The point is that most stuff isn’t provided in high kBit/s. Spotify, e.g., defaults to 96 kBit/s if you are streaming mobile. You can set the bitrate, and the highest value is 320 kBit/s. Virtually noone can hear the difference between that and a studio recording.
So I gather from the style of artifacts that the AI is attempting to imitate a two-dimensional photographic snapshot or portrait in one step?
I wonder how much spookier this might become if the AI instead starts by constructing a synthetic 3D model of a fictitious person. You might expect a lot of the background-related confusion to be resolved immediately, for one thing.
Uhhh. . . how many other people got this sponsored ad?
For a second I thought it was part of the article.
I also noticed that the background is a good indicator if it’s fake or not. The more realistic background is likely the real image and the vaguely blurry is fake.
Dashboard hit after multiple plastic surgeries.
This topic was automatically closed after 5 days. New replies are no longer allowed.