[enhancement]: Stable Diffusion 3

invoke-ai / InvokeAI

InvokeAI is a leading creative engine for Stable Diffusion models, empowering professionals, artists, and enthusiasts to generate and create visual media using the latest AI-driven technologies. The solution offers an industry leading WebUI, supports terminal use through a CLI, and serves as the foundation for multiple commercial products.

https://invoke-ai.github.io/InvokeAI/

Apache License 2.0

22.29k stars 2.32k forks source link

[enhancement]: Stable Diffusion 3 #6508

Open sammcj opened 3 weeks ago

sammcj commented 3 weeks ago

Is there an existing issue for this?

[X] I have searched the existing issues

Contact Details

No response

What should this feature add?

Stable Diffusion 3 (medium) has been released, it would be great to see this available in InvokeAI.

https://huggingface.co/stabilityai/stable-diffusion-3-medium

Alternatives

No response

Additional Content

sammcj commented 3 weeks ago

As an aside, a fantastic summary of the SD3's hostile licensing from none other than the wonderful @hipsterusername - https://youtu.be/r8oA42saTNI

So just be aware that once this lands in InvokeAI - it may not be the right model for you if you're in a commercial environment. Hopefully for the hobbyists and non-commercial artists it can still be a useful tool in the chest once we get past the quality issues.

sammcj commented 2 weeks ago

After spending several days working with SD3 in Comfy I’m actually quite excited for SD3 support to land in InvokeAI despite the negativity around the model and its licensing.

Replicate has good tutorials today that provides guidelines on the basic differences when using SD3 vs SDXL - https://replicate.com/blog/get-the-best-from-stable-diffusion-3

Looking forward to this as another tool in the chest.