Load your own Stable Diffusion Model

[oalexdoda]

Is there a way or feature planned to add your own model?

[ee0pdt]

Seems linked to #9

[oalexdoda]

Partially, but referring to actually adding your own model vs. using the predefined one.

[ee0pdt]

Do you mean fine tuning? That’s probably better than trying to make your own model

[airwindows]

Anything along the lines of textual inversion https://huggingface.co/docs/diffusers/training/text_inversion would be a huge deal for me and what I'm doing :) I believe this would qualify as 'adding your own model' and I think it'd be very welcome on a standalone Mac app version.

I think it's very likely artists capable of taking advantage of this technology might be found on Mac, needing to run a standalone app of Stable Diffusion…

[designium]

Could we train using https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_textual_inversion_training.ipynb#scrollTo=tAZq3vFDcFiT and then put the "fine tune" model in some folder inside Diffusion Bee?

[joshdance]

Being able to load your own .ckpt files generated by Dreambooth would be awesome.