3d coloring style

Medium (920px wide max)

Wide - use max window width - scroll to see page ⇅

Fit all of image in window

set default image size: small | medium | wide

Download (new tab)

previous 2 of 2

show original thumbnails

by

bbmbbf

Gallery

First in pool

AI assisted

2 in pool

Last in pool

By using a sketch, with a character model trained with your own art, and combined with a 3d model trained using renders (using your own renders or even screencaps from a movie), you can generate images with a 3d appearance. In fact, it is possible to choose a percentage of how 2d or 3d you want the image to look. In the present example, the image of Annie has been finished by the AI, using the model trained by me and combined with a basic sketch (it's not even artline), you can let the computer complete the work, achieving a result that would be very difficult for me to imitate just by coloring.

I know the results is far from perfect, is not even post processed (retouching, it's just raw generated). it was made just to ilustrate a point,

I must add that I have not been banned or prohibited from posting AI art on Inkbunny (the moderators here are the best and quite understanding), it has been my own decision because I am uncomfortable with some opinions (and maybe I myself annoy other artists who do not appreciate the use of technology). I am making this post because I wanted to show that, aside from using it for scenarios/seeking inspiration... in fact, one can save the coloring process using this tool. At the moment it is not efficient because you have to train a model for many hours for each character, but the possibility exists and I demonstrate it with this image, when the technology improves and everything is much faster, I am sure that many artists will save the work of coloring their work using not a 3d style like on this experiment, but their own coloring style. (The creator of Stable Diffusion said that a 30 times faster version will be released soon)

To comply with the AI regulations, I will provide some information about the creation process.

AI image generator : Stable Diffusion
Type : AI assisted (I provided the sketch and trained my ownn model)
Model : This is a combination of 4 annie models (I was yet learning so my first models was not providing good results, like annie having an arm always up, basically mixed all my models and added a 3d render 3d renter model on the mix that also trained.
Prompt : "Annie, purple skirt" (The model always try to draw her nude, so I must specify the clothes)
Method : IMG2IMG (using my own sketch)

I'm not planning to post more AI stuff for now, but i feel like need to ilustrate my friends who are spreading bad information about this technology.

Keywords