Rec.ommendations for Display Gamma

1886.jpg

Rec.ommendations for Display Gamma

Old news really as this white paper from the International Telecommunications Union (ITU) standardizing gamma for reference monitors in HDTV was released in March 2011! But the topic of Gamma Correction as it pertains to Display Gamma is one that comes up constantly in conversation with colleagues in both post and acquisition. So much so that as I was combing through old NegativeSpaces articles detailing monitor calibration, I felt that a post on Gamma for broadcast monitors is one well warranted.

The ITU, which is located in Geneva, Switzerland, is the inernational governing body for broadcast, transmission, and television system standards. And like most international bodies, the ITU has no real authority and can only make rec.ommendations, such as Rec. 601 for digitally encoding interlaced analog video signals, and Rec. 709 for HDTV broadcast which after only a few years of widespread implementation is about to be superceded by Rec. 2020, a guideline for UHDTV, higher than HD resolution television systems (read 4k and beyond).  

So soon? Yep.

sony_uhdtv.jpg

Anyone involved with the creation of motion picture content should be aware of Display Gamma as a factor that has a profound effect on the way an image is perceived by the end user. And one that cannot be controlled by the content creator!

No point in reinventing the wheel here as a wonderful knowledge base is readily available at the excellent site, Cambridge in Colour, a learning community for photographers. This snippet from their article provides the best explanation and examples of Display Gamma I've been able to find. 

display_gamma_CIC.jpg

Here's a link to the full article, Cambridge in Colour - Understanding Gamma Correction. Additonally, their article on Input and Output Tone Curves is an excellent companion to this topic. 

Gamma is an inevitable part of the digital imaging process and can mean different things depending on where in the chain we're looking - file gamma, camera, display, etc. But as we see in the above example, what gamma effects is the midtones. That is - grayscale which is neither black nor white but somewhere in-between. In all optical perception, the separation between midtones and black or white is ultimately what defines the contrast. And as we can see in the most simple terms in the examples above, the exact same image seen with various display gammas will be profoundly different. 

This topic is a doozy and one I have no intention of delving too deeply into. To any academic reading this, they will no doubt find what I've written to be overly simplistic and only glossing the surface of a very complex topic. That's ok. There are plenty of great scientific and very thorough bodies of work already out there on the topic. Being a technician for motion pictures that are primarily broadcast bound, my concern is Display Gamma for HD Television Systems. The problem is the current "rulebook" for HDTV, that is Rec. 709, actually makes no mention of gamma. Because of legacy technologies - NTSC, sRGB, and others - we've arrived at a defacto "average" display gamma of 2.2. Despite this number not ever being specified by the ITU in Rec. 709, it is the display gamma you are likely to find on virtually any HDTV found in the home, on laptop and LCD computer displays, and many tablets and smart phones. And again, this number of 2.2 is a ballpark figure with many devices measuring at odd numbers such as 2.15 for the iPad Mini for example. The long and short of it is Display Gamma is one of the most critical components in digital imaging as it pertains to the end viewer experience and one that's in need of a universal standard. 

Enter ITU-R BT. 1886

(Another way these recommendations are listed is ITU-R BT."#", "BT" denoting "Broadcast Television". "ITU-R BT.709", reads "Radiocommunications Sector of International Telecommunications Union Recommendation for Broadcast Television #1886", the truncated version being BT. 1886 or simply Rec. 1886.)

Why this new recommendation? (from the white paper)

"This Recommendation specifies the reference electro-optical transfer function (EOTF) that the 

displays used in HDTV programme production should follow in order to facilitate consistent picture 

presentation. The reference EOTF is specified as a simple equation, with exponent function, based 

on measured characteristics of the Cathode Ray Tube (CRT)."

Because...

a) that reference displays play a crucial role in television programme production as they are used as reference for picture presentation; 

b) that the characteristics of reference displays should be unified to ensure consistent picture presentation of programmes produced for use in broadcasting; 

c) that historically, picture presentation characteristics were determined based on Cathode Ray Tube (CRT) characteristics, and the opto-electronic transfer characteristics were implicitly based on the CRT physical characteristics; 

e) that CRT reference displays are no longer available; 

f) that the electro-optical transfer function (EOTF) of CRT displays differs amongst manufacturers, amongst models, and amongst regions, as well as varying with the settings of contrast and brightness; 

g) that for the consistency of picture presentation, it is desirable that newly introduced display technologies have an EOTF that closely matches that of the CRT; 

h) that the reference EOTF for a non-CRT display has not been defined in any ITU-R Recommendation; 

j) that Recommendation ITU-R BT.709, provides specifications for the opto-electronic transfer characteristics at the source, and a common electro-optical transfer function should be employed to display signals mastered to this format

Buried in Annex 1 (page 2) is this very important bit of business -

γ (Gamma): Exponent of power function, γ = 2.40

As of Rec. 1886, the ITU is now recommending a Display Gamma of 2.4 for reference monitors in HDTV Systems.

But what does this mean for acquistion and post production where our work (for now) will almost universally be viewed on 2.2 displays?

The reality is, if you're monitoring and mastering at 2.4, contrast in these images will always seem slightly lifted on 2.2 displays. So for example, if the room we're coloring in has a Sony OLED using a gamma of 2.4, then our deliverables will look quite a bit different when seen on our client's 2.2 gamma iPad or MacBook Air. Rec. 1886 was needed but is far from universal implementation. Ask many a technician in both the field or in post if they had even heard of it and you would likely hear "no" more often than not. My own personal rule of thumb is that in situations where I'm handling the deliverables myself and those deliverables are for 2.2 displays, I set my monitors to 2.2. When I'm working with a facility and a colorist whose monitor is set to 2.4, then my monitors are 2.4. It's very important to be on the same page with whoever gets the stuff next and it will be different from project to project. The Sony OLED's fortunately have several choices for User Display Gamma which have made these monitors indispensable tools in today's digital imaging workflow chaos. 

At the end of the day, Rec. 1886 is just that, another rec.ommendation, but a step in the right direction.

UPDATES:

My Wireless HD Video article (Cutting the Cord) has been updated with several more systems as per reader suggestions. I got great feedback on this post and the content was improved because of it. Thanks for that. I truly appreciate it and makes the hours I spend researching and writing a blog post totally worth it. 

The new site is coming along great albeit terribly slow. Working TV hours, I have to be realisitic about just how long these projects take. That said, If I can go live with the new version before 2014 shows up, I'll be feeling great. I've been writing this site in the form of HD Cinema on Blogger since 2007 and on this domain since 2010. For better or for worse, I can't shake this project and it's something I'll probably continue to do for as long as I'm involved in this business. Looking forward to getting its next iteration out there.