DetailFlow: 1D Coarse-to-Fine Autoregressive Image Generation via Next-Detail Prediction

TL;DR

DetailFlow introduces a novel coarse-to-fine 1D autoregressive image generation approach that models images through next-detail prediction, achieving high quality and efficiency with fewer tokens and faster inference.

Contribution

The paper proposes a new resolution-aware token sequence and parallel inference mechanism, significantly reducing tokens and accelerating image generation in autoregressive models.

Findings

Achieves 2.96 FID on ImageNet 256x256 with 128 tokens

Runs nearly 2x faster than previous methods like VAR and FlexVAR

Outperforms prior models in quality with fewer tokens

Abstract

This paper presents DetailFlow, a coarse-to-fine 1D autoregressive (AR) image generation method that models images through a novel next-detail prediction strategy. By learning a resolution-aware token sequence supervised with progressively degraded images, DetailFlow enables the generation process to start from the global structure and incrementally refine details. This coarse-to-fine 1D token sequence aligns well with the autoregressive inference mechanism, providing a more natural and efficient way for the AR model to generate complex visual content. Our compact 1D AR model achieves high-quality image synthesis with significantly fewer tokens than previous approaches, i.e. VAR/VQGAN. We further propose a parallel inference mechanism with self-correction that accelerates generation speed by approximately 8x while reducing accumulation sampling error inherent in teacher-forcing supervision. On the ImageNet 256x256 benchmark, our method achieves 2.96 gFID with 128 tokens, outperforming VAR (3.3 FID) and FlexVAR (3.05 FID), which both require 680 tokens in their AR models. Moreover, due to the significantly reduced token count and parallel inference mechanism, our method runs nearly 2x faster inference speed compared to VAR and FlexVAR. Extensive experimental results demonstrate DetailFlow's superior generation quality and efficiency compared to existing state-of-the-art methods.