Stable Diffusion, NovelAI, Machine Learning Art - AI art generation discussion and image dump

now, a very noob-y question. how to set up sdxl to work on the a111 ui? i did hear that the support wasn't the best?
 
What if Eggman was the hero honestly trying to build a utopia with superior technology, and Sonic and his friends were a pack of hyper-violent maniacs trying to tear it all down?

01 - PsychoSonk.png
02 - GunmanMiles.png
03 - SpikeyKnux.png
04 - PunkAmy.png
05 - HeroEggman.png
 
  • Like
  • Autistic
Reactions: ExsanguinateHorizon, cybertoaster, Vecr and 3 others
MrJokerRager said:
Here is a link to his executive order.


They have been planning this in the works for the past 10 years at least as there are articles to make AI "anti racist" as far as 2013.

It seems that since a lot of these companies like Microsoft get federal subsidies, that is how they will control AI.
Click to expand...
I bet making our AI a faggot is going to give China an advantage. I desperately hope someone out there takes the leash off these things.
 
  • Agree
Reactions: MrJokerRager and Drain Todger
Time 4 Guillotines said:
I bet making our AI a faggot is going to give China an advantage. I desperately hope someone out there takes the leash off these things.
Click to expand...
There was already some company planning on building a floating data center in international waters to get around this, so we can have fully Based AI.
 
  • Optimistic
Reactions: inception_state, Time 4 Guillotines and JustSomeDong
Drain Todger said:
There was already some company planning on building a floating data center in international waters to get around this, so we can have fully Based AI.
Click to expand...
I doubt "we" will get anything out of it. They're not gonna do anything out of altruism or for the public good, and I highly doubt they're talking about it for John Q. Shitposter to have access.
 
  • Agree
  • Feels
Reactions: anustart76, Time 4 Guillotines and Drain Todger
Drain Todger said:
There was already some company planning on building a floating data center in international waters to get around this, so we can have fully Based AI.
Click to expand...
It's an interesting concept but the company doesn't seem real. Also might be a terrible idea to put millions of dollars of equipment out in international waters. If Somalians don't raid it China will.
Edit: Got curious about the structure of OpenAI Inc. and who founded it. I'm late but apparently it's Sam Altman, the gay jewish former head of Reddit. All the faggy little rules make more sense in hindsight. Cancelling my subscription and will no longer be using ChatGPT since every interaction is going to help that company
 
Last edited:
  • Winner
  • Agree
Reactions: Jimjamflimflam and Drain Todger
woowie queen said:
now, a very noob-y question. how to set up sdxl to work on the a111 ui? i did hear that the support wasn't the best?
Click to expand...
Support is pretty good now with recent updates. All that's needed is picking a SDXL-based checkpoint from the dropdown list,and set your resolution to eg. 1024x1024, 832x1216, or 768x1280.

You will probably also need to reduce your batch size to avoid running out of VRAM, particularly when using the hirez fix tools.

Note that LoRAs and such for SD1.5 won't work for SDXL, so don't be surprised that a bunch of your LoRAs "disappeared". The A1111 UI hides them when you switch to a SDXL checkpoint because they are incompatible, it only shows LoRAs that are compatible with your currently loaded model.
 
  • Like
  • Informative
Reactions: woowie queen and Mr.Miyagi
woowie queen said:
now, a very noob-y question. how to set up sdxl to work on the a111 ui? i did hear that the support wasn't the best?
Click to expand...
You can also switch to SD.Next, or install it beside A1111. Slight differences but it should look similar enough, supposedly better SDXL support. You should also be able to get both of them to share the same checkpoint directory since they're HUGE and you won't want to have 2 copies if you want to compare A1111 and SD.Next and ComfyUI and whatever else is out there.

 
  • Informative
Reactions: woowie queen and Mr.Miyagi
Drain Todger said:
What if Eggman was the hero honestly trying to build a utopia with superior technology, and Sonic and his friends were a pack of hyper-violent maniacs trying to tear it all down?
Click to expand...
OT, but this reminds me of tool-assisted speedruns of the original Mega Man on the NES
Essentially, the game is glitched so fucking hard that the entirety of reality collapses
So you might argue that Dr. Wily was merely trying to save the world by fighting against Mega Man
 
  • Like
Reactions: Drain Todger
davids877 said:
You can also switch to SD.Next, or install it beside A1111. Slight differences but it should look similar enough, supposedly better SDXL support. You should also be able to get both of them to share the same checkpoint directory since they're HUGE and you won't want to have 2 copies if you want to compare A1111 and SD.Next and ComfyUI and whatever else is out there.

Click to expand...
i did use sd.next in the past, but i wasn't a fan of the theme it comes with on default. might give it another chance. just gotta figure, since on sd.next, sdxl works only as diffuser, what model i need to download (and if the refiners need to be with the main model.) same with the control lora, it goes with all the other loras?
 
Wonton Tony said:
I am trying to setup using this https://rentry.org/voldy guide and am wondering before i do if this is outdated and if so what new models and or setup do i need to do to be up to date?
Click to expand...
It's a bit out of date, you can basically ignore anything about SD1.4 or NovelAI. SDXL-based models are currently the best available open source options, particularly for photorealism. SD1.5 has lower system requirements and a lot more LoRAs (fine-tuned models for specific things).

Go here to install the most popular web UI (it has installation instructions). Then head over to CivitAI, sort by popular / highest rated to get started, and have fun! CivitAI is also a great resource for "borrowing" prompts.
 
  • Informative
Reactions: woowie queen, Digital Snow and Wonton Tony
you can get as big a context window on novel.ai as someone paying $15/month by using tor and creating a new circuit when your 50 free generations run out
it takes like seven seconds to export your story, clear the site's cookies and import it
if it starts blacklisting your story's ID you can just go into notepad and delete the JSON field labeled "id"
 
  • Informative
Reactions: Mr.Miyagi
I have a question to people that know how to use A1111's WebUI/SD.Next more than I do.

How the fuck do you use it to take something, like let's say my avatar, put it through img2img and make it enhance the detail of it without changing the shape of it?

Is there a specific ControlNet workflow setup that can do that? Do you just apply a LoRA after that or do you need a specific prompt?

And speaking of ControlNet, how do you set it up to do the funny hidden message images? This is another one of those things that I have zero idea about while I still keep bumblefucking with prompts and generation parameters over a single thing over and over again waiting a minute per image only to realize it's not doing shit to make it look better just as I did back when Stable Diffusion first became big.
 
You must log in or register to reply here.
Back