Please define the 10-bit codec

[Khalua]

I noticed a lot of you are talking about the new ''10-bit'' codec/s . Can someone explain in simple words what is it, why should we use it, etc...

Thanks.

[Cornucopia]

Video codecs work in particular color-spaces. 8 bit (the consumer standard) uses 8 bits per color primary (each of Red, Green, and Blue)***. This gives a total possible # of levels for each color primary = 256. Or total # possible colors/levels overall = 256 * 256 * 256 = 16million.

10 bit is a common pro level which assigns 10 bits to each color primary, so the total levels per primary are 1024. Total possible # colors/levels overall are 1024 * 1024 * 1024 = Almost 1.1 BILLION! This allows the original video to retain much more of the nuances of fine gradations between levels/colors. Less visible banding, less need for dithering. More "natural" looking.
Most of the time this is used to increase the "fine-ness" of the colors/levels, although it could also be used to increase the latitude/dynamic-range.

Note that there are other, even HIGHER levels of bitcount codecs/storage: 12bit, 14bit, 16bit, 32bit and Floating point. These are usually only used for things like Digital Intermediates and CGI for Film.

How should you use it? It's pointless to use these unless you:
A. Have a higher-bitcount master to start with that you want to retain the greate pristine levels of.
And/or
B. Have some MAJOR complicated video generation/mixing/processing operations where you don't want to incur the normal losses buildup that happen during those operations.

For those of us that are used to calling a file "24-bit color", if you multiply those codec #s by 3, it'll give you an indication of how big they are: 8bit=24bit color, 10bit=30bit color, 12bit=36bit color, 14bit=42bit color, 16bit=48bit color, 32bit=96bit color...

Scott

***(Alternately, it could work on Y, U, and V, but that's getting much more complicated..)

[Khalua]

Thank you very much for the detailed explanation. Now is pretty clear that I don't need it.