IC-Custom: Diverse Image Customization via In-Context Learning

IC-Custom Team

Introduction

IC-Custom introduces a novel approach to image customization through in-context learning, enabling diverse and flexible image generation scenarios.

Position-aware

Precise control over object placement with reference images and target backgrounds

Position-free

Generate new images while preserving reference object identity

IC-Custom Method Overview

Figure 1: IC-Custom methodology overview

Position-aware Image Customization

Hover over the images to see the generated results - Our model seamlessly integrates reference content into target scenes

Position-free Image Customization

🖱️ Hover over any image to see the generated result and text prompt - Our model creates images based on text prompts while maintaining reference identity

Reference Image Generated Image

"wandering through a lush jungle..."

Click to enlarge
Reference Image Generated Image

"Strolling in the pond..."

Click to enlarge
Reference Image Generated Image

"on the blanket..."

Click to enlarge
Reference Image Generated Image

"...rests on a rustic wooden table..."

Click to enlarge
Reference Image Generated Image

"The yellow alarm clock is perched on a snowy mountain peak at sunrise..."

Click to enlarge
Reference Image Generated Image

"A cat is lying on some ancient books..."

Click to enlarge

BibTeX

@article {li2025ic,
title ={IC-Custom: Diverse Image Customization via In-Context Learning},
author ={Li, Yaowei and Li, Xiaoyu and Zhang, Zhaoyang and Bian, Yuxuan and Liu, Gan and Li, Xinyuan and Xu, Jiale and Hu, Wenbo and Liu, Yating and Li, Lingen and others},
journal ={arXiv preprint arXiv:2507.01926},
year ={2025}
}
Copied!