I mentioned in my post about the grainy, pinkish, full-screen version of Carrie, that it might be nice to see what AI could do with those images.
I ran one “full frontal” image through two AI programs and it turned out that the idea wasn’t so nice after all.
AI did a so-so job at cleaning up the image, but it strongly suggested that what I thought was a full-frontal turned out to be a merkin. I discarded one of the images because it just did too much smoothing (so much that she seemed to be laughing), but the one I tossed suggested that you could actually see the thin strap holding the merkin up. I don’t know about that – no strap was visible in the full rear shots, but be that as it may …
The one I kept (below) didn’t show the strap, but it resolved the grainy image into a shape that suggests “patch,” not “pubes.”
I’m sorry I asked.
Click on the image to see the full size pic.
If you believe the stories, she could’ve been played by Carrie Fisher…
It’s probably more likely what you’re seeing here is the direction the AI-interpreted cleanup of the image went was in the direction of “patch/merkin” rather than that’s what it actually was. The AI probably wasn’t trained with a lot of real naked images and doesn’t parse the difference.
Also, with the version that gave you a strap– it doesn’t really know the difference between “thick bush” and “bikini bottoms”, so it picks a direction to go in.