When we will be able to run it locally ?

[2blackbar]

Hey i have gtx1080ti , will the 1024 model run on it or im getting hopes up? Im using discord SD bot currently but would love to try inpainting.

[Utopiah]

Indeed, would be nice to have a Dockerfile then interfaces like gradio or even the entire setup via cog being able to integrate with other tools, e.g Gimp, Krita via their plugin system or even tools for children e.g Scratch.

[litevex]

Hey i have gtx1080ti , will the 1024 model run on it or im getting hopes up? Im using discord SD bot currently but would love to try inpainting.

Going by VRAM, the 512 model will run. We don't currently know how large the 1024 model is, but if it's the same size it should run too.

[Lufffya]

When is the model expected to be available for download，Expecting!!!

[Iuryck]

Gonna comment here to follow

[bibyts]

What's the link to Discord Stable Diffusion bot?

[melMass]

Discord Stable Diffusion bot

Stable Diffusion is entirely free, but if you aren’t already in the Discord server, you can’t join.

[johndpope]

@Utopiah - fyi Dockerfile - https://github.com/CompVis/stable-diffusion/pull/24

[Utopiah]

Thanks @johndpope, just waiting for the open source / freely accessible weights now then.

[breadbrowser]

Thanks @johndpope, just waiting for the open source / freely accessible weights now then.

they will most likely will not.

edit: this was before it was public.

[nathanshipley]

Thanks @johndpope, just waiting for the open source / freely accessible weights now then.

they will most likely will not.

The weights have already been released for researchers already and will be released to the public "soon." Emad talks about this pretty constantly: https://twitter.com/EMostaque/status/1559588668637122561

[Utopiah]

@SirFrasi do you have information contradicting https://github.com/huggingface/diffusers/releases/tag/v0.2.3 availability today?

[danielbulhosa]

Is there a notebook or code example of inpainting?

[Utopiah]

Now we can, see https://huggingface.co/CompVis

PS: in case you get confused use the checkpoints for CompVis, not HuggingFace Diffusers, so here https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/raw/main/sd-v1-4.ckpt (you need to login first, read then agree to terms from the model card) otherwise you'll need to use https://github.com/huggingface/diffusers/ instead AFACT.

PPS: if like me you get stuck in Out of Memory Error check https://github.com/CompVis/stable-diffusion/issues/54 and the recommended https://github.com/lstein/stable-diffusion , works for me on a 2080ti in Docker.

[bibyts]

I'm a little confused. I went to this link and downloaded the file but it didn't download a .ckpt file in Gitbash: https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/raw/main/sd-v1-4.ckpt

Where can I find the .ckpt file to use? I was trying to download the oid sha256 from the link above: oid sha256:fe4efff1e174c627256e44ec2991ba279b3816e364b49f9be2abc0b3ff3f8556

[hcrawfdev]

@bibyts Do you have git-lfs installed? https://docs.github.com/en/repositories/working-with-files/managing-large-files/installing-git-large-file-storage After you do, you should be able to run git clone https://huggingface.co/CompVis/stable-diffusion-v-1-4-original and it will include the large .ckpt file.

[Utopiah]

I went to this link and downloaded the file but it didn't download a .ckpt file in Gitbash: https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/raw/main/sd-v1-4.ckpt

@bibyts you can find it on https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/tree/main and use the web interface to download it. As mentioned in the ReadMe you need a HuggingFace account to accept the license.

[karray]

Do I have to use the sd-v1-4.ckpt weights for the inpaint.py example?

[breadbrowser]

demo here https://huggingface.co/spaces/stabilityai/stable-diffusion

[Utopiah]

@breadbrowser that's not local.

[wolfgangmeyers]

Do I have to use the sd-v1-4.ckpt weights for the inpaint.py example?

I think this is a different model that doesn't have the weights released yet.

[breadbrowser]

@breadbrowser that's not local.

but it does run without error and hassle

[karray]

I think this is a different model that doesn't have the weights released yet.

I handled it. Here is what I did.

[wolfgangmeyers]

I think this is a different model that doesn't have the weights released yet.

I handled it. Here is what I did.

Oh nice. Going to test this out. Wonder if there will be another version that allows text guidance.

[Dav1d-Fn]

@SirFrasi do you have information contradicting https://github.com/huggingface/diffusers/releases/tag/v0.2.3 availability today? @Utopiah My mistake, I didn't mean stable diffusion per se but, the discord project from stability ai, which now only works with credits at dreamstudio

[Utopiah]

@breadbrowser that's not local.

but it does run without error and hassle

Ok and that's useful but this issue, issue #1 is specifically about it being run locally.

[TechnologyClassroom]

This issue can be closed. The models were released through huggingface last week.

[Utopiah]

@TechnologyClassroom , yep, posted https://github.com/CompVis/stable-diffusion/issues/1#issuecomment-1222948267 7 days ago.

[breadbrowser]

@breadbrowser that's not local.

but it does run without error and hassle

Ok and that's useful but this issue, issue #1 is specifically about it being run locally.

https://www.kaggle.com/code/rey6868/stable-diffusion-public