David Heath |
July 26th, 2010 11:47 AM |
Quote:
Originally Posted by Mike Marriage
(Post 1552090)
It [35Mb VBR 4:2:0] ISN'T HDV and any argument saying it is is factually incorrect. It is far more robust.
|
Exactly. Saying it and HDV are the same thing is comparable to saying rough table wine and finest champagne are also one and the same thing, just because both are alcoholic drinks produced from grapes via a fermentation process!
Quote:
I plan on testing 35Mb VBR 4:2:0 against 50Mb CBR 4:2:2 as it may prove to be more robust than that just without the 4:2:2 colour.
|
Whilst I agree that the 35Mb codec is pretty robust, theory suggests that the 50Mb version should be even more so. At first sight, it should seem that because 4:2:2 is 33% more pixels than an equivalent 4:2:0 codec, it should need an extra 33% higher bitrate to remain the same compression standard.
In practice, that's not true for two reasons. Firstly, the luminance channel is allocated a higher relative bitrate than chroma, and secondly doubling the number of chrominance pixels doesn't even need the chrominance bitrate to be doubled. It's possible to interpolate the extra chroma pixels from the 4:2:0 ones to a reasonable accuracy - the extra bitrate is then only needed to correct the interpolation, not define the samples from scratch.
So of the 15Mb difference between the two codecs, a lot less than the 12 Mbs that simple theory would suggest is actually needed to give 4:2:2 - the rest is available for general compression improvements.
|