search

songweige / sd-webui-rich-text

117 stars 2 forks source link

Can't I just pull checkpoints from the Stable-diffusion folder and use them? #12

Open carnivar10 opened 9 months ago

carnivar10 commented 9 months ago

I don't know anything about the code, so I'm wondering if it's possible to change it if it's a simple task.

Nrgte commented 9 months ago

Yeah the extensions should just use the already selected model from A1111.

carnivar10 commented 9 months ago

ask

Are you saying that if I set the top checkpoint, the image will be created with the top checkpoint regardless of the Model ID checked at the bottom? the intent of the question was to generate an image based on the top checkpoint

Nrgte commented 9 months ago

I don't know, but I agree that the expectation is that it would generate the image with the top checkpoint. I don't know why there is a need for the bottom settings.

JPW0080 commented 9 months ago

The model format appears to be diffusers, ample storage space would be in order for conversions.

Acephalia commented 9 months ago

Can confirm that the diffusers model downloaded into the extension directory when I tried to generate so it is using that and not the set checkpoint in A1111.

Note: There is some disconnect error with downloading the diffusers model so you will most likely have to press generate multiple times to get it fully loaded the first time you run.

songweige commented 9 months ago

Thanks for the great comments. I agree that a single loading button would be great. But as also found be JPW0080, the current code has a different implementation so the checkpoint is not directly shared. That's why two different checkpoint selectors.

Nrgte commented 9 months ago

Thanks for the great comments. I agree that a single loading button would be great. But as also found be JPW0080, the current code has a different implementation so the checkpoint is not directly shared. That's why two different checkpoint selectors.

Is it possible to change this or is this a fundamental underlying limitation of the format? If it's the later, are there any good guides you recommend to convert models from one format to the other?