deleted by creator

This is the prompt:

A {cyberpunk|dieselpunk|steampunk|solarpunk} wallpaper, black and {violet|crimson|blue|dark green|gray} colors, evening {landscape|cityscape|towers} Surreal Cubist Expressionism A distorted, fragmented figure in a red and orange hue, its features melting and shifting like wax, swirls in a vortex of swirling purple and green patterns. The background is a deep, blood-red, with stark white accents and splashes of primary colors. In the foreground, jagged shards of glass and shattered metal protrude from the canvas, as

The first part was written by hand, the part starting Surreal Cubist Expressionism was created by Llama3. Model is Stable Diffusion XL, more specifically AlbedoBase XL.

Well, sometimes you don’t want to do. But yeah, overall you’re right.

ChatGPT uses Dall-E, which wouldn’t be my first choice. Its only advantage over Stable Diffusion is that you can use natural human language. But learning to prompt Stable Diffusion is not that hard.

Edit: And Flux beats both Dall-E and Stable Diffusion. And you can also use natural language with that, if I recall correctly.

I’ve seen some “normal” looking person made by AI, though it’s definitely a minority of the images.

One of my favourites that’s not known very much is Meg Myers. She puts raw emotion into her music.

Well, yes. Not sure what did I miss?

Same! I never understood the criticism. Yeah, all of his dialogues are awkward. But they perfectly fit his character.

When you order a chosen one and a Chewbacca from Wish.

Does someone consider Jedi the good guys? I found them awful since I was a kid.

I was seriously waiting for “So uncivilized”.

Sorry for the late reply! I couldn’t check when I noticed the comment and then I forgot. The model is AlbedoBase XL (SDXL).

Here are the full generation parameters:

Prompt: artistic, abstract, lines, light colors

Negative prompt: ugly, deformed, dark

Model: AlbedoBase XL (SDXL)

Sampler: k_dpmpp_sde

Karras: Yes

CFG scale: 7

Size: 1024x512 px

Steps: 30

CLIP skip: 1

If you don’t want to start with local software, I can recommend AI Horde (for example through Horde NG) which is a service where you get free access to a cluster of volunteer Stable Diffusion (and Stable Cascade) workers.

You can use those models commercially and all of them can be used locally. The models are owned by those who created them, if you’re asking about a model you modify and train yourself, then it really depends on the original license, but IIRC for Stable Diffusion, you own the model, but you also have to license it under the same conditions as the original model.

Who owns the generated image is much less clear and really depends on the country. For example in my jurisdiction such an image is not a copyrightable work, meaning no one owns it.

Anyway, using AI Horde is a good start because it does not require that you learn how to use the models locally but at the same time it’s not a watered down service like most AI services are. And it’s completely free.

Not sure what you mean? That I posted it to 3 separate communities?

deleted by creator

I probably forgot to turn off the face fixer, it’s an additional model that runs after the image has been generated and tries to make faces look more realistic.

I mean, it was obviously a mistake, not a spam.

It’s an old image, but I suspect that is the case, I often forget to turn off the facefixer models.

Used the word “kakáč” (shitter) instead of “kartáč” (brush).

Stop projecting, maybe? No, I’m not forced. But I want to, because I refuse to pay any more money to such a shitty company. I paid for the product, now I’m gonna use it the way I want to.

Price is not my reason for pirating. Seriously, people are different from you, stop assuming that everyone has the same motivation as you do.