Constant Luminance

Since 1953, we have been using a block diagram for color video that is different from the one that a color scientist would prefer to use. The principles of color science dictate that we mix linear RGB (tristimulus signals) to make true luminance, denoted Y. If a video system were to operate in this way, it would adhere to the Principle of Constant Luminance. But in video we depart from that principle, and implement an engineering approximation: We mix nonlinear ("gamma corrected") RGB to make what I call luma, denoted Y'. (Many video engineers carelessly call this luminance.) To form luma, we use the coefficients that a color scientist would use to form luminance, but we use them in a different block diagram than the color scientist expects: We apply gamma correction before the mixing, instead of after. This alteration in the block diagram introduces a few image artifacts that are usually fairly minor. The departure from the theoretically correct order of operations is apparent in the dark band seen between the green and magenta color bars of the standard video test pattern.

Details are available in Chapter 8 of Digital Video and HDTV Algorithms and Interfaces.

The issue of constant luminance (or lack of it) is intimately intertwined with gamma correction. Gamma has unjustifiably acquired a bad reputation. I presented a paper on the topic, The rehabilitation of gamma, at a SPIE/IS&T conference in 1998.That paper outlines the Principle of Constant Luminance. As you can deduce from its title, that paper concentrates on the reproduction of lightness (which is related to luminance, which is related to luma). It merely outlines the color issues.

I presented the related issue of choosing luma coefficients for conventional video, DTV, ATV, and HDTV, in a SMPTE paper in 1998: Luminance, luma, and the migration to DTV. The so-called paper is virtual at this moment, having not been actually finished in that medium! However, the abstract of the presentation is available:

For the truly courageous, an audiotape of the session is available through SMPTE. The opening paragraph of this note is the first paragraph of that paper's abstract.

Some fragments of the paper-in-progress are available. Start with the brief technical note Errors due to nonconstant luminance. If you STILL want to keep going, access the links at the bottom of that page.

All of this will be tied together within a month or two, and then (eventually) released as the written version of the SMPTE paper.

• The rehabilitation of gamma (Acrobat PDF format, 1223508 bytes)
  
  
• The magnitude of nonconstant luminance errors (Acrobat PDF format, 89246 bytes)
  
  
• Transforms between luma coefficient sets (Acrobat PDF format, 87907 bytes)