Clubberlang to Mildlyinteresting 1 week ago (+9/-0) (web.archive.org)
x0x7 1 points 1 week ago
Or he just went to the mall in the early morning, filmed outside his apartment when people were at work, and generally filmed a lot and included all segments where he happened to not have people in view for a while, and also pointed his camera away from people.
People tend to face other people. I'm guessing you do because hopefully you're normal. At any given moment if you face any of the other directions there's often not people there. Just because there is often a person within 30 feet of you doesn't mean you don't have angles without people in them.
[ + ]
[ - ] observation1 0 points 1 week ago (+0/-0)
Far too many videos (not mentioned in this link), that clearly exhibit he is going places that are not possible. They are during the day and, even with covid, would have people
[ + ]
[ - ] Nagasaki 0 points 1 week ago (+0/-0)*
It is a modern deep learning approach. Train an algorithm to not only remove people, but fill in the image perfectly with photorealistic fill that matches the scenery. I am nearly certain that this is just some AI researcher having fun. It would be impossible to edit the videos by hand, and there are enough of them that it's unlikely that they went to an empty beach etc.
This is so effective that it must be taking in ~2 minutes of relatively still video and spitting out a short 10 second cleaned video, or maybe it's state of the art and the input is also only 10 seconds.
Agreed. Impressive either way.
Although, editing out the videos by hand may not be as difficult, if there was some aid from the program.
I think it is not as "streamlined" as the final cut makes it appear.
[ + ]
[ - ] obscenity 0 points 1 week ago (+0/-0)
or he's an actual time traveler as occams razor suggests
[ + ]
[ - ] DaijoubuYabai 0 points 1 week ago (+0/-0)
This must be a new definition of "Occam's Razor".