Lightning-Fast Image Inversion and Editing for Text-to-Image Diffusion Models

TL;DR

This paper introduces a fast, guided Newton-Raphson based method for image inversion in diffusion models, enabling real-time image editing and interpolation with high accuracy and efficiency.

Contribution

It formulates diffusion inversion as a root-finding problem and develops a guided Newton-Raphson method that achieves rapid, high-quality reconstructions and edits in diffusion models.

Findings

Inverts images within 0.4 seconds on an A100 GPU.

Achieves high-quality image reconstructions and edits.

Improves image interpolation and rare object generation.

Abstract

Diffusion inversion is the problem of taking an image and a text prompt that describes it and finding a noise latent that would generate the exact same image. Most current deterministic inversion techniques operate by approximately solving an implicit equation and may converge slowly or yield poor reconstructed images. We formulate the problem by finding the roots of an implicit equation and devlop a method to solve it efficiently. Our solution is based on Newton-Raphson (NR), a well-known technique in numerical analysis. We show that a vanilla application of NR is computationally infeasible while naively transforming it to a computationally tractable alternative tends to converge to out-of-distribution solutions, resulting in poor reconstruction and editing. We therefore derive an efficient guided formulation that fastly converges and provides high-quality reconstructions and editing. We showcase our method on real image editing with three popular open-sourced diffusion models: Stable Diffusion, SDXL-Turbo, and Flux with different deterministic schedulers. Our solution, Guided Newton-Raphson Inversion, inverts an image within 0.4 sec (on an A100 GPU) for few-step models (SDXL-Turbo and Flux.1), opening the door for interactive image editing. We further show improved results in image interpolation and generation of rare objects.