Does the shift of the entertainment industry to 4K UHD resolution displays matter for CG creators?
linvanchene
Posts: 1,382
edited and removed by user
Post edited by linvanchene on
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
So, 4K would be 3840 x 2160, and 8K would be 7680 x 4320? There are people currently producing images with these sorts of dimensions, presumably with a view to printing them. Hivewire recently ran a contest where the image had to be 3000 x 3000, and the damn' thing took forever to render, but again I think they asked for this size because they planned to print the winners.
There has to be a plateau, really. I would imagine most pictures would actually look worse in higher definition, because unless you are actually prepared to work at all the extra detail, and unless that extra detail is worth looking at, people will just tire of it. I could spend all day looking at an Ansel Adams photograph, but I don't want to look at high definition holiday snaps.
My last PC competition entry was actually purposefully blurred and reduced in size because I thought the original sharp 2k version was too bloated for the subject matter.
i am no expert here, but I DO have a 20 MP pro DSLR camera. IMHO Still pictures can justify that extra resolution since anyone can look close at the finest details, but when in motion, the eye cannot discern the same level of detail. look back a generation or 2. you watch a video tape, it looks ok. Now pause it and the picture is awful. unpause and the picture looks a LOT better again.
so if the image is a video, with motion, then the 4K will probably not be that big a deal.
but look at what went on with 3D...that will probably not come back until you get naked eye 3D.
Something else to think about is how much of the object will actually be shown in the render and would it need more resolution than is already available. Say I am doing a 4000x4000 render of an office worker sitting in front of a window with a city scape behind her. How much of the pixels in the image will actually be given over to the worker? How much will be made up of the distance shot of the city? Unless the image is rather close up shot of her then the odds are that she won't get more than half of the pixels and of that a percentage of them will go to other things like hair, clothes etc. Even if her face takes a full 1000 pixels (which is a quarter of the image) that is still below the number of available pixels used for most face textures currently.
Naked eye 3D may be on smarphones before 4K displays become the majority. There are companies working on that now.
It took a few years for Blue Ray to become mainstream, and that was much less of an investment than 4K, IMO. There are new solid state memory technologies on the horizon that may obsolete DRAM, Flash, and spinning hard drives in one fell swoop.
If professional artists need to produce 4K art and video, there's always Amazon Cloud EC2. If you routinely render with Luxrender, EC2 is a way to boost capacity and pay only for what you use. For that matter, Amazon Web Services can deliver the completed content with high bandwidth too.
4K is still a niche market, IMO. Blue Ray, on the other hand, was specifically targeted at consumers, many of whom already had 1080P TVs. Sony's decision to put a BD player in every PS3 was brilliant. That got more BD players into homes than any of the standalone player manufacturers did. Only time will tell on 4K.
Edit: Interesting article on Wikipedia: http://en.wikipedia.org/wiki/4K_resolution
I disagree about the video tape ever looking okay (assuming you are talking about VHS). Our first VCR when I was young was a Beta, and it forever made me a video snob, when the market shifted and we eventually ended up getting a VHS, the video quality decrease was very noticeable, though it had in theory better sound, and for sure had longer tapes (with a Beta, many rentals came on two tapes). I would have been an early adopter of Laser Disc, had I been able to afford it. I was an early adopter of DVD and HD. I don't really have it in my budget at the moment to go 4K. I don't buy into 3D at all.
I think an interesting thing that is going to play out over the next few years that may affect film and CG more than pixel resolution, is temporal resolution. We have been trained to think of 24 frames per second as "film-like", and many people grouse about anything faster than that as looking like "video" But 24 frames was not chosen for any artistic reason, it was basically the fewest number of frames that were needed to sync a sound track to when "talkies" came about. The silent films that came before were an even lower frame rate, in many cases not even uniform since they were hand cranked cameras. I've liked many of the tech demos I have seen of faster frame rates. I would have liked to have seen how it looked on a theatrical feature, but I couldn't since I am boycotting Peter Jackson's Hobbit movies (but that's a different discussion). People I have talked to about the process have mixed reactions, but I wonder how much of it is just because it is not what we are used to. Cameron may very well try it in the Avatar sequels, since he has promoted faster film rates for years.
Better temporal resolution can clear things up, in some cases, better than pixel resolution ever will. Action scenes aren't a blur, and they don't have to slow it down for the camera. Movement doesn't blur out the background, think of a picket fence seen from a moving car for example. It may actually help CG, because part of the reason CG can look so bad, is that it did not look like film (no grain), often having the film scanned to a digital intermediate, the effects added, and then printed back out to film, decreasing the quality a bit with each step. But, in this day of the Alexa and the Red and the CineAlta, most things are digitally filmed, processed and projected. The CG has gotten better and better over the years, so it can look very good. The thing that really makes CG glaringly obvious are poor animation, the difficulty of making things look like and behave like they have real mass.
Which is a long winded way of saying is I think higher resolutions will help CG more than hurt it.
I've been rendering at a bit higher than 4k for my stuff of late (5120x2880). I use LuxRender, and these renders take two to five days to converge sufficiently depending on scene elements/construction. I have also been experimenting with much higher resolutions for print (9000x6000). These consume 18-20GB of RAM for the LuxRender process and take around ten to fourteen days to converge. And my primary render node has some beef to it: http://fav.me/d79h74h
There was a document Disney released about the compute power they had on hand to render Frozen. 30,000 cores. Only large studios currently have the budget for server farms like that... Even AWS EC2 gets expensive fast when rendering at higher resolutions, given the cost on the bigger compute instances types required to render these resolutions.
Rendering at higher resolutions, my biggest observation is that the 4k face maps for figures do OK, but the torso/limb maps really start to show their lack of resolution because so much of the figure has been crammed into a single UV map. I wish DAZ had broken up Genesis/G2F into more intelligent UV maps rather than sticking with the old legacy layouts.