View Full Version : Scarlet 4:4:4...fantasy
02-20-2008, 03:08 PM
Ok I've read many times that although red is an awesome camera and all, it isn't true 4:4:4 at 4k due to the bayer sensor, but can be at lower res(not like 3k is low res, but you get the point)(If I'm off, I'm sure someone will defend red in the blink of an eye, but I'm pretty sure thats true). However, I have no idea if you can classify red raw as 4:4:4,4:2:2, and so forth. So if scarlet uses a bayer sensor and it is native 2k, we would not be able to get 2k 4:4:4, right, but maybe 720p or something? But RGB is a very different story I'm sure. Just trying to understand the video world and its limitations.
And I sure hope scarlet comes out awesome!:biggrin:
02-21-2008, 03:04 PM
Any clarification on this ?
02-21-2008, 03:54 PM
You can't call bayer raw data as x:y:z numbers. Think about it - in a 3chip camera, you start with 4:4:4 RGB, convert to 4:4:4 Y'CbCr, then throw away resolution in Cb and Cr, to get 4:2:2 say. In Bayer, we just record what the sensor sees, and reconstruct what the original RGB at pixel would have been.
In any well engineered camera, you won't have "full" resolution luma anyway, as you should be optically filtering to avoid aliasing, so although you sample and record it as "4", it will measure 3.5 or something.
With all these systems, what you have to do is look at the end result and see if you like it or not. All the numbers and figures are otherwise meaningless to the end result.
02-21-2008, 04:30 PM
If you want to understand the fundamental difference between Bayer CMOS v's 3 CCD imaging it might be helpful to take a sheet of paper and a pin.
For the CMOS prick the paper with 100 x 100 small holes. For the 3-CCD prick the paper with 33 x 33 small holes. Now go back and do that again, and now do it again. But make sure that the holes are in exactly the same location as the original ones or you will make a big mess.
Look at your piece of paper. Each one has the same number of holes (pixels) One has a fine pattern across the sheet of paper, the other one has a courser structure.
That's why descriptions like 4:4:4 etc don't mean anything, you are using the language of 3 pin holes on top of each other for a totally different sampling structure.
But if you look at the two hole patterns and consider how each one might represent a real world object seen through it that may be a clue as to why we like a Bayer structure .... :-)