TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Cold Diffusion: Inverting Arbitrary Image Transforms Without Noise

9 点作者 lnyan将近 3 年前

2 条评论

MathYouF将近 3 年前
After an excellent all day workshop at CVPR this year explaining diffusion in more detail, it seemed pretty clear that any noise function could be used. I&#x27;m not sure if it should have been obvious, but I felt this paper coming from a mile away after seeing that.<p>I wonder to what degree certain parts of diffusion dictate using certain noise functions, and how much this paper truly challenges how we understand them. Cool to see it was researched.<p>Next idea: it seems like a lot of steps could be skipped by using things like momentum during the inference time. I&#x27;m sure OpenAI has already implemented several clever tricks like that in production for DallE.<p>I&#x27;m working on (various, non-diffusion) methods for 2D drawing to 3D output right now.
ggm将近 3 年前
Shame neither a Pons, nor a Fleischmann are on the author list. I love how people can intrude contextual puns and jokes into paper titles.