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

A giant skeleton eating cows

OIG.jpeg OIG1.jpeg

wtf.jpeg
 
Sweet baby Jesus. I believe I've seen that cover before, can't quite put my finger on it.
It looks like the cover for the old Jeanne d'Arc game.

Sadly this thing refuses to make my avatar into anything sucks to use a character with the name genocide in it....

It has no issues making demonic disney characters though.
1696477688196.jpeg
 
It tried to recreate the Source Engine style without being prompted to do so, that's nuts
People were having stable diffusion draw anime characters in styles from other shows.
This doesn't work as well as current anime checkpoints but it still knows.

cirno.jpgcirno-bebop-1.jpg
Cirno from Touhou in the style of Cowboy Bebop


motoko3.jpgmotoko1.jpg
Motoko Kusanagi in pokemon art style


ash2.jpgash1.jpg
Ask Ketchum in Ghost in the Shell art style

Sometimes it doesn't work well.

ryuko-matoi.jpgryuko-matoi-2.jpg
Ryuuko Matoi. One Piece art style

batman2.jpgbatman1.jpg
Batman in One Piece art style

This was kind of funny.
batman3.jpg
 
This was my inspiration.
IMG_2183.jpeg
I used image recognition in ChatGPT to generate a prompt based off of this image. It wrote:
Generate a minimal graphic design poster with a blend of vaporwave and cyberpunk aesthetics. Feature holographic liquid patterns combined with geometric grids, and incorporate neon-lit Japanese characters. The colour palette should dominate in neon purples, blues, and pinks. Include futuristic fonts and a barcode. Ensure the output is digital art, not a physical representation.
And generated these. Pretty impressive.

IMG_2447.jpeg
IMG_2446.jpeg
IMG_2444.jpeg
 
I've asked it boring shit like "a birthing person" and wasn't impressed with the multiculturalism generic stuff or phrases or terms you may hear like "raining cats & dogs"
i didn't find the images produced to be very interesting.
However I Really like the differences you can achieve simply by adding or removing a word or changing the meaning just by adding in overt/subtle or asking for emphasis on certain elements you like, I achieved a lot of these just by swapping hallucinatory with psychotomimetic.

"creation of a universe in a hallucinatory style with emphasis on colors and mind bending themes"
_763195f2-f646-42b9-a49b-8544ff02f276.jpeg

"creation of a multiverse in a psychotomimetic style with emphasis mind bending creations"
_4c88e42c-7ef8-4fca-8478-1ddcc2532048.jpeg _613cd52a-d6f3-4315-aa73-03d447208152.jpeg _660dc20e-38a2-4b45-9448-c061b585e0de.jpeg

"creation of multiverses in a psychotomimetic style with emphasis mind bending creations and horror psychedelic explosions"
_dc9a4b1f-280b-4223-ba83-a10e47809a28.jpeg _8486ed61-dd81-4143-af64-01ed4a61b751.jpeg _6e07284b-352d-49c3-9b1f-5c0dfd7f823e.jpeg _0c60e8fd-1307-47c5-99f6-5034c7a99505.jpeg

"a humanoid creature with mind expanding psychedelic explosions of personalities making more planets in a psychotomimetic style with emphasis mind bending creations with subtle horror elements and overtly cute elements mixed"
_582f0c36-f42b-453e-b2ed-641b38f105e3.jpeg _4aa93dfe-2bb3-4394-80c2-a8c466413bb6.jpeg _2995a406-9011-4708-91a6-d008ec781f51.jpeg

"a godly higher being with mind expanding psychedelic explosions of personalities that are making planets in a psychotomimetic style with emphasis mind creations with overtly subtle horror elements and super cute elements mixed"
_23832420-69a8-49b3-ba5b-f7a774007142.jpeg _187492bb-4b7b-4107-973f-6a5511f58942.jpeg _6f88e4df-7553-456d-b795-ee1fe91c1a60.jpeg

I just really enjoy the psychedelic art style it creates, it can also do horror quite well
 
I read earlier in the thread there might already be a torrent for the dalle 3 model available? if so can someone point me in the right direction to find it? Currently use SDNext with SDXL and a few other models but would be more than willing to jump os's / systems to have that locally.
 
  • Optimistic
Reactions: Roland TB-303
  • Winner
Reactions: Nottafed
Back