正在加载视频...

视频加载失败

Introducing Opendream, a better interface for diffusion models 👨‍🎨 Opendream brings much needed features to your workflow, such as: - writing extensions as single Python functions 👀 - layering and non-destructive editing 🍰 - saving and sharing 💾 And it's all open-source.

36,158 次观看 • 2 年前 •via X (Twitter)

10 条评论

Varun Shenoy 的头像
Varun Shenoy2 年前

Writing a new extension is as easy as writing a Python function and decorating it with `𝚘𝚙𝚎𝚗𝚍𝚛𝚎𝚊𝚖.𝚍𝚎𝚏𝚒𝚗𝚎_𝚘𝚙`. We’ve already written a few extensions you can try out, including ControlNet, @Meta's Segment Anything, and @OpenAI's DALL-E.

Varun Shenoy 的头像
Varun Shenoy2 年前

The biggest benefit to Opendream is non-destructive editing. Any operation is a function of other layers, and you do not lose access to them once you use them.

Varun Shenoy 的头像
Varun Shenoy2 年前

Check it out on Github right here: Like what you see or interested in adding an extension? Feel free to drop a star and make a PR!

Varun Shenoy 的头像
Varun Shenoy2 年前

Built with Shashank Rammoorthy, Rahul Shiv, and Eric Lou for our final project in CS 348K, one of the *best* classes I've ever taken.

Varun Shenoy 的头像
Varun Shenoy2 年前

Exciting to see Opendream at the top of HN 🥳

Jay Sethi 的头像
Jay Sethi2 年前

Looks sick Varun!

iamrobotbear (bk) 的头像
iamrobotbear (bk)2 年前

This is crazy cool, any chance to be able to implement SD XL & possibly a Google colab notebook to run it in? Cheers!

Varun Shenoy 的头像
Varun Shenoy2 年前

Yup, SDXL should be a straightforward extension!

Pruthvi Shetty 的头像
Pruthvi Shetty2 年前

Looks great @varunshenoy_ , thanks for sharing it!

Saurabh Patil 的头像
Saurabh Patil2 年前

this is amazing @varunshenoy_

相关视频