt
retrain UNO on the authors' video-derived multi-human data (ablation)UNO · Multi-human benchmarkFace-Sim: 0.1474 -> 0.2879Improved
MultiCrafter vs next-best method on multi-human fidelityMultiCrafter · Multi-human benchmarkFace-Sim: 0.4117 -> 0.5284Improved
MultiCrafter vs OmniGen2 on aestheticsMultiCrafter · Multi-human benchmarkAES (HPSv2): 0.3103 -> 0.2915Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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