AccDiffusion: An Accurate Method for Higher-Resolution Image Generation

TL;DR

AccDiffusion introduces a novel patch-wise image generation method that decouples prompts and employs dilated sampling, significantly reducing object repetition and improving high-resolution image quality without additional training.

Contribution

The paper presents the first decoupling of patch prompts and dilated sampling techniques for accurate, high-resolution image generation without training.

Findings

Reduces object repetition in high-res images

Improves global consistency in patch-wise generation

Outperforms existing methods in quality and accuracy

Abstract

This paper attempts to address the object repetition issue in patch-wise higher-resolution image generation. We propose AccDiffusion, an accurate method for patch-wise higher-resolution image generation without training. An in-depth analysis in this paper reveals an identical text prompt for different patches causes repeated object generation, while no prompt compromises the image details. Therefore, our AccDiffusion, for the first time, proposes to decouple the vanilla image-content-aware prompt into a set of patch-content-aware prompts, each of which serves as a more precise description of an image patch. Besides, AccDiffusion also introduces dilated sampling with window interaction for better global consistency in higher-resolution image generation. Experimental comparison with existing methods demonstrates that our AccDiffusion effectively addresses the issue of repeated object generation and leads to better performance in higher-resolution image generation.