[Tig] ProRes444 necessary when grading BW

Bob Friesenhahn bfriesen at simple.dallas.tx.us
Sun Mar 17 16:20:44 GMT 2013

On Sat, 16 Mar 2013, David HJ. Lindberg wrote:

> In my case I'm coming from colored raw files converted to colored ProRes444
> which I then will grade to black and white. But now I understand that 444
> would give me more grey values per pixel than 422 would.

444 would not give you more actual grey values but it may provide more 
perceptually gray values than 422 if the gray is allowed to include a 
tiny bit of color (not-apparent to the non-expert viewer).  This 
assumes a full-bandwidth luma channel, which is available in both 444 
and 422.  Is black and white true gray or is it really highly 
desaturated color?

If RGB samples are all the same (R=G=B) then the image can be 
processed in RGB-form as 'grey' and will remain as such as long as all 
of the channels receive the same treatment.  It is more efficient to 
use just one channel for grey though.

> "but a matrix of RGB in a roughly 70/20/10 mix."
> Do you mean it's 70% of the green channel and 20, 10 of either red and
> blue? If so, which is 20 and which is 10?

21.26% red, 71.52% green, 7.22% blue.

This formula is only important when converting from RGB to Re.709 grey 
('luma').  It has no importance if all the pixels are pure grey.  You 
can apply it repeatedly to grey pixels and still get the same answer.

Bob Friesenhahn
bfriesen at simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer,    http://www.GraphicsMagick.org/

More information about the Tig mailing list