So I work at a home improvement store, and one of my co-workers does some contracting work on the side. He is trying to encourage one of his neighbors to put some simple small-park kind of stuff on a plot of land he owns so that he (my co-worker) can pick up some extra business installing it.

He's seen me messing around with Stable Diffusion on some web apps at work on my down time, and he asked if it was possible if I could take a photo of the site and use AI to insert some of these elements into it so that he could show it to this potential client and maybe sell it to him that way.

"Sure," I said, thinking to myself, 'I can just use inpainting to blend this stuff into the image pretty seamlessly. Easy-peasy.'

It took me almost a full day of on-again, off-again work to get a picnic table I could live with. But I CANNOT get any model, any prompt, anything to make a swing set that I can live with. I've been pecking away at this problem for several days now, and every single attempt at a swing set has resulted in something that is mangled, twisted, or some terrible hybrid of OTHER playground equipment that my co-worker definitely doesn't want in the scene.

At this point I'm just working on it for the challenge, but I admit that I'm stumped. Short of training my own Lora, does anyone have any advice on how to make a coherent swing set to bring into this image? >_< Yes, this is a silly problem to have, I admit that, but I've also learned a great deal about how Stable Diffusion 'thinks' in the last few days, so I consider it a learning experience. Still, does anyone else have any ideas?

Just wanted to put this out there for anyone else who was in the same position, as I'd spent some time banging on this to find a functioning combination and would have appreciated having had success reports myself.

Running Debian Trixie, current as of July 22, 2023.

I see 512x512 speeds of about 2.2 it/s, which is significantly slower than an lower-end Nvidia card than I'd used and significantly (about 1/8th the speed) that other people have reported getting the same XT 7900 XTX card running at on Linux), so there is probably more work for me to do, but it's definitely running on the GPU and is much faster than running on the CPU, so I know that this combination (vanilla system Python, vanilla system drivers, torch nightly in a venv) does at least work, which was something that I'd been unsure of up until now.

Running on the host, no Docker containers. Using a venv. Automatic1111 web UI, in-repository drivers, 2.1.0.dev20230715+rocm5.5 torch via pip installed in a venv, standard system Python 3.11 (i.e. did not need to set up Python 3.8, as I've seen some people do). Needs the non-free-firmware apt repo component enabled; I have firmware-amd-graphics-20230515-3. Rocm 5.6 is out as of this writing from AMD, but Debian Trixie presently only has 5.5 packaged and in the repos.

I did need to install libstdc++-13-dev -- only libstdc++-12-dev being installed caused Automatic1111 to bail out with an error in not being able to find a limits C++ header when building some C++ code at runtime; some users had run into a similar error and resolved it by installing libstdc++-12-dev, which was a bit confusing. I have both clang and g++ installed. I am not terribly familiar with the AMD ROCM stack, but my understanding is that part of it (libamdhip64?) performs some compilation at runtime; it apparently remembers the binaries it has compiled, as if I removed libstdc++-13-dev after a successful run, it continued to work.

The user running the Automatic1111 frontend needed to be added to the render and video groups to have access to the requisite device files.

I did not need to have HSA_OVERRIDE_GFX_VERSION set.

As for options being passed in COMMAND_ARGS, just --medvram and --api.

--xformers does not work with AMD cards; Stable Diffusion (or Automatic1111, unsure about responsibility in the stack) apparently just ignores it there; passing it doesn't break anything.

Some --opt-sdp options, like --opt-sdp-attention, cause dramatic slowdown, I assume causing the generation to run on the CPU instead of the GPU. I'd suggest to anyone trying to get a similar environment running to not start including optimization flags until they have things working without them; this had complicated things for me.

I see 2.59 it/s, so something like 20% higher performance, without --medvram being passed to COMMANDLINE_ARGS.

I have not done extensive testing to see whether any issues show up elsewhere with Stable Diffusion.

This is a Mastodon thread I created featuring devices and computers that never were, but could have been. I think my favorite might be the HD Laserdisc player called the MOID.

https://mastodon.jordanwages.com/system/media_attachments/files/110/686/620/475/187/344/original/95ff43d76ba41dd6.jpg

Learning to inpaint and messing around with a steampunk lycoris

I thought I would share my experience this evening with the group here seeing that I still excited as hell for getting this hodgepodge to work at all.

I have been playing with the machineMl version of stable diffusion on Windows for a while now (we won't go into the reasons why, but I have the 6800XT, which is not well suited to this use case)

Automatic11111 on MachineML is dog slow, but I got some interesting results. So today I cleared out an old SSD, wired it up and installed a clean ubuntu. Following this guide I managed to get rocM running and the difference is like chalk and cheese. Or rather Impulse and Warp drive. Totally crazy!

So for you AMD Radeon users out there. There is hope. It is in Linux, but it is there.

Random illustration for a story idea; the story idea may not pan out, but I was proud of how the art came out!

Steps: 40, Sampler: Euler a, CFG scale: 10, Seed: 900795974, Size: 1536x1536, Model hash: 25ba966c5d, Model: aZovyaRPGArtistTools_v3, Denoising strength: 0.3, Clip skip: 2, Token merging ratio: 0.5, Ultimate SD upscale upscaler: 4x-AnimeSharp, Ultimate SD upscale tile_width: 512, Ultimate SD upscale tile_height: 512, Ultimate SD upscale mask_blur: 8, Ultimate SD upscale padding: 32, Version: v1.3.2

I hope it’s the open release of SDXL - the beta on the Stable Diffusion discord is getting pretty impressive.

In any case, I’d like a tea serving drone 😁.

I made a large scale working QR Code scene that points to kbin.social. I used the new QR Code Control for SD 1.5 (released here, and multiple rounds of upscale in img2img using it, controlnet tile, and the t2ia color control, starting from a QR code I made with this QR code generator.