DetailFlow: 1D Coarse-to-Fine Autoregressive Image Generation via Next-Detail Prediction
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909897297231872 |
|---|---|
| author | Liu, Yiheng Qu, Liao Zhang, Huichao Wang, Xu Jiang, Yi Gao, Yiming Ye, Hu Li, Xian Wang, Shuai Du, Daniel K. Chen, Fangmin Yuan, Zehuan Wu, Xinglong |
| author_facet | Liu, Yiheng Qu, Liao Zhang, Huichao Wang, Xu Jiang, Yi Gao, Yiming Ye, Hu Li, Xian Wang, Shuai Du, Daniel K. Chen, Fangmin Yuan, Zehuan Wu, Xinglong |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_21473 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | DetailFlow: 1D Coarse-to-Fine Autoregressive Image Generation via Next-Detail Prediction Liu, Yiheng Qu, Liao Zhang, Huichao Wang, Xu Jiang, Yi Gao, Yiming Ye, Hu Li, Xian Wang, Shuai Du, Daniel K. Chen, Fangmin Yuan, Zehuan Wu, Xinglong Computer Vision and Pattern Recognition 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. |
| title | DetailFlow: 1D Coarse-to-Fine Autoregressive Image Generation via Next-Detail Prediction |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2505.21473 |