Rick Burnett
Section Moderator
Which of course, this thread just validates that people don't really understand what they are getting with 4:2:0 and 4:2:2, because it has nothing to do with the color science (or color depth, etc). It has to do with the spacial color sampling.
Have a look here:
http://en.wikipedia.org/wiki/Chroma_subsampling
Where the picture examples are perfect of showing what happens. Look at the colors, they are all just as rich, but, at the boundaries, you have color contamination. This is why green screen work is so difficult at times with 4:2:0 with detail below the spatial frequency of the color sampling because the colors are mixed together, and knowing what to key out and what not to key out becomes difficult as both things have the same colors in them at the contamination points.
Almost everything I have seen comparing 4:2:0 and 4:2:2 have shown be that if you aren't doing visual effects, and you are not SERIOUSLY pushing the grade, 4:2:0 is more than adequate. However, if you start heavily grading or doing VFX work, that contamination of different colors starts to become bothersome or noticeable, but it has no effect on the color space.
Now, what DOES have an effect on color space is bit depth. Given 8-bit, and not 10-bit, THAT is where the rest of your freedom can come in on grading where you have to push things really far. That is why, when using an 8-bit color space, you need to get as close in general as possible in camera so you aren't pushing your colors too non-linearly and getting banding in gradients.
Have a look here:
http://en.wikipedia.org/wiki/Chroma_subsampling
Where the picture examples are perfect of showing what happens. Look at the colors, they are all just as rich, but, at the boundaries, you have color contamination. This is why green screen work is so difficult at times with 4:2:0 with detail below the spatial frequency of the color sampling because the colors are mixed together, and knowing what to key out and what not to key out becomes difficult as both things have the same colors in them at the contamination points.
Almost everything I have seen comparing 4:2:0 and 4:2:2 have shown be that if you aren't doing visual effects, and you are not SERIOUSLY pushing the grade, 4:2:0 is more than adequate. However, if you start heavily grading or doing VFX work, that contamination of different colors starts to become bothersome or noticeable, but it has no effect on the color space.
Now, what DOES have an effect on color space is bit depth. Given 8-bit, and not 10-bit, THAT is where the rest of your freedom can come in on grading where you have to push things really far. That is why, when using an 8-bit color space, you need to get as close in general as possible in camera so you aren't pushing your colors too non-linearly and getting banding in gradients.
