Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Orthogonal Adaptation for Modular Customization of Diffusion Models

About

Customization techniques for text-to-image models have paved the way for a wide range of previously unattainable applications, enabling the generation of specific concepts across diverse contexts and styles. While existing methods facilitate high-fidelity customization for individual concepts or a limited, pre-defined set of them, they fall short of achieving scalability, where a single model can seamlessly render countless concepts. In this paper, we address a new problem called Modular Customization, with the goal of efficiently merging customized models that were fine-tuned independently for individual concepts. This allows the merged model to jointly synthesize concepts in one image without compromising fidelity or incurring any additional computational costs. To address this problem, we introduce Orthogonal Adaptation, a method designed to encourage the customized models, which do not have access to each other during fine-tuning, to have orthogonal residual weights. This ensures that during inference time, the customized models can be summed with minimal interference. Our proposed method is both simple and versatile, applicable to nearly all optimizable weights in the model architecture. Through an extensive set of quantitative and qualitative evaluations, our method consistently outperforms relevant baselines in terms of efficiency and identity preservation, demonstrating a significant leap toward scalable customization of diffusion models.

Ryan Po, Guandao Yang, Kfir Aberman, Gordon Wetzstein• 2023

Related benchmarks

TaskDatasetResultRank
Multi-Concept Image Generation12-concept dataset
Text Alignment0.654
26
Text-to-Image PersonalizationConcepts dataset
CLIP-I Score0.664
14
Single-concept customizationSingle-concept customization dataset
Accuracy V188.29
7
Single-concept customizationCustom Personalization Benchmark 8 concepts (test)
V1 Score83.62
7
Multi-concept customization fusion7-concept integration
GPU Memory (MB)1.28e+3
7
Single-concept customizationSingle-Concept Customization V1-V8
Score V129.38
7
Multi-concept Generation32 concepts
DINO0.428
5
Showing 7 of 7 rows

Other info

Code

Follow for update