Talk:CIE 1931 color space

Please add new topics at the bottom

Please add new topics at the bottom

How can Y represent luminance if y chromacity coordinate changes in xyY with a change of the Y coordinate

y is defined as $y=\frac{Y}{X+Y+Z}$. Thus Y cannot represent luminance, cause the y coordinate changes if Y changes. Only if we set X = Z = 0, we get a fixed y coordinate in the xy-chromacity diagram. But as soon as X or Z have a non zero value, the y chromacity coordinate changes if you change Y. This is a contradiction. --95.91.242.234 (talk) 10:07, 23 May 2022 (UTC)

@95.91.242.234 Something is completely wrong here. I think Z represents luminosity, but x and y, as represented in the article, don't make sense to me either. 92.252.104.29 (talk) 04:47, 19 October 2023 (UTC)

@92.252.104.29 To correct myself. I just don't know what's going on. 92.252.104.29 (talk) 05:01, 19 October 2023 (UTC)

I believe you are confusing the xy of the chromaticity diagram with the XYZ primaries. The lowercase “xy” refers to the xy planar coordinates of the chromaticity diagram. The uppercase XY or XYZ refer to primaries, and not the xy coordinates. The XYZ primaries are imaginary (as an imaginary RGB). Thus, xyY refers to the xy coordinates of the chromaticity diagram, and Y is not actually the z-axis, as z = 1-x-y, but with xy specifying chromaticity, then Y can specify luminance, HK effect notwithstanding.

And yes, of course the y coordinate changes with changes in Y, consider how luminance changes affect saturation.  Myndex talk   16:44, 21 February 2025 (UTC)

Y can represent luminance, while still changing the x, y chromaticity coordinates. The reason is that luminance is mostly determined by the amount of green light. Red contributes only a little bit to luminance and blue contributes almost nothing at all. So to increase the luminance, you must add green. Unless you also add red and blue, the hue will of course change. XYZ is designed so that adding X automatically subtracts from green so that luminance stays the same. MTres19 (talk) 21:41, 24 April 2024 (UTC)

Discrepancy between LMS --> XYZ coordinates from the HPE matrix?

When I tried converting the normalized LMS color cone response to CIE 1931 XYZ, there seemed to be a big discrepancy from the S cone response to the Zbar graph. Supposedly, the Zbar should only consist of the S cone response, which tops out at 1. The matrix also does not scale the S cone response by any amount. However, the actual Zbar function's maximum is approximately 1.78 at 445nm . Have I made a mistake somewhere? 66.183.66.205 (talk) 07:14, 19 January 2024 (UTC)

Regarding Section "CIE xy  chromaticity diagram and the CIE xyY color space"

Since the gamut is defined by the projection of X,Y,Z  onto the x,y plane , it becomes a triangle, which is intrinsically convex. So the discussion of convexity is unnecessary and just adds confusion, unless one adds more sensor bands or channels as is the case in multi-spectral satellites and spectroscopy. Then the mult-stimuli projection should be convex.

Lets skip this complication for the tristimlulus and RGB cases.

MxBuck (talk) 20:08, 28 June 2024 (UTC)

Analytical approximation

In the mentioned chapter in this article, an analytical approximation is given to calculate the color matching functions. The formulas were taken from literature link [23]. In February 2025, a paper was published in the same journal, namely the "Journal of Computer Graphics Techniques" under [1], that provides a new calculation method. This method preserves the chromaticity of light near the UV and IR drastically better than the formulas given here and should therefore be more suitable for scientific calculations. I think this article should point the reader to this newer and more accurate calculation as well. 195.202.43.90 (talk) 10:25, 23 May 2025 (UTC)