Minigunny

>>17478 >nurse outfit That's something I haven't genned in forever.

>>17480 same tbh, I just wanted to for some reason

Tried a thing

>>17482 turned out pretty good all things considered. nice job!

>>17483 Thanks, definitely expected it to be a lot worse since it was just controlnet and using the same seed/prompt. Even so it's funny how massively the style can change. Here's the source template https://files.catbox.moe/mbfwoz.7z

>>17484 oh cool! I might mess around with it, if I have some time. I still got a lot I want to add to the UI, so that's still most of my free time lel.

uoh

>>17486 uoh 2

https://twitter.com/MAiJiNTHEARTIST/status/1665800119021862912 Yakuza won

>>17489 good luck to the (((yakuza))), i'm behind 3928 proxies

>>17489 Every time someone links twatter I fucking know I'm up for the most abhorrent and stupid shit ever and I still am surprised I wonder if the troons are going to shame nvidia for cuda cores any time soon. Allowing local use of AI is raycis n shiet ya know

>>17491 >raycis We can use rays for our gens now? I can finally put those raytracing capabilities to good use

ABSOLUTELY TALMUDIC

>>17491 >>17493 kikes and trannies need to be hung, good thing more people are starting to push back against trannies now so I'm hoping more of them are hung in the future to spare us from this retardation

>>17491 >shame nvidia for cuda cores It's gonna be so fucking funny, not even tech trannies understand that CUDA cores ARE the GPU.

Built in openpose editor for the official controlnet addon that lets you edit the preprocessor output: https://github.com/Mikubill/sd-webui-controlnet/discussions/1548 Could be cool for doing quick pose/hand/face fixes I guess.

>>17489 wow! im a wanted felon! (in Japan)

>>17489 So this does this shit have legal power already or not

>>17498 >>17499 Could just get companies like pixiv to pussy out which would be annoying because then I would have to find a loli friendly platform

>>17500 ugh reading the way this faggot types hurts my head but basically it's common sense copyright law, niggers can sue you if you do AI outputs that are not derivative of existing works and promoting it in a commercial sense, pixiv has already banned that and they've gone further to target accounts uploading derivative works with style/artist LORAs so it's not such a huge hit on that front.

hate this faggot but it seems like the recent Jewvidia driver updates are causing slowdowns when all the VRAM is being used because to avoid OOM errors it just goes into sharing memory with your PC's RAM https://github.com/vladmandic/automatic/discussions/1285

>>17499 >>17498 Isn't it just for COMMERCIAL use? Like if you're making shit for yourself then cool but if you make AI art that infringes on someone's IP and try to sell it then they can sue?

>>17504 Some people here seemed to do commission money. I personally do this for fun/fap so yeah it doesn't affect me yet. Would suck to see AI completely banned from everywhere because of angry niggers from twatter

>>17505 >because of angry niggers from twatter oh you sweet summer child

>>17497 Very happy with this. I was using third party extensions and while they were fine, the openpose hand implementation sucked.

redid teto. synthv outfit is pretty gatcha, not a whole lot of data yet.

>>17509 >redid teto. YEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEES >synthv outfit is pretty gatcha ...oh :(

>>17510 also thanks for reminding me i need to deactivate her and synthv before i wipe down the ssd and move it to a new build

Any tips for merging models? I kind of want to merge the chewy marshmallow model above with AOM3 or something else. Or has anyone done model mixes with it so I don't have to?

>>17510 it just tends to flub the details more than half the time and has no idea how to recreate the striped "staff" pattern on her uniform. i think a lot of the posts weren't tagged properly looking them over again.

>>17502 That would explain the slowdowns I've been having lately. Funny how just a week ago they promised 2x Stable Diffusion performance https://blogs.nvidia.com/blog/2023/05/23/microsoft-build-nvidia-ai-windows-rtx/

>>17514 Did anyone even benefit from that?

Might want to get to baking that new thread

>>17515 The "speedup" requires converting pytorch models to onnx. You can convert existing models but I haven't bothered to try because it's also not compatible with lora https://devblogs.microsoft.com/directx/optimize-directml-performance-with-olive/