- Joined
- Aug 2, 2021
Stable Diffusion, NovelAI, Machine Learning Art - AI art generation discussion and image dump
- Thread starter Near
- Start date
-
🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
- Joined
- Sep 16, 2019
If you have the GPU grunt, that plugin will probably do what you want but you'll probably need to mess with ControlNet and Outpainting. If you don't, AI Horde is a crowd sourced way of getting Stable Diffusion running. I'm not sure of a good alternative to Stable Diffusion for Img2Img or Inpainting, so if anyone has a better alternative, you should probably listen to them.
- Joined
- Aug 2, 2021
Actually been playing around with the Krita plugin and it just shows me how out of touch I am with the AI stuff.
The models seem to be working though.
There any good ai setup/ai tutorial videos?
- Joined
- Sep 16, 2019
This video and the channel was good for going over the basics for the plugin and some guides. I learned mostly by just poking around and trying things out, so I don't really have any good video guides offhand. I do have some guides on Rentry. They may a bit outdated, but they've got a ton of info
- Joined
- Mar 11, 2024
maybe i'm missing something but how is this A.I related?
- Joined
- Feb 28, 2015
mister fireworks
kiwifarms.net
- Joined
- Oct 2, 2022
Does anybody have any advice for a tech retard (me) on how to instruct the ai bot to make simple images without a bunch of dumb bullshit added to it?
for example, i wanted the bing ai to make the seinfeld logo but using another word. Being a useless luddite, i typed in "seinfeld logo using the word "stanston". It gave me 4 images with the familiar yellow/red logo using the word, but it added a bunch of weird garbage graphics around it. Of course after screaming at it to remove the other elements and JUST THE LOGO ON BLACK BACKGROUND I ended up with something even further away from the first batch. Bing also makes me start over after 5 tries so I was wondering if anyone had any advice on how to make my commands better/more specific and if theres any free ai program that is better than bing?
I also tried to use Dall-e to make an amazon style wage cage and it absolutely refused to put a human in a cage next to a robot. The could be in there together or the robot but not the human. It even shut me down completely when i asked them to make a rendering inspired by the amazon patent. I cant tell if this is just me being horrible at prompts or if its not even worth trying to get around the rules. Does anyone have a recommendation for a free, maybe newer/non pozzed program that isnt muzzled to prevent you from making fairly tame stuff like a seinfeld logo and a wage cage?
for example, i wanted the bing ai to make the seinfeld logo but using another word. Being a useless luddite, i typed in "seinfeld logo using the word "stanston". It gave me 4 images with the familiar yellow/red logo using the word, but it added a bunch of weird garbage graphics around it. Of course after screaming at it to remove the other elements and JUST THE LOGO ON BLACK BACKGROUND I ended up with something even further away from the first batch. Bing also makes me start over after 5 tries so I was wondering if anyone had any advice on how to make my commands better/more specific and if theres any free ai program that is better than bing?
I also tried to use Dall-e to make an amazon style wage cage and it absolutely refused to put a human in a cage next to a robot. The could be in there together or the robot but not the human. It even shut me down completely when i asked them to make a rendering inspired by the amazon patent. I cant tell if this is just me being horrible at prompts or if its not even worth trying to get around the rules. Does anyone have a recommendation for a free, maybe newer/non pozzed program that isnt muzzled to prevent you from making fairly tame stuff like a seinfeld logo and a wage cage?
- Joined
- Dec 28, 2014
It's AI generated. glif.app is an AI meme generator. But the only options to use it involve signing in on Google or Discord. Lmao no thanks. Someone do a Virgin vs. Chad on Kiwi Farms vs. Reddit and then another on Reddit vs. Kiwi Farms.
It's actually better and funnier than the shit talk we get from actual humans, well at least semi-humans.
It even got me mad for a couple seconds before I figured out what it was. Eerie.
Soggy Floppa
kiwifarms.net
- Joined
- Nov 4, 2023
- Joined
- Mar 11, 2024
so not only is A.I gonna steal our jobs but it'll take over our shitposting as well.
- Joined
- Dec 9, 2017
I somehow missed this a few weeks ago, but it seems as if Forge is going to become an experimental repo. If you actively use Forge, you might want to rollback to commit 29be1da, or make a backup of Forge. For those uninitiated, Forge was a fork of Automatic1111’s offering which provided a bunch of “essential” extensions prebuilt and some tweaks to the backend to make image generation more efficient. The A1111 dev branch, allegedly, has implemented these tweaks. I cannot confirm this presently as I haven’t used standard A1111 in some time, but I actively use Forge alongside ComfyUI and there was noticeable difference in generation times between Forge and A1111 on my 4090.
I do intend on making a clean install and giving the dev branch of A1111 a go tonight and will report back.
People in the comments seem pretty upset over this whole thing.
I do intend on making a clean install and giving the dev branch of A1111 a go tonight and will report back.
People in the comments seem pretty upset over this whole thing.
- Joined
- Dec 12, 2022
Having tried the dev branch, I can't say it really matches in speed. As a simple example, if you try a txt2img + non-latent upscale, A1111 will hang for a long time on starting up the upscale for some reasons. Forge only stops for 2-3 seconds at most on that part.
Maybe it's a match for iterations per seconds, but that's not helpful if it decides to hang for 25+ seconds in parts of the workflow.
- Joined
- Mar 17, 2022
On the topic of Forge, I tried it out and it runs better than (the very, very outdated pull of) A1111 on my machine, but I can't switch models on Forge without my RAM / VRAM getting maxed out on the next post-switch generation and my entire machine locking up, whereas A1111 seems to handle RAM management when model-switching much better. Has anyone else had this issue?
- Joined
- Mar 29, 2014
Seems like almost every online AI tool or generation site has "sign in with Google" (never works), is nonfunctional, has limited free use, or some combination of such BS.
Last edited:
- Joined
- Dec 9, 2017
When switching? Cannot say that I have and I cannot replicate the issue. I’ve tried to load some SD 1.5 and SDXL checkpoints to see if I could log anything weird with VRAM allocation, but it’s loading/unloading properly on my end.
- Joined
- Dec 9, 2017
Forgive the double-post. This isn’t really news, but I did find it hilarious. CivitAI has declared that they are interested in hosting in-person gatherings and events. DashCon 2.0, anyone?
Also, keep in mind that CivitAI is largely populated by coomers. Don’t really see this being anything other than a disaster, but I’ll watch intently and excitedly.
Also, keep in mind that CivitAI is largely populated by coomers. Don’t really see this being anything other than a disaster, but I’ll watch intently and excitedly.
- Joined
- Mar 11, 2024
We need boots on the ground for this if this happens.
Imagine going to listen to a panel and the guy on stage starts talking about how he made a maggot porn loRA.
- Joined
- Dec 9, 2017
“Well, you see… first… I was inspired by the Botfly Girl story from a long time ago. Then I took hundreds of pictures of maggots in decaying animal carcasses, and scoured the Internet for them. Next, I looked at each picture and created an individual text file corresponding to each picture and described them in as much explicit detail as I could….”