There are five different models:
Realistic, Semi-Realistic, Anime, Cartoon and Storybook
These are not my models. I use the realistic one so much that I wanted to be able to use it on CivitAI. (though for some reason I can't get them to be an option to use online in CivitAI's generator)
On the source page, it says the license is GPL-3.0:
https://huggingface.co/rippertnt/SD-MODELS/tree/main
Additional models, including later versions, are on this page:
https://huggingface.co/circulus
I don't know how to get those versions to work in Automatic1111, so I am only posting the 3.6 versions.
If the source of these models wants them removed, so they can put them on their own account, let me know. I think that person is here:
https://civitai.com/user/rippertnt
They have not posted anything since version 2.8 of some of the models as of me posting these merges.
For what I think is a "full" version of the realistic model, there is information in the "Metadata_for_full_merge" zip file (listed as "Training Data") on that model's page that came from the internal metadata that Automatic1111 extracted for that model. (the 4GB version, not the 2GB one) I assume all of the models here are merges, but only this file had this additional data.
Some of these models lean toward NSFW, so I am tagging them as such. If some of them don't seem to lean that way, I can adjust them later. I only ever use the realistic one, and that definitely leans NSFW, so I don't know much about the others.
I've been able to create some really unique imagery using the realistic model. I don't know how to do it well enough to generate flawless imagery though. Maybe others will be able to come up with prompts that work well. Over time, I will post some of the images I previously created if they can be recreated properly. I don't want to post an image that can no longer be recreated exactly as of posting. What a prompt generates changes for me over time, from software updates to messing with settings perhaps. The realistic model doesn't work as good as other realistic models, but I think it's able to handle some concepts with just a prompt better than some others I've tried.
I originally used a free generator online to create some cool imagery from the realistic model:
https://huggingface.co/spaces/Manjushri/Manju-Dream-Booth-GPU
They have since updated it to include some of the other models. You can try it there if you like, but it uses a newer version of the models.
Even before it was updated to a newer version, I couldn't exactly replicate imagery due to things I don't fully understand. Weights being applied perhaps? I don't know. Now that this model is here, maybe others can explore what it can do and that can help me as well. So yes, selfish reasons too.
Links to original sources for each model are below. I originally added the links to the "About this version" section, but then in Automatic1111 with the CivitAI extension the link is active on the model image you click to choose a model. That could get annoying if you accidentally click that. So, I'll add them here.
Realistic (Pruned):
Realistic (Full):
https://huggingface.co/rippertnt/SD-MODELS/blob/main/canvers-realistic-v3.6.safetensors
Semi-Realistic (Pruned):
https://huggingface.co/rippertnt/SD-MODELS/blob/main/v3.6/canvers-semi-v3.6-fp16-no-ema.safetensors
Anime (Pruned):
https://huggingface.co/rippertnt/SD-MODELS/blob/main/v3.6/canvers-anime-v3.6-fp16-no-ema.safetensors
Cartoon (Pruned):
https://huggingface.co/rippertnt/SD-MODELS/blob/main/v3.6/canvers-toon-v3.6-fp16-no-ema.safetensors
Storybook (Pruned):
https://huggingface.co/rippertnt/SD-MODELS/blob/main/v3.6/canvers-toon-v3.6-fp16-no-ema.safetensors
I haven't posted them, nor have I ever tried them, but version 3.5 of some of these models are also available here:
https://huggingface.co/rippertnt/canvers-image-v3.5/tree/main
Only "Semi-Realistic" is not available. For the other four, the pruned and full models seem to be available.
Description
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.









