This blog is a rewritten version of a creative cow post I made on July 25th, 2011.
Since the Color Correction Handbook was published, a single error of a tenth of a percent has emerged to cause me some small regret. I’m referring to the display gamma value I cite for digital Rec. 709 displays in chapter 1 of the book. I’m an advocate for standards in all of the postproduction work that we do, and it pains me that in attempting to clarify a confusing issue, I’ve inadvertently added just a dash more confusion.
At the time I was writing chapter 1, I was going by information regarding the inverse gamma encoded by Rec 709 compliant cameras, and so the display gamma I derived was 2.5, which fit with my understanding at the time. This value wasn’t challenged by anyone, and it was difficult to find a quoted gamma standard for Rec 709 that was mathematically derived, so this is what I went with. Unfortunately for me, in light of new information it appears I was off by .1 in that section.
Since publication, it has been brought to my attention via Charles Poynton’s own open letter to the industry that there, in fact, has never been a formal display gamma standard for Rec 709, as gamma was an implicit characteristic of CRT displays that was simply “built-in.” The actual gamma employed by CRT displays is quoted by Poynton as falling between 2.3 and 2.4, but this was never a published standard.
Of course, now most colorists putting together brand new suites are getting digital displays of one kind or another, and these digital displays lack the inherent gamma characteristics of CRT. For a variety of reasons that Mr. Poynton explains far better then I ever could, gamma is still a desirable and necessary characteristic in a display (even CPU displays have incorporated gamma settings for as long as flat panels have been available). However, lack of an easily referenced, published standard is confusing.
In his open letter, Poynton advocates for a published display gamma for digital broadcast displays of between 2.35 and 2.4 (he seems to be hoping that SMPTE will pick one), and peak white of 80-120 cd/m(squared).
It’s worth pointing out that the display gamma for projected digital cinema is yet another value, 2.6 (with peak white of 48 cd/m(squared)), but that assumes a completely dark viewing environment. My understanding is that higher display gamma settings represent scenes better in darker environments, whereas lower display gamma settings represent scenes better in brighter environments (which explains sRGB’s gamma standard of 2.2 for lit office/computer environments).
With that rationale, 2.4 falls in the middle for a muted “evening living room” environment. This all reinforces the importance of a carefully controlled viewing environment, where your display settings match the characteristics of the display surround, for doing color-critical work. Personally, I hope Poynton is successful and that a single gamma standard is published, as this is a confusing topic that engenders a lot of disagreement and doubt.