Which Gamma is Correct?
By Chris Heinonen on
As 2014 TVs begin coming out, I’ve been rethinking how I properly attempt to calibrate them. While measuring the grayscale and color gamut are easy, picking the correct gamma is hard to decide. There are many factors that come into play, from the room and environment to the capabilities of the display itself. It makes a large difference in how we see an image as well as you see the picture above with the different sides using different gammas. To try to come to a consensus on what gamma is correct for a review, I talked to other reviewers and experts to get their feedback.
I reached out to David Katzmaier of CNet, Kris Deering of Sound & Vision, Vincent Teoh of HDTVTest.co.uk, Stacey Spears of Microsoft and co-author of the Spears & Munsil Blu-ray discs, and Derek Smith of SpectraCal. Derek provided the paper that SpectraCal has published on BT. 1886 Gamma, which can be found here, that explains their position on the benefits of choosing it.
Q: What Gamma do you target when calibrating a display? Why?
Kris Deering: “Depends on the display and where it is located. For family rooms with ambient light I typically stick to 2.2. For theater rooms I’ll use either BT. 1886 if possible or 2.3. This gives the image a bit more of a contrasty look and punch.“
David Katzmaier: “I’ve been targeting a 2.2 gamma ever since I began calibrating TVs for systematic comparison. That’s what I was originally taught to use in ISF training and elsewhere. During a CalMan training class I asked about their use of 2.4 and higher gammas in darker rooms, and was told “2.2 is still the standard,” which was later reinforced by people familiar with Hollywood studio monitors calibrated to 2.2. “
“I also discussed 2.2 vs. 2.4 with colleagues like Kevin Miller, defending my choice of 2.2 in comparison lineups because in a room full of TVs, the other adjacent screens spoil the “dark room.” I was told it made sense.”
“I have decided to calibrate to BT.1886 for 2014, however, based primarily upon the strong recommendation of Spectracal (see “BT.1886 10 Questions, 10 Answers”), and the fact that after the Calman 5.2 update, BT. 1886 is the program’s new default. It makes more sense to me, especially in light of the non-zero black level of most displays. There’s some pain for me in the switch: namely, I have to recalibrate all of the comparison displays to the new gamma, and my meter, a CS-2000, takes a very long time to measure black (a step I could skip during some rough runs with power-based gammas…not so with 1886, which I believe requires a 0% measurement for every run). “
Stacey Spears: “If I know a particular show was graded on a 2.2 display, I will use 2.2, otherwise I just leave at 2.4.”
Vincent Teoh: “Currently, I target 2.4 with the lower-end of the greyscale (we’re talking 5% to 10% video signal) aligned to 2.1-2.2 if the display offers sufficient controls (10- or even 20-point white balance). Because I’m so used to the rich, high image contrast look that 2.4 gamma provides, I’m finding BT.1886 a bit washed out, since even with the Panasonic ZT/ VT’s black level of 0.003 cd/m2, the gamma tops out at 2.35 according to the BT.1886 formula.”
“Of course, 2.4 gamma does make shadow detail a bit less obvious, hence the targeting of 2.1-2.2 at low IREs on a capable display.”
Q: If you have a standard gamma you target but you find a display looks better at a different gamma during testing, do you use measurements from that different gamma, or from your target one for consistency?
VT: “For reviews, I target one value, i.e. 2.4 for consistency and to level the playing field. LCD TVs with 0.15 cd/m2 blacks belong in the conservatory, and don’t deserve to be salvaged with contrast enhancer/ black expander/ S-curve gamma. :-)”
DK: “I don’t publish pre-calibration measurements, and I don’t have much use for them beyond deciding which settings to use prior to calibration. For example, in deciding which gamma preset to calibrate in, I typically choose the one that’s flattest and comes closest to my target. Regardless, I always target the same gamma (now BT.1886) to create a level playing field for comparison.”
KD: “For measurements I see what the display can do. Since there is no “standard” for gamma per se, I will usually shoot for a 2.2 gamma for a flat panel or a 2.3/2.4 for a projector depending on its contrast performance. I will report which one I ultimately used and use that as my measurement for the review.”
Q: I encounter many displays now that ship reasonably well calibrated to a specific gamma (usually 2.2) but are non-adjustable. If you see a display that comes set this way, do you measure it at its preset gamma point, or your standard gamma knowing that will cause all the measured values to be worse?
