Here's another one of C-Ko's drawings brought to life by ChatGPT.
Prompt: This drawing of a man is holding a sword in his right hand and a fish in his left hand. Render a photo-realistic image of this drawing. It's okay if it's deformed.
ChatGPT misinterpreted some details, so I annotated the image.
Prompt: I have added labels to the drawing of the man; I've labeled his mustache, teeth (mouth), and chin. With these parts identified, re-render the image of the man.
The shirt was still being weird, so I took another go at it.
Prompt: Render the image again, but don't let the shirt touch the sides of the head. Notice how in the original drawing the man has a narrow neck, so the rendering should also have a narrow neck.
A month later I had to mess with “perfection” and try to get the image closer to the drawing.
Prompt: Render the image of the man again, but have the mouth curve all the way up to the height of the eyes like in the original drawing. Be sure to show the teeth. It's okay if there are more teeth than a human normally has.
I made the mistake of generating this image in the same conversation as C-Ko's Portrait of A-Ko, so when I tried to tweek that one with the following prompt, it resulted in a hyrbid of C-Ko's two drawings.
Prompt: Going back to “the drawing with superfluous details removed,” notice that the nose should actually be on the far left of the face, to the left of the mouth. Re-render the image with the nose to the left of the mouth. Also, have the mouth curve like in the drawing, so that the corner of the mouth is slightly between the nose and eyes with the teeth running all the down to the bottom of the head (so that it can be implied that the other corner of the smile is on the far side of the head).
