CompSlider: Compositional Slider for Disentangled Multiple-Attribute Image Generation

2025-08-31

AI Summary

Fine-grained, multi-attribute (e.g., age, expression) co-editing in text-to-image (T2I) generation remains challenging due to attribute entanglement; existing slider-based methods suffer from cross-attribute interference and lack independent, precise control. Method: We propose CompSlider—a framework that enables decoupled attribute editing without fine-tuning the base diffusion model. It introduces a latent-space conditional prior modeling scheme coupled with a slider-adaptation mechanism, jointly optimized via a disentanglement loss and a structure-consistency loss to achieve attribute disentanglement while preserving geometric integrity. Contribution/Results: CompSlider is compatible with both T2I and text-to-video (T2V) generation, supporting high-fidelity, efficient multi-attribute composite editing. Extensive experiments validate its effectiveness across diverse attributes and demonstrate successful generalization to video generation, establishing new state-of-the-art performance in controllable generative editing.

Technology Category

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

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

In text-to-image (T2I) generation, achieving fine-grained control over attributes - such as age or smile - remains challenging, even with detailed text prompts. Slider-based methods offer a solution for precise control of image attributes. Existing approaches typically train individual adapter for each attribute separately, overlooking the entanglement among multiple attributes. As a result, interference occurs among different attributes, preventing precise control of multiple attributes together. To address this challenge, we aim to disentangle multiple attributes in slider-based generation to enbale more reliable and independent attribute manipulation. Our approach, CompSlider, can generate a conditional prior for the T2I foundation model to control multiple attributes simultaneously. Furthermore, we introduce novel disentanglement and structure losses to compose multiple attribute changes while maintaining structural consistency within the image. Since CompSlider operates in the latent space of the conditional prior and does not require retraining the foundation model, it reduces the computational burden for both training and inference. We evaluate our approach on a variety of image attributes and highlight its generality by extending to video generation.

Problem

• Disentangle multiple attributes in slider-based image generation
• Enable reliable independent manipulation of image attributes
• Reduce computational burden for training and inference

Innovation

• Generates conditional prior for multi-attribute control
• Introduces disentanglement and structure loss functions
• Operates in latent space without retraining foundation model