Install Guide
The day has arrived. The launch of Umi AI is upon us! After nearly two weeks of frantic work and assistance from my excellent coder, Umi AI is ready to reach the hands of the people!
What is Umi AI? Well, it's not actually an AI. It is a mod for Automatic1111's WebUI, like a mod for Skyrim, as an example. What I've done is create a 'Wildcard System' that uses the WebUI interface to allow for simple and accessible character randomization.
Want to know more about Umi AI? Read this post to see more details.
Take careful note of the PC requirements.
-
Nvidia GPUs only, not AMD. If you have an AMD GPU you're basically screwed here.
-
WebUI now supports Linux and MacOS. However, you will have to refer to Voldy's Guide if you want to install on those systems.
-
If your computer is old, or it sucks, you can't use the WebUI to generate characters. Not sure if your rig sucks? Download Speccy and post your specs in the Umi AI Discord to be sure before you spend multiple hours wasting your time on a fruitless task. We'll let you know if your pc is good enough.
To install WebUI, there is a simple method now available! This method is only for Windows. If you are on Linux or Mac, you will have to refer to Voldy's Guide.
Step 1: Download the Autoinstaller.exe file from this Github page:
https://github.com/EmpireMediaScience/A1111-Web-UI-Installer/releases
As an example, you can find the download link on whatever the newest release is, here:
When you run the .exe, you will end up spending about 5 minutes or less actually clicking buttons, but about 30 minutes downloading things, depending on your internet speed. If your internet is really slow, you should be warned you may have to download 10 or even 20 GBs of data.
The autoinstaller is great because it installs all the annoying dependencies and other crap automatically. Just keep clicking along and following directions until you reach this page:
Select where you want WebUI installed. It's up to you. I recommend putting it as close to the root/top level of a drive as possible.
Clean install only matters if you've installed WebUI in the past and maybe screwed some stuff up. This will erase it all and start over.
You can also install the Stable Diffusion model, version 1.5. As of this moment, v2.0 already exists, but 1.5 still has its merits. Personally, I recommend leaving this checked so you have a default model installed. I will link you to many other models later.
Once you finish installing and downloading stuff, the installer will close. A folder will open up at the install location with a Launcher shortcut. Run it.
I recommend toggling the options like so:
Auto Update WebUI: CHECK
Auto Update Extensions: CHECK (Useful for Umi AI!)
Clear Generated Images: UNCHECK
Low VRAM: If you have an older graphics card, you will want to check this. If you're not sure, join the Umi AI Discord and post your specs there. Our team will tell you if your card is good for low, medium, or high VRAM. This is a very important option.
Enable Xformers: Depends on your card, see above.
Convert Model: See above.
You can also add 'additional arguments' here. This is going to be HIGHLY dependant on the card and model you use. But you can add these later, so just leave it blank for now.
Finally, Launch WebUI.
Do you like waiting? I hope so. You're going to be waiting a LONG while. This black screen will probably require about 30 minutes to download everything it needs to run. (You only need to do this once.) The installed will need to install 'torch' and 'torchvision' and a buuuunch of other stuff that makes AI work.
So, we wait.
And wait.
And wait some more.
...
Still waiting.
If you get an error at this point, join the Umi AI Discord and post the error there. Otherwise, we should be done!
With WebUI having popped up, check out the tabs along the top.
Find the Extensions tab, then the Install from URL sub-tab. Paste the following line, as shown above, then hit install:
https://github.com/Klokinator/UnivAICharGen.git
With that, Umi AI will be installed.
Now, how do you use Umi AI?
What features does it have?
What Models, Checkpoints, Embeddings, and other things can you add to WebUI?
What settings should you choose?
How do you prompt?
What is txt2img? Img2img? Inpainting? Outpainting??
I will be writing detailed guides for all of these things, moving forward.
For right now though, I advise that if you run into any issues or have any questions that you visit the Umi AI Discord!
>>>>>>>Visit the Umi AI Discord!<<<<<<<
Next post, learn about what makes Umi AI worth using in the first place!