Concept Sliders: LoRA Adaptors for Precise Control in Diffusion Models

TL;DR

This paper introduces Concept Sliders, a novel method for precise, interpretable control over image attributes in diffusion models using low-rank parameter directions, enabling targeted edits and concept transfer.

Contribution

The paper proposes a new approach to create interpretable concept sliders for diffusion models, allowing precise, continuous control over image attributes with minimal interference and easy composition.

Findings

Sliders outperform previous editing techniques in targeted edits.

Sliders can transfer concepts from StyleGAN for intuitive editing.

Method improves image quality issues like object deformation and distorted hands.

Abstract

We present a method to create interpretable concept sliders that enable precise control over attributes in image generations from diffusion models. Our approach identifies a low-rank parameter direction corresponding to one concept while minimizing interference with other attributes. A slider is created using a small set of prompts or sample images; thus slider directions can be created for either textual or visual concepts. Concept Sliders are plug-and-play: they can be composed efficiently and continuously modulated, enabling precise control over image generation. In quantitative experiments comparing to previous editing techniques, our sliders exhibit stronger targeted edits with lower interference. We showcase sliders for weather, age, styles, and expressions, as well as slider compositions. We show how sliders can transfer latents from StyleGAN for intuitive editing of visual concepts for which textual description is difficult. We also find that our method can help address persistent quality issues in Stable Diffusion XL including repair of object deformations and fixing distorted hands. Our code, data, and trained sliders are available at https://sliders.baulab.info/