Ceci est une ancienne révision du document !
In part six An Introduction to Stable Diffusion, we will look at models and how to install them in Stable Diffusion. For Easy Diffusion, 2.5.41, go to Image Settings and select Model. Then click on the drop-down arrow which brings up the default Model sd-v1-4 (there are no other choices.)
A variety of other Models exist and may be added. Some can be found at the links below: https://github.com/easydiffusion/easydiffusion/wiki/Custom-Models https://www.reddit.com/r/stablediffusion/wiki/models/ https://civitai.com/?types=Checkpoint https://cyberes.github.io/stable-diffusion-models/ https://rentry.org/sdmodels
Looking at Model pages gives you an idea of their purposes, which often have a theme or focus that may or may not be appropriate for your design. Because they may be large, they might take a while to download, so you might want to be sure the models are appropriate for your needs before beginning that process. It’s always good practice to check the Models for viruses via Clam AV or other anti-virus programs. Stable Diffusion uses the CheckPoint (name.ckpt) and Safetensors (name.safetensors) versions of Models. In Easy Diffusion the file type is not shown in the Model drop-down choice.
Another smaller tweak to Models can be accomplished by changing the Variable AutoEncoder (VAE). These are mostly used for solving smaller issues such as multiple legs, fingers or poorly drawn faces which may also be addressed using negative prompts. Examples for dealing with problems are: 1))), 2)), 3), 4), out of frame, extra fingers, mutated hands and similar prompts. It’s expected that the upcoming model initially named SDXL 1.0 will resolve these problems and also include the ability to add desired text. That release may be available before this article is published. Presumably the update will be installed in a fashion similar to other models as described below. After installation, added new models will be listed as below.
Similarly, Automatic1111 has the same options although their type is indicated in the drop-down menu. Given that Models are mostly derivative, pared down or otherwise modified versions of Models first made available and sometimes large, careful selection is appropriate. Hopefully, we may cover the process of modifying Models in a future article.
Once you have selected and downloaded additional Models, you need to move/copy them to the appropriate location. The locations for downloading the Easy Diffusion installation are: easy > easy-diffusion > models > stable-diffusion. Likewise for Automatic111: automatic1111 > stable-diffusion-webui > models > stable-diffusion. Restart Stable Diffusion and the new Models should be available.