4:4:4 vs 4:2:2 a one stop difference?

[Mike Brennan]

Is there a concensus in the difference in dynamic range between 4:4:4 and 4:2:2? (or 3:1:1 for that matter).

 

1/2 to 1 stop?

 

Due to ..... :)

 

 

 

Mike Brennan

[Alvin Pingol]

I posted a similar topic to this a little while ago. It may be of use to you:

 

http://www.cinematography.com/forum2004/in...showtopic=1381&

[John Sprung]

4:2:2 and all that happens in the digital part of things. The CCD's would be the same, and dynamic range depends only on the CCD's. On a three chip camera, there's one CCD each for red, green, and blue. There's no way to make CCD's for Y, Pr, and Pb -- at least I've never heard of anyone trying to do it. It is an interesting idea, though. You could send most of the light straight through to a luminance CCD, and split off smaller amounts with yellow and cyan filters to CCD's of the same physical size, but with pixels twice as wide and twice as high -- two stops more sensitive than the luminance CCD. This would be a 4:2:0 system, since there's no raster scanning involved it makes no sense not to do two dimensional subsampling. We need to run this by a Larry Thorpe type who really knows video camera design.

 

At least for now, luminance and color difference is done with electrons, not photons. So the pixels are the same size on all three chips. Dynamic range depends on the sensitivity of the CCD's, and bigger CCD's can have bigger pixels and proportionally smaller dead spaces between the pixels. The bigger the pixels, the more photons you can throw at them.

 

 

 

-- J.S.

[Mike Brennan]

  There's no way to make CCD's for Y, Pr, and Pb -- at least I've never heard of anyone trying to do it.  It is an interesting idea, though.

-- J.S.

<{POST_SNAPBACK}>

 

Thanks John,

I wasn't thinking along those lines but I'll share the credit for the Y, Pr Pb ccd idea with you:)

 

I was referring to any possible loss of dynamic range caused by in-camera processing post A/D.

 

Possible (subtle) differences in highlight or shadow detail between 4:4:4, 4:2:2 (recorded 10 bit) and 3:1:1 recorded 8 bit.

 

I'm enquiring about dynamic range, but perhaps referring to *recorded dynamic range* would be more to the point.

 

Does rounding and quantisation chop a little off the dynamic range?

 

In respect to recording 4:4:4 are we really seeing an improvement in dynamic range?

 

 

Mike Brennan

[Phil Rhodes]

Hi,

 

No. Colour subsampling affects sharpness, particularly of highly saturated objects. It has no intrinsic affect on dynamic range.

 

Phil