John Galt (Panavision) and Larry Thorpe (Canon), two well respected industry veterans, gave along talk at the Hollywood Post Alliance about how to really understand what is going on in digital cameras. It went over so well they revised & expanded it and did another one at Panavision, entitled Demystifying Digital Camera Specifications. This is looooooong, but HIGHLY recommended if you want to really understand what is going on. As a Red owner, I couldn’t help but notice that the press release mentioned something about “pixels aren’t resolution,” which is true, but clearly a response to the Red One’s 4K resolution. I haven’t had a chance to watch them all yet, so I can’t fairly state whether there is an agenda at work (it would be somewhat fair to presume that Panavision and Canon cameras won’t be looked on TOO unkindly), but these two guys are well known and respected, and have been doing this stuff for a long time. Chime in with comments about your take on it all - I’m prepping for a big client demo, no time to watch today…
-mike
(Page 1 of 1 pages for this article )
There’s something like an hour+ of material here and a lot of it gets super technical (despite the presenters protestations that it wouldn’t).
Basically it boils down to this: Pixel count does not equal resolution because of Modulation Transfer Function, which has an effect at every stage (light through lens, lens to chip, all the way to the projector lens and it hitting the screen in the theatre.) In addition, single sensor bayer cameras are essentially optically 4:2:0 cameras because they record 1/2 the red & blue data compared to the green channel.
I guess you COULD take their claims as a swipe at Red IF you consider Red as the only single sensor Bayer camera out there, though John Galt did describe 4K as marketing hype at one stage, though not in reference to a particular camera (but couldn’t this also apply to Dalsa).
In clip 7 he describes a new chip for the Genesis with 5760 pixels across, and the RGB “photosites” arranged in columns, so it produces a true 4:4:4 1920*1080 RBG image. At the end he talks about an experimental Panavision camera that he describes as a true 4K camera, “which is 12K in the new math” (i.e. 12K of pixels but arranges to create 4K RBG rather than Bayer).
The point they’re making is similar to the arguments over the Panasonic HVX200 using DVCproHD at 4:2:2 at 960*720 and the HDV cams using MPEG2, 4:2:0 at 1440*1080
Posted by Dylan Pank on 05/10 at 12:35 PM
