A standalone, locally installed version of Stable Diffusion from KMVD SD GUI is now in version 1.7.1 over at itch.io. I have used this GUI since 1.5 for testing and article research. To recap for those that aren’t aware of this little jewel, it requires no installation. Just download, unzip, find the exe file, and get started. It is just that simple and many are taking advantage of this free or “pay what you will” Stable Diffusion interface.
If you want to get into using an AI generator, then this is a simple way to do just that. It also comes with a model file that covers a lot of ground in terms of excellent image generation in many styles. Mind you, this is Stable Diffusion, Version 1.5. The recently released Version 2 has some limitations added to it that took away artist styles and mimicked people. We are not focusing on that version as it seems to be a step backward with the new limitations. Version 2 is a story for another time and now is certainly not a bad time to snag a GUI that works with 1.5 to retain those options.
One thing that some new users miss is the fact that there are other, free models out there from places like Hugging Face that offer specific models trained on certain conditions like classic animation, The Spiderverse, Cyberpunk, Elden Ring, Archer (yes, the TV series), Redshift, Robo and Van Gogh to name a few. These models are more specifically trained in these types and styles of animation, gaming, and painting.
Same Prompt: “Raven Haired Heroine” with specific models.
The above image represents the same prompt using the various models including the default model. These specifically trained models can produce some amazing results and some unexpected ones as well.
You can download these models over at Hugging Face. You will have to click on a model and drill down to the ckpt file which is the model file. The Stable Diffusion 2 model was added to this repository one day before I wrote this article so you can download it too even though we’re focusing on the version 1.5 models.
Clicking on the model banner/link will take you to the Model Card page. Look for a menu in the top portion of the page that says: Model Card Files Community and click on the Files link. The ckpt file may be in this location or in one of the subfolders. After downloading you can add the location of the model file to the GUI or copy model files over to your existing GUI within the Data/Models folder. I chose the latter option so I would have all my model files in one location.
From here you can go into the Settings menu and select the model in the Stable Diffusion Model dropdown box. Just using these specifically trained models doesn’t mean you are going to get Robots every time you use the Robo model or an Archer-like model as that will depend on your written prompt and the weight given to it.
The Model Selection Drop-Down Menu in NMKD Stable Diffusion GUI
It may sound complicated but there aren’t that many controls to play around with and experimentation is lots of fun so you can eventually tweak your prompt to bring you something more like what you have in mind.
Once you learn to drill down to the ckpt file on the Hugging Face website you can then search for any model files that may fit a certain need. Or… just download them for fun and see what happens. With enough experimentation you can be creating model specific renderings without having to use long prompts with lots of positive and negative wording.
M.D. McCallum, aka WarLord, is an international award-winning commercial graphics artist, 3D animator, published author, project director, and webmaster with a freelance career that spans over 20 years. Now retired, M.D. is currently working part-time on writing and select character development projects. You can learn more about MD on his website.