# MultiRef: Controllable Image Generation with Multiple Visual References

arXiv: https://arxiv.org/abs/2508.06905 · HTML: https://mlsanity.com/paper/2508.06905/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Builds on 34 of its 56 references that are in the map.
- First posted on arXiv 9 Aug 2025.

## Builds on (most cited first, 30 of 34)

- [Scaling Rectified Flow Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2403.03206/) (2024)
- [LAION-5B: An open large-scale dataset for training next generation image-text models](https://mlsanity.com/paper/2210.08402/) (2022)
- [DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation](https://mlsanity.com/paper/2208.12242/) (2022)
- [GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment](https://mlsanity.com/paper/2310.11513/) (2023)
- [Grounded SAM: Assembling Open-World Models for Diverse Visual Tasks](https://mlsanity.com/paper/2401.14159/) (2024)
- [MagicBrush: A Manually Annotated Dataset for Instruction-Guided Image Editing](https://mlsanity.com/paper/2306.10012/) (2023)
- [Evaluating Text-to-Visual Generation with Image-to-Text Generation](https://mlsanity.com/paper/2404.01291/) (2024)
- [Generative Multimodal Models are In-Context Learners](https://mlsanity.com/paper/2312.13286/) (2023)
- [TIFA: Accurate and Interpretable Text-to-Image Faithfulness Evaluation with Question Answering](https://mlsanity.com/paper/2303.11897/) (2023)
- [Scaling Synthetic Data Creation with 1,000,000,000 Personas](https://mlsanity.com/paper/2406.20094/) (2024)
- [AnyDoor: Zero-shot Object-level Image Customization](https://mlsanity.com/paper/2307.09481/) (2023)
- [MLLM-as-a-Judge: Assessing Multimodal LLM-as-a-Judge with Vision-Language Benchmark](https://mlsanity.com/paper/2402.04788/) (2024)
- [OminiControl: Minimal and Universal Control for Diffusion Transformer](https://mlsanity.com/paper/2411.15098/) (2024)
- [Emu Edit: Precise Image Editing via Recognition and Generation Tasks](https://mlsanity.com/paper/2311.10089/) (2023)
- [Diffusion Self-Guidance for Controllable Image Generation](https://mlsanity.com/paper/2306.00986/) (2023)
- [VITON-HD: High-Resolution Virtual Try-On via Misalignment-Aware Normalization](https://mlsanity.com/paper/2103.16874/) (2021)
- [LayoutDiffusion: Controllable Diffusion Model for Layout-to-image Generation](https://mlsanity.com/paper/2303.17189/) (2023)
- [Subject-driven Text-to-Image Generation via Apprenticeship Learning](https://mlsanity.com/paper/2304.00186/) (2023)
- [Versatile Diffusion: Text, Images and Variations All in One Diffusion Model](https://mlsanity.com/paper/2211.08332/) (2022)
- [Large-scale Classification of Fine-Art Paintings: Learning The Right Metric on The Right Feature](https://mlsanity.com/paper/1505.00855/) (2015)
- [ControlNet++: Improving Conditional Controls with Efficient Consistency Feedback](https://mlsanity.com/paper/2404.07987/) (2024)
- [GenArtist: Multimodal LLM as an Agent for Unified Image Generation and Editing](https://mlsanity.com/paper/2407.05600/) (2024)
- [Can We Generate Images with CoT? Let's Verify and Reinforce Image Generation Step by Step](https://mlsanity.com/paper/2501.13926/) (2025)
- [UniReal: Universal Image Generation and Editing via Learning Real-world Dynamics](https://mlsanity.com/paper/2412.07774/) (2024)
- [FreeControl: Training-Free Spatial Control of Any Text-to-Image Diffusion Model with Any Condition](https://mlsanity.com/paper/2312.07536/) (2023)
- [Learning to generate line drawings that convey geometry and semantics](https://mlsanity.com/paper/2203.12691/) (2022)
- [Human-Art: A Versatile Human-Centric Dataset Bridging Natural and Artificial Scenes](https://mlsanity.com/paper/2303.02760/) (2023)
- [Instruct-Imagen: Image Generation with Multi-modal Instruction](https://mlsanity.com/paper/2401.01952/) (2024)
- [ACE: All-round Creator and Editor Following Instructions via Diffusion Transformer](https://mlsanity.com/paper/2410.00086/) (2024)
- [StyleBooth: Image Style Editing with Multimodal Instruction](https://mlsanity.com/paper/2404.12154/) (2024)

## Built on by (most cited first, 9 of 9)

- [Emu3.5: Native Multimodal Models are World Learners](https://mlsanity.com/paper/2510.26583/) (2025)
- [FastFit: Accelerating Multi-Reference Virtual Try-On via Cacheable Diffusion Models](https://mlsanity.com/paper/2508.20586/) (2025)
- [MultiBanana: A Challenging Benchmark for Multi-Reference Text-to-Image Generation](https://mlsanity.com/paper/2511.22989/) (2025)
- [MICON-Bench: Benchmarking and Enhancing Multi-Image Context Image Generation in Unified Multimodal Models](https://mlsanity.com/paper/2602.19497/) (2026)
- [Garments2Look: A Multi-Reference Dataset for High-Fidelity Outfit-Level Virtual Try-On with Clothing and Accessories](https://mlsanity.com/paper/2603.14153/) (2026)
- [MultiBind: A Benchmark for Attribute Misbinding in Multi-Subject Generation](https://mlsanity.com/paper/2603.21937/) (2026)
- [StructGen: Disambiguating Multi-Reference Image Generation via Structured Context Modeling](https://mlsanity.com/paper/2607.15619/) (2026)
- [VIF-Bench: Evaluating Visual Instruction Following in Multi-Reference Image Generation](https://mlsanity.com/paper/2609.37709/) (2026)
- [It's Always 10:10: Reference Images Break a Bias That Prompts Only Dent](https://mlsanity.com/paper/2610.11320/) (2026)

## Abstract (arXiv)

Visual designers naturally draw inspiration from multiple visual references, combining diverse elements and aesthetic principles to create artwork. However, current image generative frameworks predominantly rely on single-source inputs -- either text prompts or individual reference images. In this paper, we focus on the task of controllable image generation using multiple visual references. We introduce MultiRef-bench, a rigorous evaluation framework comprising 990 synthetic and 1,000 real-world samples that require incorporating visual content from multiple reference images. The synthetic samples are synthetically generated through our data engine RefBlend, with 10 reference types and 33 reference combinations. Based on RefBlend, we further construct a dataset MultiRef containing 38k high-quality images to facilitate further research. Our experiments across three interleaved image-text models (i.e., OmniGen, ACE, and Show-o) and six agentic frameworks (e.g., ChatDiT and LLM + SD) reveal that even state-of-the-art systems struggle with multi-reference conditioning, with the best model OmniGen achieving only 66.6% in synthetic samples and 79.0% in real-world cases on average compared to the golden answer. These findings provide valuable directions for developing more flexible and human-like creative tools that can effectively integrate multiple sources of visual inspiration. The dataset is publicly available at: https://multiref.github.io/.