Triple Underscore

ROCm is flaky in regular consumer GPUs at the best of times. I’m surprised you could even get that far on a steam deck.

Try the command line arg --opt-sdp-attention. You might also want to try out --medvram or --lowvram (4GBs is considered low when it comes to AI). although I have a feeling it’s just because of the custom nature of the deck’s APU.

Your best bet would be to search for builds of ROCm, PyTorch and Torchvision that are specifically made for the deck, if such things even exist.

For anybody confused: The Lemmy version of the image (the one with pictrs in the URL) won’t have it, it strips out the EXIF data (or wherever else the prompt is stored). You have to download the image from catbox (click on the title, not the thumbnail), which seems to keep the data.

While I went through the effort I may as well throw it in for future reference:

A blonde 25yo news anchor unexpectedly (riding a vibrator:1.1) to orgasm on live TV, (sweaty, spent, disheveled, clenching muscles:1.1), full body view
Negative prompt: child, childish, teen, easynegative, closeup, doll, penis
Steps: 30, Sampler: DPM++ 2M SDE Karras, CFG scale: 7, Seed: 639212525, Size: 512x512, Model hash: c4625c7cd3, Model: qgoPromptingreal_qgoPromptingrealV1, Version: v1.3.0

I don’t think that’s necessary. Not sure though.

It’s probably “Add python to environment variables.” They must have changed the wording on that at some point.

Not on Windows right now so I can’t confirm, but you probably forgot to pick “Add Python to PATH” or whatever the option is in the installer. Try running the Python installer again, maybe it’ll let you add it without needing to uninstall & reinstall

Edit: If you’re on Nvidia, there seems to be a simpler install method now: https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-NVidia-GPUs (Method 1)

how much space was the download(s)?

On my end, it’s sitting at ~64GB (with btrfs compression shenanigans), though 60 of those are from all the models I have installed. The download would probably be ~2GB, even less if you disable downloading the “default” models with --no-download-sd-model and instead pick models off of Civit or wherever manually.

Edit: Should have mentioned. Most full models are between 2-4 GBs each. Some can be 5+ but they tend to be “full” versions intended for merging & such. LoRAs are generally smaller. Depending on how much they’re pruned they’ll be anywhere between 10-100 MBs each.

how confusing is the software to use?

There’s definitely a learning curve, yes. But there’s plenty of resources (and more importantly, examples) out there.

what kind of limitations does the software have? can i do multiple people? monsters? futa? etc.

As long as you have the correct models set up it can generate basically anything. At least with anime models, monsters and futa are a given. Your main issue will probably be multiple people, although there are solutions to that. (See the multidiffusion upscaler GitHub repo on the main post)

By Web UI it means that the graphical part of it – where you write your prompt and hit generate – is running inside your browser and not as a separate window or command line. Everything is kept on your own computer unless you explicitly tell it to open up remote access.

It’s all free, yes.

For the how: Aside from the webui (or whatever else you’re running the stable diffusion code on), you’ll need models that instruct it what to create. The builtin models are terrible for porn.