A Noise is Worth Diffusion Guidance
Guardado en:
| 发表在: | arXiv.org (Dec 5, 2024), p. n/a |
|---|---|
| 主要作者: | |
| 其他作者: | , , , , , , , , , , |
| 出版: |
Cornell University Library, arXiv.org
|
| 主题: | |
| 在线阅读: | Citation/Abstract Full text outside of ProQuest |
| 标签: |
没有标签, 成为第一个标记此记录!
|
MARC
| LEADER | 00000nab a2200000uu 4500 | ||
|---|---|---|---|
| 001 | 3141682707 | ||
| 003 | UK-CbPIL | ||
| 022 | |a 2331-8422 | ||
| 035 | |a 3141682707 | ||
| 045 | 0 | |b d20241205 | |
| 100 | 1 | |a Ahn, Donghoon | |
| 245 | 1 | |a A Noise is Worth Diffusion Guidance | |
| 260 | |b Cornell University Library, arXiv.org |c Dec 5, 2024 | ||
| 513 | |a Working Paper | ||
| 520 | 3 | |a Diffusion models excel in generating high-quality images. However, current diffusion models struggle to produce reliable images without guidance methods, such as classifier-free guidance (CFG). Are guidance methods truly necessary? Observing that noise obtained via diffusion inversion can reconstruct high-quality images without guidance, we focus on the initial noise of the denoising pipeline. By mapping Gaussian noise to `guidance-free noise', we uncover that small low-magnitude low-frequency components significantly enhance the denoising process, removing the need for guidance and thus improving both inference throughput and memory. Expanding on this, we propose \ours, a novel method that replaces guidance methods with a single refinement of the initial noise. This refined noise enables high-quality image generation without guidance, within the same diffusion pipeline. Our noise-refining model leverages efficient noise-space learning, achieving rapid convergence and strong performance with just 50K text-image pairs. We validate its effectiveness across diverse metrics and analyze how refined noise can eliminate the need for guidance. See our project page: https://cvlab-kaist.github.io/NoiseRefine/. | |
| 653 | |a Random noise | ||
| 653 | |a Noise generation | ||
| 653 | |a Image quality | ||
| 653 | |a Image reconstruction | ||
| 653 | |a Image enhancement | ||
| 653 | |a Image processing | ||
| 653 | |a Noise reduction | ||
| 700 | 1 | |a Kang, Jiwon | |
| 700 | 1 | |a Lee, Sanghyun | |
| 700 | 1 | |a Min, Jaewon | |
| 700 | 1 | |a Kim, Minjae | |
| 700 | 1 | |a Jang, Wooseok | |
| 700 | 1 | |a Cho, Hyoungwon | |
| 700 | 1 | |a Sayak, Paul | |
| 700 | 1 | |a Kim, SeonHwa | |
| 700 | 1 | |a Cha, Eunju | |
| 700 | 1 | |a Jin, Kyong Hwan | |
| 700 | 1 | |a Kim, Seungryong | |
| 773 | 0 | |t arXiv.org |g (Dec 5, 2024), p. n/a | |
| 786 | 0 | |d ProQuest |t Engineering Database | |
| 856 | 4 | 1 | |3 Citation/Abstract |u https://www.proquest.com/docview/3141682707/abstract/embedded/6A8EOT78XXH2IG52?source=fedsrch |
| 856 | 4 | 0 | |3 Full text outside of ProQuest |u http://arxiv.org/abs/2412.03895 |