# UniCustom: Unified Visual Conditioning for Multi-Reference Image Generation

Source: [UniCustom: Unified Visual Conditioning for Multi-Reference Image Generation](<https://www.sota2.com/research/paper/6a0405d7dbc836b87d827828>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-15T05:00:12.430Z

Authors: Yiyan Xu, Qiulin Wang, Wenjie Wang, Yunyao Mao, Xintao Wang, Pengfei Wan, Kun Gai, Fuli Feng

Published: 2026-05-12T13:10:05.000Z

## Abstract

Multi-reference image generation aims to synthesize images from textual instructions while faithfully preserving subject identities from multiple reference images. Existing VLM-enhanced diffusion models commonly rely on decoupled visual conditioning: semantic ViT features are processed by the VLM for instruction understanding, whereas appearance-rich VAE features are injected later into the diffusion backbone. Despite its intuitive design, this separation makes it difficult for the model to associate each semantically grounded subject with visual details from the correct reference image. As a result, the model may recognize which subject is being referred to, but fail to preserve its identity and fine-grained appearance, leading to attribute leakage and cross-reference confusion in complex multi-reference settings. To address this issue, we propose UniCustom, a unified visual conditioning framework that fuses ViT and VAE features before VLM encoding. This early fusion exposes the VLM to both semantic cues and appearance-rich details, enabling its hidden states to jointly encode the referred subject and corresponding visual appearance with only a lightweight linear fusion layer. To learn such unified representations, we adopt a two-stage training strategy: reconstruction-oriented pretraining that preserves reference-specific appearance details in the fused hidden states, followed by supervised finetuning on single- and multi-reference generation tasks. We further introduce a slot-wise binding regularization that encourages each image slot to preserve low-level details of its corresponding reference, thereby reducing cross-reference entanglement. Experiments on two multi-reference generation benchmarks demonstrate that UniCustom consistently improves subject consistency, instruction following, and compositional fidelity over strong baselines.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Multi-Reference Image Editing on MICo-Bench](<https://www.sota2.com/research/sota/multi-reference-image-editing-on-mico-bench.md>): Multi-Reference Image Editing; MICo-Bench; Object Score: 54.3.

- [Subject-Consistent Image Generation on OmniContext](<https://www.sota2.com/research/sota/subject-consistent-image-generation-on-omnicontext.md>): Subject-consistent image generation; OmniContext; Fidelity (Single, Character): 8.06.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.12088>)

[PDF](<https://arxiv.org/pdf/2605.12088.pdf>)
