Kdistrobox rm nameofbox
Nwould be a question for @PcChip
EOops meant to reply to @PcChip
Pjust trying to run the ./webui.py from AUTOMATIC1111
Pit has known issues with current versions of python according to reddit posts
ETry using comfyui instead
EAuto1111 is a mess
Nwe should actually collaborate and see if we can get Stable Diffusion running in a distrobox. We already have configurations for doing ollama, it would be cool to add stable diffusion to our set of ujust commands.
Pi wonder if automatic1111 would have worked if i had specified ubuntu:22.04
Pinstead of latest
Pmight just be that easy 
Ndepends if it has an older python I would assume.
Pi will try that tomorrow - thanks a ton for the suggestion!!
EI've seen people rage at comfyui because it's always running the latest pytorch. Very updated. :Hellmo:
EInstall comfyui manager along with it to handle custom_nodes extensions
Ndo you do your install containerized or directly on the host?
BYou try it yet? I have an air 1s arriving in a couple days.
J@Noel @PcChip this is more what you want
Jwhen people say "use a container" we mean using a container, not installing things inside a distrobox.
Noh this is sick.
JNLooks like there is no need to reinvent the wheel.
Jthis is why I make fun of distro people btw, ya'll always default to "I'm supposed to install a package and do things by hand."
JAnd I'm more like "someone lazier than me put this on docker hub"
Nfor sure. It would be cool to get this going in a quadlet like we do for Ollama.
Nthis one looks the most promising.
NSince it supports multiple different UIs.
Jyeah, that does look like a great default.
Aor in my case "somebody lazier than me made a flatpak of it lol"
NLooks like it doesn’t support AMD GPUs 
NFrom their wiki: “Unfortunately, AMD GPUs #63 and Mac #35 are not supported, contributions to add support are very welcome!!!!!!!!!!”
Jah bummer
K
Klove that the resolve container just werks
ANMay be potentially easier to reinvent the wheel, but who knows.
NWe definitely would want to support both Nvidia and AMD GPUs if we can.
Nespecially for image generation.
Pwouldn't it just work with zluda?
EOh that one has invoke, which has a super nice ui
Jwe have invoke,
ujust invokeaiNHow does invoke compare to something like comfy?
5Certain Irishmen
KNot as soft, more summoning
NI should know better than posting serious stuff in the Bazzite channel 
Jnow we just need to find the right vscode plugin so you can just run your own copilot off your fancy GPU and then we'd be done.
Ahttps://github.com/rjmacarthy/twinny i used this with ollama for a while myself
GitHub
The most no-nonsense, locally or API-hosted AI code completion plugin for Visual Studio Code - like GitHub Copilot but completely free and 100% private. - rjmacarthy/twinny
Jyeah!!!
