@Background Pony #6541
Style has no exclusive rights under IP law. Artist tags on derpi represent copyright. Even if the style was identical an artist tag would be incorrect.

Also, this art looks nothing like greg rutkowski’s or alphonse mucha’s work.

Stable Diffusion can’t actually reference any images at all (unless you do img2img, which this is not). The only things it has access to at generation time are the trained weights, the CLIP embeddings (prompt), and the starting image (gaussian noise in this case). Prompting with artists names in most cases just nudges the CLIP embeddings towards “good art”.

@Background Pony #6541
The effect it has can be very far abstracted from anything you’d normally call an artist’s style. For example, if an artist tends to compose pictures with backlighting, using them in a prompt may influence it to make images composed that way. Likewise, including an artist who works mostly in grayscale might merely lead to a result being grayscale. It’s hardly clear-cut what you’ll get.

Besides, they did list what they wrote as the prompt, which could be thought of as the credit you’re expecting.

@Background Pony #6541
thats disingenuous. By typing an artists name, you prompt the AI to reference and copy the style of said artists.

She looks adorable

This looks really nice.

Dude, Fluttershy in hooded trenchcoat looks good. <3

@Background Pony #6541
That’s a trick question. Those artists never did draw ponies or similar, also, it’s an AI text-to-image generation, so basically all that line told to AI is to try mimicking their style.

Don’t you feel that greg rutkowski and alphonse mucha require artist credits seeing how you prompted their gallerys to assemble this?