PureCC: Pure Learning for Text-to-Image Concept Customization

2026-03-08

AI Summary

This work addresses the challenge that existing text-to-image concept customization methods often degrade the pretrained model’s original capabilities when incorporating new concepts. To mitigate this interference, the authors propose a decoupled learning framework that leverages a frozen feature extractor to provide clean concept representations as implicit guidance, while a trainable flow-based model learns to predict the original conditional distribution. The approach introduces a dual-branch training architecture and an adaptive guidance scale λ* to enable high-fidelity multi-concept customization without compromising the model’s general generative capacity. Experimental results demonstrate that the method achieves state-of-the-art performance in preserving the base model’s behavior, effectively balancing customization fidelity with model generalization.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Multimodal Learning
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Existing concept customization methods have achieved remarkable outcomes in high-fidelity and multi-concept customization. However, they often neglect the influence on the original model's behavior and capabilities when learning new personalized concepts. To address this issue, we propose PureCC. PureCC introduces a novel decoupled learning objective for concept customization, which combines the implicit guidance of the target concept with the original conditional prediction. This separated form enables PureCC to substantially focus on the original model during training. Moreover, based on this objective, PureCC designs a dual-branch training pipeline that includes a frozen extractor providing purified target concept representations as implicit guidance and a trainable flow model producing the original conditional prediction, jointly achieving pure learning for personalized concepts. Furthermore, PureCC introduces a novel adaptive guidance scale $\lambda^\star$ to dynamically adjust the guidance strength of the target concept, balancing customization fidelity and model preservation. Extensive experiments show that PureCC achieves state-of-the-art performance in preserving the original behavior and capabilities while enabling high-fidelity concept customization. The code is available at https://github.com/lzc-sg/PureCC.

Problem

• concept customization
• model preservation
• text-to-image generation
• personalized concepts
• original model behavior

Innovation

• decoupled learning
• dual-branch training
• adaptive guidance scale
• concept customization
• model preservation