Originally Posted by Trat
I don't know Scott, to me Limited shows the same colors as Full...
What tests have you done to prove that? The basis behind your arguments have been a pinch thin.
After doing the math yesterday it would take only 10bits for YCbCr to not have any color loss when Limited/Limited is used. If it's 8bits then there is quite a bit of loss. What's interesting is the loss is on the conversion from RGB to YCbCr (within the PS3) and not on the conversion from YCbCr to RGB (within the set). Here's a small run of numbers to demonstrate it:
RGB = 23, 23, 23 => YCbCr = 36, 128, 128 => RGB = 23, 23, 23
RGB = 24, 24, 24 => YCbCr = 37, 128, 128 => RGB = 24, 24, 24RGB = 25, 25, 25 => YCbCr = 37, 128, 128 => RGB = 24, 24, 24
RGB = 26, 26, 26 => YCbCr = 38, 128, 128 => RGB = 26, 26, 26
This is what I'm getting at. This is digital, simply "expanding" a digital signal does not imply infinite resolution. IF the YCBCr is 10bit this will not happen, if it's 8 bit it will happen. I am curious what test you have done to truly know you are not loosing colors. The first test I did was pretty hard to verify but it did work for the six 14 color series samples I tried it on.
I need to produce an easy to use color banding test or we need to determine what video controller the PS3 is using.