The Forum
Contests => Past 2023 Caption Contests => Past Contests => March Caption Contest => Topic started by: Lisbeth on April 02, 2023, 11:06:11 PM
-
Inspiration came last minute this time but boy, am I excited about your opinion?
The girl in the photo is what happens when AI image generation works with 3D renders of Felicia. A far fetch from a real photo still but wow, these things are getting good...
-
Can’t comment but would love to see what renders you used.
-
Hi hi!
I used a pretty rushed one for this experiment: see below. The hair looks terribly and there's lots clipping errors in the pic, but it was enough of a foundation for the AI to work with.
The rest was a mix of using different key words and using interim image results as seeds for further processing. It's also a fun game of sorts between allowing your AI to stray further away from the original image (better results but drastic differences) and sticking close to what you uploaded (worse results but more true to your original upload).
This experiment also taught me a lot about what it takes to make AI generated images more appealing. Apparently, placing a model in front of an empty wall is not as helpful as I thought it would be because the AI tends to add random details to that blankness :D :D :D
-
That is a big shift on some parts while others not so much. It looked like the AI kept the same general body shape, a little bustier maybe. A fun thing to try out and I remember watching a video somewhere that you could do consistent characters through AI too. So someone could find ummmm someone they like and use her in different stories.
-
One of the biggest gripes I have with AI is stability really. The version of Licy you see in the cap is but one of several - some very skinny, some busty, some asian, some white, some weirdly proportioned and with a smile that will haunt my nightmares.
It's cool for one-offs but creating an identical character over several images seems to be tough - at least for now...