Summary of Blue Noise For Diffusion Models, by Xingchang Huang et al.
Blue noise for diffusion models
by Xingchang Huang, Corentin Salaün, Cristina Vasconcelos, Christian Theobalt, Cengiz Öztireli, Gurprit Singh
First submitted to arxiv on: 7 Feb 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Graphics (cs.GR); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper introduces a novel class of diffusion models that incorporate correlated noise within and across images during training and sampling. By utilizing time-varying noise and blue noise, the model improves generation quality compared to traditional Gaussian white noise methods. The framework also allows for correlation between images in a mini-batch, enhancing gradient flow. The authors evaluate their method on various datasets using the FID metric, achieving improvements over existing deterministic diffusion models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about creating a new way for computers to generate images by adding special kinds of noise to the training process. Normally, these computer programs use random noise, but this new approach uses noise that’s connected between different parts of an image. This makes it easier for the program to learn and improve its skills. The authors tested their method on different datasets and found that it did a better job than other methods at generating images. |
Keywords
* Artificial intelligence * Diffusion
