@Background Pony #5F90
If only “pinkie pie” was changed to a different pony, with everything else the same, would you get the same pose? Like the rest of the main six like this. 🙂
@Background Pony #5F90
Maybe I could learn to incorporate this tool into my drawing process somehow, such as creating rough concepts and whatnot to save time and develop ideas. But for now I just need to process my thoughts and feelings about the whole thing.
Though personally I think it’s still not quite up to the standards of human artists, at least not without help. The real power of Stable Diffusion (and similar image generation models) is in churning out lots of reasonable quality stuff fast.
I did check your Twitter previously but didn’t see any additional info in the tweet featuring this image. After digging through some of your previous posts though now I see what you’re referring to.
Custom model fine tuned on a rig with 8 RTX A6000s, presumably using an image dump from Derpibooru. Cool stuff.
Wait, this was done with Stable Diffusion? Wow. Now that I look closely I can see the artifacts, but upon seeing this my first though was “that’s a really unique style”. Didn’t even cross my mind that it might have been done by AI.
I have so many questions. Was this just done with a prompt given to the vanilla Stable Diffusion 1.4 model? Or did you need to use a custom tuned model to get these results? Or perhaps textual inversion? Do you mind sharing the prompt?
@Background Pony #5F90
If only “pinkie pie” was changed to a different pony, with everything else the same, would you get the same pose? Like the rest of the main six like this. 🙂
Maybe I could learn to incorporate this tool into my drawing process somehow, such as creating rough concepts and whatnot to save time and develop ideas. But for now I just need to process my thoughts and feelings about the whole thing.
Your car is faster than you, and a plane can fly. And yet, they are just machines.
Imported from Derpibooru - Posted by AstraliteHeart
@Background Pony #5F90
I’m aware. I just didn’t want to believe it at first. Kudos on training this software.
Most days I would spend hours tearing my hair out to attempt to draw something close to this level.
Maybe I could learn to incorporate this tool into my drawing process somehow, such as creating rough concepts and whatnot to save time and develop ideas. But for now I just need to process my thoughts and feelings about the whole thing.
@Background Pony #5F90
This HAS to be an artist’s re-rendition of that previous ai generated wedding pinkie. It HAS to be. If it isn’t and it is ai then I’ll just smash in my drawing tablet cuz what’s the point now??
It is not a re-rendition, this is as real AI as it gets - not a collage of others work, not a copy-paste, but a computer learning to draw. Please do not smash your drawing tablet, you have been given a new shiny tool to play with, embrace the possibilities.
Imported from Derpibooru - Posted by AstraliteHeart
How about now
https://i.imgur.com/nLvkneQ.jpg
(fixed left eye)
I’m guessing that text was the prompt passed into the AI to generate this image.
Because there’s no actual artistry involved, there’s only pretty aesthetics.
Why does that suck?
If only “pinkie pie” was changed to a different pony, with everything else the same, would you get the same pose? Like the rest of the main six like this. 🙂
Wut
pinkie pie, sugestive, anthro, full body, portrait, wedding dress, veil, intricate, highly detailed, digital painting, artstation, concept art, smooth, sharp focus, illustration, Unreal Engine 5, 8K, art by artgerm and greg rutkowski and alphonse mucha
https://en.wikipedia.org/wiki/Pulled_rickshaw
False equivalence but ok.
I’m aware. I just didn’t want to believe it at first. Kudos on training this software.
Most days I would spend hours tearing my hair out to attempt to draw something close to this level.