Talk:Rec. 2020

Coverage of 1932 Standards

I removed a line stating rec2020 coverage of CIE 1932 color space as well as adobe and rec709. These figures referred to a SID paper, which in turn refrenced Cambridge University Press, Cambridge, 1932. So if the figures are to be used they should cite the original article. I could not access a copy of the original article and based on my calculations the numbers given were incorrect. — Preceding unsigned comment added by 146.115.5.122 (talk) 14:01, 11 December 2015 (UTC)

The Cambridge paper from 1932 was about the CIE 1931 color space. As for the calculations they are referenced and are from an NHK paper that was presented at a SMPTE conference. --GrandDrake (talk) 06:15, 13 December 2015 (UTC)

Rec. 2020 color space

The color space is the primary colors and the white point. There are several transfer functions that can be used with the Rec. 2020 color space. --GrandDrake (talk) 03:59, 5 April 2016 (UTC)

The color space could be defined by the chromaticity coordinates of the primaries, the white point, and the electro-optical conversion function, but that does not in any way elevate the 1931 XYZ chromaticity diagram (which is 2 dimensional) to a color space (which has three dimensions). The diagram at the top of the page as of this date is not of a color space, it is a chromaticity diagram. A dimension is missing for it the diagram to be of a color space; it is merely a chromaticity diagram. Lovibond (talk) 23:07, 5 April 2016 (UTC)

Color space has been defined by the primary colors and white point while the transfer function can vary and scRGB even has a dynamic range that changes depending on the bit depth. The Rec. 2020 color space can be used with the Rec. 2020 transfer function, SMPTE ST 2084, and Hybrid Log-Gamma. Also color spaces such as CIE 1931 color space, CIE 1964 color space, and Lab color space don't define a transfer function. Also there are color spaces that define a transfer function but don't define a brightness. --GrandDrake (talk) 00:02, 6 April 2016 (UTC)

None of your statement addresses the fact that the CIE 1931 chromaticity space is not a color space. The figure near the top of the article is in chromaticity space, not color space.

The CIE 1931 color space does indeed include what you refer to as a transfer function, though implicitly. It is radiometrically linear. The CIE 1964 U* V* W* color space likewise has a transfer function, based on Reilly's cube root approximation. The CIE 1976 color spaces CIELAB and CIELUV both have as their transfer function a slight modification of the cube root formula used in the 1964 U* V* W* color space. Nevertheless, none of this bears on the fact that the figure is in chromaticity space, not color space. Accordingly, your most recent edit restored a factual inaccuracy that had been corrected, and it therefore should be reverted. Lovibond (talk) 14:04, 6 April 2016 (UTC)

So what is the fixed brightness range for CIE 1931 color space, CIE 1964 color space, and Lab color space? Many color spaces don't have a fixed brightness range which is why they are shown with a 2D diagram. Also to support HDR by the end of the year Rec. 2020 will have three transfer functions. --GrandDrake (talk) 02:51, 8 April 2016 (UTC)

Irrelevant. As I pointed out earlier, this does not make the (x, y) chromaticity diagram into a color space. Do you actually believe the (x, y) chromaticity diagram is a color space? Lovibond (talk) 02:52, 9 April 2016 (UTC)

A color space is a combination of color primaries and white point which is shown as a 2D image. Also even with a defined brightness range the encoding method (RGB, YCbCr, etc...) changes the maximum brightness of the primary colors. Dolby uses the term color volume to refer to a combination of color space, transfer function, and encoding method which can be used to make a 3D image of colors at all ranges of supported brightness. --GrandDrake (talk) 03:45, 12 April 2016 (UTC)

Y' as Weighted Sum

Under "RGB and luma-chroma formats", Y' is described as a weighted average. I believe it should be described as a weighted sum. See the equation given. — Preceding unsigned comment added by 172.91.176.10 (talk) 11:41, 9 March 2020 (UTC)

No, it is a weighted average, because it can be described as then devided by 1. Who cares though. 109.252.90.66 (talk) 14:29, 3 February 2021 (UTC)

Wrong Cb / Cr formulas

The RGB and luma-chroma formats section appears to contain wrong formulas:

YCbCr are calculated as C′B = 2⋅(B′−Y′)/(1−KB) and C′R = 2⋅(R′−Y′)/(1−KR)

The YCbCr article seems to provide the correct formulas:

C′B = (1/2)⋅(B′−Y′)/(1−KB) and C′R = (1/2)⋅(R′−Y′)/(1−KR)

which could be translated to C′B = (B′−Y′)/2⋅(1−KB) and C′R = (R′−Y′)/2⋅(1−KR), also.

--91.92.16.181 (talk) 17:32, 1 December 2020 (UTC)

Fixed. 109.252.90.66 (talk) 14:26, 3 February 2021 (UTC)