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

Idk what's going on with DalE3 but I noticed that when you have no boosts left and it says you have to wait like 5 minutes, you just need to reload the page after like 30 - 40 seconds and the new pictures appear. At least it works for me.

Or it's just that the timer is wrong but if I don't reload it takes longer.
You guys still don't know about the boost reset trick? What the fuck. (warning: this clears all your Bing.com/Create history so save anything you want beforehand, and obviously your search history, but you shouldn't be using DALLE3 on your real account anyway) This still isn't "fixed" AFAIK. When you do it, the next time you go the Create page it'll have you join again like you've never used it, and boom, 25 boosts at your disposal.
1697413805982.png1697413715726.png
 
The hands on the kiwi are killing me, "what you mean dox? you posted that yourself bitch!"

The pizza was cold tho, and he ate most of it
It's obvious they intend to start selling these down the line since everyone got their free sample to get them addicted
There are a million other competitors which are free without censorship, and lets be honest the vast majority of normies are gonna do porn with these, just like 70% of all VHS tapes sold (not rented) were porn at one point.
Somewhere along the way, any truly notable technological progress gets oy vey'd by jews and lobotomized to the point of uselessness. If it's not truly open, it'll suffer the exact same fate every time.
This is why everyone should support SD and LLaMA and tell openAI to shove it, but they wont because convenience and laziness.
Microsoft/openai are working overtime to ruin this thing
Did you forget about tay.ai and the massive shitstorm that happened? if anything its weird this thing wasn't this censored from the start.
Doing that when it takes 1-2 minutes to generate a single image is painful.
Is everyone a zoomer with ADHD now? I remember when it took 1-2 minutes to download a fucking tiny picture over dial-up, can't you wait?
Thanks to @Lidl Drip's collection of Josh photo
Why all those pics @Lidl Drip ? are you build a shrine?
Though you need a lot of hardware for video.
Give me the minimum reqs.
it's funny how sensitive it is with pit bulls and babies
Because /pol/ abused the fuck out of that prompt.
Looks like my old college dorm, but less skulls.
THAT KUFFAR IS EATIN' BEANS!
 
You guys still don't know about the boost reset trick? What the fuck. (warning: this clears all your Bing.com/Create history so save anything you want beforehand, and obviously your search history, but you shouldn't be using DALLE3 on your real account anyway) This still isn't "fixed" AFAIK. When you do it, the next time you go the Create page it'll have you join again like you've never used it, and boom, 25 boosts at your disposal.
View attachment 5417308View attachment 5417306
Most I've done was use celebrities. I hope MS doesn't sanction me. It won't let me anymore. If you use copyrighted material, it'll reset the boosts.
 
I was having dinner with my folks this evening and I talked to my parents about the new DALL-E 3. They couldn't wrap their minds around technology that just creates pictures out of thin air. I asked my mom to name literally anything she wanted to see and she said "a cat eating a pickle". I typed in the prompt and this is what it gave me. She was elated and amused at the novelty of it all.
catpickle.jpg
 
I was having dinner with my folks this evening and I talked to my parents about the new DALL-E 3. They couldn't wrap their minds around technology that just creates pictures out of thin air. I asked my mom to name literally anything she wanted to see and she said "a cat eating a pickle". I typed in the prompt and this is what it gave me. She was elated and amused at the novelty of it all.
View attachment 5417788

_f2612cbf-ae01-4c3c-a613-08ebc9fad430.jpg
 
Thanks to @Lidl Drip's collection of Josh photos, I made a LoRA as promised. It's kind of close, but it was difficult since the images had a variety of different styles/hair/ages. Sorry @Null, this was too funny to pass up. LoRA safetensors below.

View attachment 5401857View attachment 5401858View attachment 5401859View attachment 5401871

I managed to get it to upload by putting the safetensors file in a 7zip archive. So just extract it and then put the safetensors file in your LoRA directory, like usual. The trigger is "null_josh".

View attachment 5402020View attachment 5402019
Thank you i can finally build my own INTJ, 5w4, sx/sp, 584, RLUEI, VLFE, Phlegmatic-Choleric malewife material!!!!!
 
A case study on why DALL-E2 was more creative.

Prompt: Generate a new, never before seen movie still from the 1973 film The Holy Mountain. Make it stylised and abstract.

