InverFill: One-Step Inversion for Enhanced Few-Step Diffusion Inpainting

TL;DR

InverFill introduces a one-step inversion technique that enhances few-step diffusion inpainting by injecting semantic information into the initial noise, achieving high-fidelity results efficiently without retraining.

Contribution

The paper presents InverFill, a novel one-step inversion method that improves few-step diffusion inpainting by incorporating semantic information, eliminating the need for training or heavy optimization.

Findings

Significantly improves image quality in few-step inpainting.

Matches the performance of specialized inpainting models at low NFEs.

Does not require real-image supervision or extensive retraining.

Abstract

Recent diffusion-based models achieve photorealism in image inpainting but require many sampling steps, limiting practical use. Few-step text-to-image models offer faster generation, but naively applying them to inpainting yields poor harmonization and artifacts between the background and inpainted region. We trace this cause to random Gaussian noise initialization, which under low function evaluations causes semantic misalignment and reduced fidelity. To overcome this, we propose InverFill, a one-step inversion method tailored for inpainting that injects semantic information from the input masked image into the initial noise, enabling high-fidelity few-step inpainting. Instead of training inpainting models, InverFill leverages few-step text-to-image models in a blended sampling pipeline with semantically aligned noise as input, significantly improving vanilla blended sampling and even matching specialized inpainting models at low NFEs. Moreover, InverFill does not require real-image supervision and only adds minimal inference overhead. Extensive experiments show that InverFill consistently boosts baseline few-step models, improving image quality and text coherence without costly retraining or heavy iterative optimization.