Echoing one of the commenters: It's WITCHCRAFT.
That's incredible! I want it. For the Witchcraft version of Photoshop, I will pay the money.
Fred ,'Smile Time'
Got a question about technology? Ask it here. Discussion of hardware, software, TiVos, multi-region DVDs, Windows, Macs, LINUX, hand-helds, iPods, anything tech related. Better than any helpdesk!
Echoing one of the commenters: It's WITCHCRAFT.
That's incredible! I want it. For the Witchcraft version of Photoshop, I will pay the money.
I know, right? If it was a week later, I'd think they was punkin' us.
I was slack-jawed with amazement until the 'remove the road and replace it with plants that don't appear anywhere else in the photo' segment. Even in full screen mode, I could not see where the randomly shaped, densely situated plants came from.
Is that even possible for something like Renderman, much less PS?
I'm not surprised. I've been thinking for years that eventually we'd have this feature.
I mean, it's very cool (and would be fun to play with), but my first reaction is, "It's about time!"
I was slack-jawed with amazement until the 'remove the road and replace it with plants that don't appear anywhere else in the photo' segment.
It looked like it took one type of plant and replaced the whole road with it.
Even in full screen mode, I could not see where the randomly shaped, densely situated plants came from.
I thought I saw the plant they used near the road. Huh. May have to watch closer.
I don't recall them saying how powerful that OSX machine they were rendering on was, I'm assuming top of the line.
The last one is the one that makes me either a) scared or b) disbelieving. Because the mountain that gets filled in on the lower right hand corner takes a dip significantly AFTER the fill-edge. Unless content-aware-fill has been educated on the different between mountains and, say, rivers I can't imagine any way for it to realize that mountains behave that way.
I'm not removing the "we'll see" ticky box from my mind until I see it IRL.
Because the mountain that gets filled in on the lower right hand corner takes a dip significantly AFTER the fill-edge. Unless content-aware-fill has been educated on the different between mountains and, say, rivers I can't imagine any way for it to realize that mountains behave that way.
If you look a third of the way over to the left from that mountain, there's a dip at the same angle that the program used when it created that new dip.
The big "Wow" thing I want is something Microsoft developed. They're working on a program that can take two ore more different pictures of something (or a video of an object) and based on the difference, construct an internal 3-D model of the object (or at leas of the portion of the object that's visible to the camera).
That's something that's been sorta' a Holy Grail of "machine vision" since the 1950s. And I wants it.