MultiCrafter: High-Fidelity Multi-Subject Generation via Spatially Disentangled Attention and Identity-Aware Reinforcement Learning

• URL: http://arxiv.org/abs/2509.21953v1
• Date: Fri, 26 Sep 2025 06:41:43 GMT
• Title: MultiCrafter: High-Fidelity Multi-Subject Generation via Spatially Disentangled Attention and Identity-Aware Reinforcement Learning
• Authors: Tao Wu, Yibo Jiang, Yehao Lu, Zhizhong Wang, Zeyi Huang, Zequn Qin, Xi Li,
• Abstract summary: Multi-subject image generation aims to synthesize user-provided subjects in a single image.<n>Existing methods are limited by their reliance on simple reconstruction-based objectives.<n>We propose MultiCrafter, a framework that ensures high-fidelity, preference-aligned generation.
• Score: 28.841076643572933
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Multi-subject image generation aims to synthesize user-provided subjects in a single image while preserving subject fidelity, ensuring prompt consistency, and aligning with human aesthetic preferences. However, existing methods, particularly those built on the In-Context-Learning paradigm, are limited by their reliance on simple reconstruction-based objectives, leading to both severe attribute leakage that compromises subject fidelity and failing to align with nuanced human preferences. To address this, we propose MultiCrafter, a framework that ensures high-fidelity, preference-aligned generation. First, we find that the root cause of attribute leakage is a significant entanglement of attention between different subjects during the generation process. Therefore, we introduce explicit positional supervision to explicitly separate attention regions for each subject, effectively mitigating attribute leakage. To enable the model to accurately plan the attention region of different subjects in diverse scenarios, we employ a Mixture-of-Experts architecture to enhance the model's capacity, allowing different experts to focus on different scenarios. Finally, we design a novel online reinforcement learning framework to align the model with human preferences, featuring a scoring mechanism to accurately assess multi-subject fidelity and a more stable training strategy tailored for the MoE architecture. Experiments validate that our framework significantly improves subject fidelity while aligning with human preferences better.

