Artists have a style and the artistic merit allows them to keep creating good art, but an AI can be fleeting. Like it’s been dreamed up.
Most professional prompters sooner or later settle on just one or two image styles. Just look at their galleries, for example: kalmar, FoylerTF. Isn’t it true that the style of the images is similar? I often find out who generated it simply by the style of the posted AI image.
I’m still trying out different styles, searching the one that’s most convenient for me, but in general, I can already distinguish two main styles of the images I generated (this style is not included in them, it’s just an experiment).
Rarely do I find a style from an AI that I really like that has more than one image
The same with living artists. Many times, having found a stunning image, I went to the artist’s gallery - and there was only a sequence of endless YCHs or adopts.
In general, there are still quite a few prompters, much less than artists, so you also meet those you like hundreds of times less often. In the next five years the situation will significantly improve.
Yeah but I wanted more of this style, haha.
Well, maybe later I will generate more images in the same style. Currently there are plans to create a two-panel comic about Ember in trouble (slightly similar to this generation with Gilda). Then I plan to create an anthro image of Queen Novo, because there are quite a few good images of her. Then, perhaps, I will return to this style.
Actually I do, I meant it more that there isn’t more. Artists have a style and the artistic merit allows them to keep creating good art, but an AI can be fleeting. Like it’s been dreamed up. Rarely do I find a style from an AI that I really like that has more than one image, and even when it does, AI particularities usually end up meaning that there’s one good one I really like and then 3-4 slightly cursed images in the usual AI madness way.
Maybe it’s due to the work you’ve done cleaning it up, but I was really impressed by this one. Well done.
pictures of princesses created by AI in a variety of styles
Good prompting, then some inpainting with the same AI and about two hours of manual fixing of clearly visible AI artifacts in a graphic editor (Krita).
I wonder if the AI set could do full body shots of the princess sisters.
Why not? Here on Derpibooru you can find some full body pictures of princesses created by AI in a variety of styles.
Dang, I was really hoping this was an artist. Well prompted, I guess? What a style. I wonder if the AI set could do full body shots of the princess sisters.
@Background Pony #98A3
In this case you will have to wait at least a week. Because the model that generates in this style unfortunately does not know anything about Chrysalis.
I’m currently collecting a dataset for training changeling LoRA, but this will take a noticeable amount of time.
In general, there are still quite a few prompters, much less than artists, so you also meet those you like hundreds of times less often. In the next five years the situation will significantly improve.
In this case you will have to wait at least a week. Because the model that generates in this style unfortunately does not know anything about Chrysalis.
I’m currently collecting a dataset for training changeling LoRA, but this will take a noticeable amount of time.
nothing against your pick, I would prefer maybe Chrysalis and Cadance in this style
Thank you, I always try to add more emotions to my images when I generate them.
Would you like to see a Big Mac + Applejack pair in the same style?