I think that's more of advertisement bullshit than reality. Let's take the example with the road and the desert. Only a human can understand where's the road and where are bushes. How does a piece of retail software know what to replace the road with? It can recognize any object on any photo with 100% accuracy and JUDGE what to replace it with? Excuse me, but universities around the world are wasting billions trying to teach machines object recognition and they still fail.
Of course, a software CAN recognize SOME shapes and SOME objects, but it's more like Pleo or Aibo level than intelligent image editing.
So why exactly would Adobe make a video like that then? If they faked it with pre-set pictures or anything, soon you'd have customers crying about false advertising.
Well look around you? How many products promise you'll become healthier, stronger etc etc. Your juice bottle has '100% natural juice' written on it but in fact it has 50% of natural stuff in it. There's always a smaller print somewhere telling you that 'yes, you can loose 100lbs of fat with our product but you need to spend 2 hrs per day @ the gym too'.
They're always lying a bit in the ads, that's wha I mean. Sure, that thing will remove lens flares and some other stuff, but you sure can't click on a tree or on road a get a desert instead of it. That is bullshit. Because for the machine it's just thousands of pixels, not a road or a tree. It can't possibly know where the road ends unless Adobe has achieved 100% object recognition and that tech should be immediately expropriated by DARPA and all its creators locked down in a dungeon beneath Pentagon.
Well look around you? How many products promise you'll become healthier, stronger etc etc. Your juice bottle has '100% natural juice' written on it but in fact it has 50% of natural stuff in it. There's always a smaller print somewhere telling you that 'yes, you can loose 100lbs of fat with our product but you need to spend 2 hrs per day @ the gym too'.
They're always lying a bit in the ads, that's wha I mean. Sure, that thing will remove lens flares and some other stuff, but you sure can't click on a tree or on road a get a desert instead of it. That is bullshit. Because for the machine it's just thousands of pixels, not a road or a tree. It can't possibly know where the road ends unless Adobe has achieved 100% object recognition and that tech should be immediately expropriated by DARPA and all its creators locked down in a dungeon beneath Pentagon.
That's a terrible comparison, really: completely unrelated. Yeah there's a lot of fluff and exaggeration in the health industry, but that has zero to do with Photoshop. The guy in the video clearly shows how the tool works, and anyone can replicate the few mouseclicks required perfectly - that's completely different from health food or similar BS that are borderline rip-offs anyway. So either the tool works just as advertised, or the video is doctored.
And what exactly would be Adobe's motivation to do the latter? Get a few sales? Then have their reputation tainted for not delivering what they promised? Extremely unlikely.
Nowhere does it say that the tool relies on object recognition btw (the road was outlined by the selection already, so it might just do some advanced seamless blending that is aware of surrounding areas) - and to be honest, I don't care how it works, either. This is clearly one of the main selling points of CS5, and from the looks of it, it will be amazing.
Oh and I'm not 100% convinced yet, either; those videos are too small to really judge the before and after results - original images would be best to compare. But even if there are artifacts and such, it's still a huge timesaver.
I just tried this in the CS5 beta that I got, works pretty well after all. But the build he uses in the video is probably newer, so it probably works better too.
Watercooled 5950X | AORUS Master X570 | Asus RTX 3090 TUF Gaming OC | 64Gb RAM | 1Tb 970 Evo Plus + 2Tb 660p | etc etc
I think that's more of advertisement bullshit than reality. Let's take the example with the road and the desert. Only a human can understand where's the road and where are bushes. How does a piece of retail software know what to replace the road with? It can recognize any object on any photo with 100% accuracy and JUDGE what to replace it with? Excuse me, but universities around the world are wasting billions trying to teach machines object recognition and they still fail.
Of course, a software CAN recognize SOME shapes and SOME objects, but it's more like Pleo or Aibo level than intelligent image editing.
The guy selected the road and told the software to delete it and fill it in. The software examines the surrounding pixels and fills in it accordingly. There's no object recognition going on. It's basically a smart clone stamp. It doesn't seem very far fetched to me and I'm actually surprised this hasn't been implemented already.
Indeed, when I look at the results I get, it does look very similar to the clone stamp and/or the patch tool. Different selections also yeild varying results, even selecting the same object a little differently can make a pretty big difference, not surprisingly
Watercooled 5950X | AORUS Master X570 | Asus RTX 3090 TUF Gaming OC | 64Gb RAM | 1Tb 970 Evo Plus + 2Tb 660p | etc etc
...because they already have! Dun dun duunn!
No, seriously. Why wouldn't they?
Sounds exactly like the douchebag move Steve Jobs or Bill Gates would pull off.
But I suppose buying Adobe would be way too fucking expensive to be profitable.
Signature/Avatar nuking: none (can be changed in your profile)
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
Another small step in making VR indistinguishable from RL 
omfg photoshop has brainz!
