Squeezing Capacity from Multimodal Large Language Models for Subject-driven Generation

• URL: http://arxiv.org/abs/2605.26111v1
• Date: Mon, 25 May 2026 17:59:35 GMT
• Title: Squeezing Capacity from Multimodal Large Language Models for Subject-driven Generation
• Abstract summary: Existing approaches often encode text and reference images separately.<n>Recent frameworks that connect multimodal models and diffusion models improve instruction following, but largely overlook identity preservation.<n>We condition diffusion models on Multimodal Large Language Models that jointly encode text and reference images, and augment it with VAE-based identity conditioning.<n>Our approach harmonizes multimodal understanding with identity preservation, mitigates copy-paste issues, and achieves superior performance regarding human preference on subject-driven image generation.
• Score: 22.419513267677278
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: Subject-driven image generation aims to synthesize new images that preserve the identity of the given subject while following textual instructions. Existing approaches often encode text and reference images separately. This limits cross-modal reasoning abilities and causes copy-paste artifacts. Recent frameworks that connect multimodal models and diffusion models improve instruction following, but largely overlook identity preservation. To address these limitations, we condition diffusion models on Multimodal Large Language Models (MLLMs) that jointly encode text and reference images, and augment it with VAE-based identity conditioning. A novel Dual Layer Aggregation (DLA) module is designed to aggregate multi-level MLLM features for optimal conditioning, and a multi-stage denoising strategy is applied to progressively balance the semantic information from MLLM and fine-detail identity from VAE during inference. Extensive experiments demonstrate that our approach harmonizes multimodal understanding with identity preservation, mitigates copy-paste issues, and achieves superior performance regarding human preference on subject-driven image generation. Our project website is available at https://zsh2000.github.io/squeeze-mllm-subject-gen/.

Related papers

• Multimodal Large Language Models for Multi-Subject In-Context Image Generation [56.20395856287325]
  Music is the first MLLM specifically designed for textbfMUlti-textbfSubject textbfIn-textbfContext image generation.<n>We enhance the model's understanding of multi-subject semantic relationships through a vision chain-of-thought mechanism.<n>By incorporating complex subject images during training, we improve the model's capacity for chained reasoning.
  arXiv  Detail & Related papers  (2026-04-08T15:37:42Z)
• Growing Visual Generative Capacity for Pre-Trained MLLMs [60.826355079902505]
  Bridge is a pure autoregressive unified MLLM that augments pre-trained visual understanding models with generative ability.<n>We propose a semantic-to-pixel discrete representation that integrates compact semantic tokens with fine-grained pixel tokens.
  arXiv  Detail & Related papers  (2025-10-02T00:40:02Z)
• Query-Kontext: An Unified Multimodal Model for Image Generation and Editing [53.765351127477224]
  Unified Multimodal Models (UMMs) have demonstrated remarkable performance in text-to-image generation (T2I) and editing (TI2I)<n>We introduce Query-Kontext, a novel approach that bridges the VLM and diffusion model via a multimodal kontext'' composed of semantic cues and coarse-grained image conditions encoded from multimodal inputs.<n> Experiments show that our approach matches strong unified baselines and even outperforms task-specific state-of-the-art methods in several cases.
  arXiv  Detail & Related papers  (2025-09-30T17:59:46Z)
• Lavida-O: Elastic Large Masked Diffusion Models for Unified Multimodal Understanding and Generation [63.50827603618498]
  We propose Lavida-O, a unified Masked Diffusion Model (MDM) for multimodal understanding and generation.<n>Lavida-O presents a single framework that enables image-level understanding, object grounding, image editing, and high-resolution text-to-image synthesis.<n>Lavida-O achieves state-of-the-art performance on a wide range of benchmarks including RefCOCO object grounding, GenEval text-to-image generation, and ImgEdit image editing.
  arXiv  Detail & Related papers  (2025-09-23T17:05:46Z)
• Generative Multimodal Pretraining with Discrete Diffusion Timestep Tokens [66.02261367232256]
  Multimodal Large Language Models (MLLMs) aim to unify visual comprehension and generation.<n>Existing approaches rely on spatial visual tokens, where image patches are encoded and arranged according to a spatial order.<n>In this paper, we build a proper visual language by reconstructing diffusion timesteps to learn discrete visual tokens.
  arXiv  Detail & Related papers  (2025-04-20T16:14:28Z)
• MaVEn: An Effective Multi-granularity Hybrid Visual Encoding Framework for Multimodal Large Language Model [49.931663904599205]
  MaVEn is an innovative framework designed to enhance the capabilities of Multimodal Large Language Models (MLLMs) in multi-image reasoning. We show that MaVEn significantly enhances MLLMs' understanding in complex multi-image scenarios, while also improving performance in single-image contexts.
  arXiv  Detail & Related papers  (2024-08-22T11:57:16Z)
• TIE: Revolutionizing Text-based Image Editing for Complex-Prompt Following and High-Fidelity Editing [23.51498634405422]
  We present an innovative image editing framework that employs the robust Chain-of-Thought reasoning and localizing capabilities of multimodal large language models. Our model exhibits an enhanced ability to understand complex prompts and generate corresponding images, while maintaining high fidelity and consistency in images before and after generation.
  arXiv  Detail & Related papers  (2024-05-27T03:50:37Z)
• Many-to-many Image Generation with Auto-regressive Diffusion Models [59.5041405824704]
  This paper introduces a domain-general framework for many-to-many image generation, capable of producing interrelated image series from a given set of images. We present MIS, a novel large-scale multi-image dataset, containing 12M synthetic multi-image samples, each with 25 interconnected images. We learn M2M, an autoregressive model for many-to-many generation, where each image is modeled within a diffusion framework.
  arXiv  Detail & Related papers  (2024-04-03T23:20:40Z)

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.