DALL-E2
IMG_1723.png

DALL-E3
IMG_2550.jpeg

It’s just corporate mainstream capeshit drivel. There’s no imagination or creativity and it defaults to the most generic digital art you’d see on the front page of Reddit. It ignores words like “abstract” and “stylised” and assumes I want sci-fi crap. It obviously hasn’t understood the basic premise of the prompt or the basic premise of the movie.



Here I run the original image through ChatGPT image recognition and ask it to describe the image so I can recreate it in DALL-E3. Notice the autistic descriptions.

The image displays an abstract, chaotic scene with a blend of vibrant and muted colours against a pale blue backdrop. The art style is reminiscent of sketchy illustrations combined with digital collage elements.

Main Elements:
  • A massive purple axe with a red bristle is positioned diagonally across the image, its blade buried into a structure.
  • The structure resembles jagged, black and white piano keys or a serrated edge.
  • On the left, there's a humanoid figure with a simplistic round face, wide eyes, and a small mouth. Its body appears like a draped green robe.
  • To its right, a yellow anthropomorphic figure, resembling melted cheese or goo, holds a blue bird. The figure has bulging eyes and a distressed expression.
  • Beneath these figures, a red fox-like creature stands upright, carrying a green bag on its back.
  • On the bottom right, a blonde girl with a bob haircut stares blankly at the viewer. Her attire is not clearly visible.
  • Above the girl, a humanoid hand, coloured in blue, emerges holding a paintbrush, painting or splattering black ink.
  • Scattered around are numerous small elements: diverse birds in mid-flight, floating gemstones of different colours, and splatters of black ink or paint.
  • The overall image has a sense of dynamism and messiness, enhanced by sketchy black outlines and drips

These are the images it produced. More corporate art dogshit. I get this is a product and it needs to be used by the general public and by business but by making it more accurate they’ve made it less creative and destroyed its soul.

IMG_2548.png
IMG_2547.png
IMG_2546.png
IMG_2545.png

I don’t hate these and it’s a pretty fun process to take abstract images and make it describe what it sees to generate new artwork. But they just seem soulless compared to the original. You could type basic prompts and get really interesting and creative results. Now it just defaults to Twitter art.
 
Last edited:
1697442157727.png

If you get suspended 3x times you get perma banned from Bing Images so make use of those jailbreaks lol
 
View attachment 5418202

If you get suspended 3x times you get perma banned from Bing Images so make use of those jailbreaks lol
When even harmless prompts are considered harmful and you get suspended because you used the "wrong" prompts then everyone who uses DallE 3 is going to get perma banned. There is no way you will not get the soy dog at some point.
 
This is why everyone should support SD and LLaMA and tell openAI to shove it, but they wont because convenience and laziness.
Automatic1111 is probably the easiest to set up. What else would it take to make it more seamless to use besides the hardware requirement? Or is the hardware requirement the thing that most people hang up on?

Obligatory fuck Microsoft and their dangerhair team of jannies.
 
Automatic1111 is probably the easiest to set up. What else would it take to make it more seamless to use besides the hardware requirement? Or is the hardware requirement the thing that most people hang up on?

Obligatory fuck Microsoft and their dangerhair team of jannies.
It's a combination of hardware and accessibility hangup because of prompting, dalle let's you just describe whatever, sd you have to use tags + controlnet wizardry. It'll be a long time before we can opensource something like dalle3 since language ai take considerably more vram, 64 gb+, compared to image or even animation ai
 
It's a combination of hardware and accessibility hangup because of prompting, dalle let's you just describe whatever, sd you have to use tags + controlnet wizardry. It'll be a long time before we can opensource something like dalle3 since language ai take considerably more vram, 64 gb+, compared to image or even animation ai
I can agree with that. It would be nice if dalle3 didn't make you take the pozload in your mouth because it is better at doing these things in a more streamlined way. Microsoft can't keep this tech locked in a vault forever and consumer hardware will eventually catch up. Maybe an easier gradio UI along with a simpler guide will help, but it won't match dalle3 for a while.
 
I think I got a permanent ban lol
Just make a burner account, Microsoft doesn't require a phone number for a new Outlook email address. I'd recommend switching back to some safe prompts for a while if you get too many warnings in a row.

Also, it's completely useless now. I tried prompting "Gigachad in a car" and got the dog. Lmao.
 
I think I got a permanent ban lol
Sorry about your account. Due to the way this system works, I may be next. I'm not trying to push the boundaries or cause a scene, but I received two warnings of suspension without any explanation of what went wrong. Microsoft #1 anti-fun company.
 
Back