HOME | DD

Description
I'm always looking to see what the current state-of-the-art is in computer-assisted artistic style-transfer. I'm getting really impressed with how far and fast this is progressing in image editing software.
Here, I started with an old 3d scene from 2021 that I rendered in DAZ Studio and was never happy with. (When I do these kinds of tests, I always start from my worst unfinished images since I've got so much bad artwork sitting around and eating up my disk space.)

I cropped the original image to just the head, applied a comic style-transfer filter using Photoleap, and then finally cleaned up things a bit more in Photoshop.

I've never seen a Photoshop comic action that I liked before, but this approach worked. The figure's position and expression, the facial features and shaping, the lighting and hair shape... all were relatively well maintained.

At this point I don't know whether to call something like this AI-assisted or not. There was no text-to-image part at all. But facial feature recognition helped with the style transfer, and the comic style itself probably (?) has an AI model component to it. (I'm guessing.)

Description
I'm always looking to see what the current state-of-the-art is in computer-assisted artistic style-transfer. I'm getting really impressed with how far and fast this is progressing in image editing software.
Here, I started with an old 3d scene from 2021 that I rendered in DAZ Studio and was never happy with. (When I do these kinds of tests, I always start from my worst unfinished images since I've got so much bad artwork sitting around and eating up my disk space.)

I cropped the original image to just the head, applied a comic style-transfer filter using Photoleap, and then finally cleaned up things a bit more in Photoshop.

I've never seen a Photoshop comic action that I liked before, but this approach worked. The figure's position and expression, the facial features and shaping, the lighting and hair shape... all were relatively well maintained.

At this point I don't know whether to call something like this AI-assisted or not. There was no text-to-image part at all. But facial feature recognition helped with the style transfer, and the comic style itself probably (?) has an AI model component to it. (I'm guessing.)