Related papers

• From Sparse Decisions to Dense Reasoning: A Multi-attribute Trajectory Paradigm for Multimodal Moderation [59.27094165576015]
  We propose a novel learning paradigm (UniMod) that transitions from sparse decision-making to dense reasoning traces.<n>By constructing structured trajectories encompassing evidence grounding, modality assessment, risk mapping, policy decision, and response generation, we reformulate monolithic decision tasks into a multi-dimensional boundary learning process.<n>We introduce specialized optimization strategies to decouple task-specific parameters and rebalance training dynamics, effectively resolving interference between diverse objectives in multi-task learning.
  arXiv  Detail & Related papers  (2026-01-28T09:29:40Z)
• PSR: Scaling Multi-Subject Personalized Image Generation with Pairwise Subject-Consistency Rewards [86.1965460124838]
  We propose a scalable multi-subject data generation pipeline.<n>We first enable single-subject personalization models to acquire knowledge of multi-image and multi-subject scenarios.<n>To enhance both subject consistency and text controllability, we design a set of Pairwise Subject-Consistency Rewards.
  arXiv  Detail & Related papers  (2025-12-01T03:25:49Z)
• OneCAT: Decoder-Only Auto-Regressive Model for Unified Understanding and Generation [91.45421429922506]
  OneCAT is a unified multimodal model that seamlessly integrates understanding, generation, and editing.<n>Our framework eliminates the need for external components such as Vision Transformers (ViT) or vision tokenizer during inference.
  arXiv  Detail & Related papers  (2025-09-03T17:29:50Z)
• MOSAIC: Multi-Subject Personalized Generation via Correspondence-Aware Alignment and Disentanglement [13.100620283631557]
  We present MOSAIC, a representation-centric framework that rethinks multi-subject generation.<n>Our key insight is that multi-subject generation requires precise semantic alignment at the representation level.<n>We propose the semantic correspondence attention loss to enforce precise point-to-point semantic alignment.
  arXiv  Detail & Related papers  (2025-09-02T05:40:07Z)
• FocusDPO: Dynamic Preference Optimization for Multi-Subject Personalized Image Generation via Adaptive Focus [10.615833390806486]
  Multi-subject personalized image generation aims to synthesize customized images containing multiple specified subjects without requiring test-time optimization.<n>We present FocusDPO, a framework that adaptively identifies focus regions based on dynamic semantic correspondence and supervision image complexity.
  arXiv  Detail & Related papers  (2025-09-01T07:06:36Z)
• Mitigating Attention Hacking in Preference-Based Reward Modeling via Interaction Distillation [62.14692332209628]
  "Interaction Distillation" is a novel training framework for more adequate preference modeling through attention-level optimization.<n>It provides more stable and generalizable reward signals compared to state-of-the-art RM optimization methods.
  arXiv  Detail & Related papers  (2025-08-04T17:06:23Z)
• Multi-Level Aware Preference Learning: Enhancing RLHF for Complex Multi-Instruction Tasks [81.44256822500257]
  RLHF has emerged as a predominant approach for aligning artificial intelligence systems with human preferences.<n> RLHF exhibits insufficient compliance capabilities when confronted with complex multi-instruction tasks.<n>We propose a novel Multi-level Aware Preference Learning (MAPL) framework, capable of enhancing multi-instruction capabilities.
  arXiv  Detail & Related papers  (2025-05-19T08:33:11Z)
• Nested Attention: Semantic-aware Attention Values for Concept Personalization [78.90196530697897]
  We introduce Nested Attention, a novel mechanism that injects a rich and expressive image representation into the model's existing cross-attention layers.<n>Our key idea is to generate query-dependent subject values, derived from nested attention layers that learn to select relevant subject features for each region in the generated image.
  arXiv  Detail & Related papers  (2025-01-02T18:52:11Z)
• The Superalignment of Superhuman Intelligence with Large Language Models [63.96120398355404]
  We discuss the concept of superalignment from the learning perspective to answer this question.<n>We highlight some key research problems in superalignment, namely, weak-to-strong generalization, scalable oversight, and evaluation.<n>We present a conceptual framework for superalignment, which consists of three modules: an attacker which generates adversary queries trying to expose the weaknesses of a learner model; a learner which will refine itself by learning from scalable feedbacks generated by a critic model along with minimal human experts; and a critic which generates critics or explanations for a given query-response pair, with a target of improving the learner by criticizing.
  arXiv  Detail & Related papers  (2024-12-15T10:34:06Z)
• Unlocking the Potential of Text-to-Image Diffusion with PAC-Bayesian Theory [33.78620829249978]
  Text-to-image (T2I) diffusion models have revolutionized generative modeling by producing high-fidelity, diverse, and visually realistic images. Recent attention-based methods have improved object inclusion and linguistic binding, but still face challenges such as attribute misbinding. We propose a Bayesian approach that designs custom priors over attention distributions to enforce desirable properties. Our approach treats the attention mechanism as an interpretable component, enabling fine-grained control and improved attribute-object alignment.
  arXiv  Detail & Related papers  (2024-11-25T10:57:48Z)
• CustomContrast: A Multilevel Contrastive Perspective For Subject-Driven Text-to-Image Customization [27.114395240088562]
  We argue an ideal subject representation can be achieved by a cross-differential perspective, i.e., decoupling subject intrinsic attributes from irrelevant attributes via contrastive learning.<n>Specifically, we propose CustomContrast, a novel framework, which includes a Multilevel Contrastive Learning paradigm and a Multimodal Feature Injection (MFI)<n> Extensive experiments show the effectiveness of CustomContrast in subject similarity and text controllability.
  arXiv  Detail & Related papers  (2024-09-09T13:39:47Z)

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.