[–] zevv 6y ago ↗ Interesting detail is that the set seems to span age 14 to maybe 30, but no older then that. Would it be harder to generate realistic aged faces because of the increase of details? [–] briga 6y ago ↗ This is probably more a result of there being more images of young people in the training set
[–] briga 6y ago ↗ This is probably more a result of there being more images of young people in the training set
[–] scottmcdot 6y ago ↗ Quite a few in this album are cross-eyed:https://drive.google.com/file/d/1dNN8JejjaeKj8h8617f8pwCgoaJ...https://drive.google.com/file/d/1bYdjnKlqJdhJOrZIuN-zeCwyL8K... [–] Sohcahtoa82 6y ago ↗ Cross-eyed is the least of my concerns on that first one. Yikes.
[–] [dead] aaron695 6y ago ↗ It an interesting new field.Get stolen digital goods and clean them with AI.Not saying these where created from stolen goods at all, but there's no way to tell anyway so whateves.I'd be shorting stock photo providers.
7 comments
[ 3.1 ms ] story [ 29.7 ms ] threadhttps://drive.google.com/file/d/1dNN8JejjaeKj8h8617f8pwCgoaJ...
https://drive.google.com/file/d/1bYdjnKlqJdhJOrZIuN-zeCwyL8K...
Get stolen digital goods and clean them with AI.
Not saying these where created from stolen goods at all, but there's no way to tell anyway so whateves.
I'd be shorting stock photo providers.