KD: “I will measure a display for 2.2, 2.3 and 2.4 if it has a preset gamma value. I will see which one it tracks the best and report and use that.”
VT: “I measure at my standard gamma of 2.4. If, like you say, most displays ship with 2.2 gamma, it’s a level playing field anyway in terms of pre-calibration measurements. I have to praise Sony for targeting 2.4 gamma by default out of the box in its most accurate [Cinema] mode.”
“Post-calibration, if a display doesn’t offer sufficient controls to adjust gamma nearer to 2.4 and reduce delta errors (dEs), then too bad. I’m not going to make allowances – it’s unfair on displays which do offer these controls. It would also put pressure on manufacturers to either provide the controls, or to get it right out of the box.”
DK: “I always try to calibrate/adjust a TV using the available controls to come as close as possible to CNET’s standard for comparison (D65 grayscale, BT. 1886 gamma, 40fL max light output with a 20% window, delta errors for color as low as possible, etc).”
The same measurement with different gamma targets shows different error levels.
Q: What about BT. 1886 for gamma? Have you tested this on displays you have reviewed?
KD: “I will do BT 1886 for displays in dark rooms (or with full light control) if possible. This is what I use on my own projector. This typically requires a gamma that is adjustable at multiple points though.”
VT: “I’m very used to the high image contrast appearance with plenty of pop and depth granted by 2.4 gamma. I’ve tried BT.1886 on the Panasonic ZT/ VT and several other LED LCDs with deep black levels (Samsung F9000, Sony W9), and found even those images a bit flat (to me), since on those displays gamma tops out at around 2.3 to 2.35 according to the BT.1886 formula which takes into account black level. I prefer to align the midtones and above to 2.4, and low-end to 2.1-2.2.”
DK: “Only projectors so far. I am in the process of calibrating my reference displays (ZT60, F8000, etc) to this gamma now.”
Q: Given that gamma is something that you should adjust to best match your display and environment, does it make sense to test every display with the gamma that provides it with the best overall picture and measurements instead of having a default standard?
DK: “Perhaps for calibrators in the field, but not for TV evaluations at CNET. Again, I believe in a reviews environment a standard is required for fair comparison, and to reward those displays that most closely approach that standard.”
KD: “Yes, since there is no default standard anyways. BT 1886 is a defined standard, but it doesn’t seem to be widely implemented by anyone in particular. The case would be stronger if we knew this was what the content out there was actually mastered in, but even that is a guessing game at this point. I think you should pick the best gamma for the display/environment.”
VT: “As David Katzmaier said, that would make sense in the field, but since we’re reviewing TVs in a set environment, we will stick to a standard, i.e. 2.4 gamma.”
Q: Is there anything else you think readers and consumers should know about gamma?
KD: “Gamma is very display and environment dependent. Realize that most displays can’t do a proper 2.3 gamma let alone a 2.4. A lot of displays start creating a lot of display artifacts when you adjust gamma as well, so don’t just rely on measurements to ensure the display is accurate. Gamma is another one of the main reasons that calibrations should be done by those with the know how and experience to understand what limitations your room and display may have and getting the most out of your display.”
VT: “Previously, experts disagreed about what is the correct gamma value for home displays, ranging from 2.2 to 2.5. Even though there is now a specified standard, i.e. BT.1886, in a sense the gamma target will still vary from one display to another, simply because the formula is designed to take into account each individual set’s black level and white level.”
“At the end of the day, consumers are very unlikely to know for sure whether any particular material was mastered in 2.2, 2.4 or any other gamma value. The important thing is to get a high enough image contrast without sacrificing shadow detail or colour fidelity, taking into consideration the viewing environment. Bright room, 2.2; dark room, 2.4.”
DK: “One important point that should be conveyed is the move from previous power-based gammas to BT. 1886 in the production realm. As the Spectracal paper puts it: “BT.1886 is quickly being adopted in the field, particularly in the broadcast, production, and post-production industries.” So it only make sense to begin calibrating to this gamma for displays to mostly accurately capture the original source.”
SS: “It will become easier with UHD content because it will start at 2.4, unless Dolby gets its way and they throw out gamma and go with what our eye sees, which is what Dolby Vision does. I kinda hope that happens.”
“For live sports, 2.2 is correct since it comes straight from the camera. Even then, I leave at 2.4.”
“In a bright room, 2.2 is probably a safer choice for the average consumer.”
As you can see, even when it comes to a review, people can’t come to a uniform agreement. In the end, I will probably side with David Katzmaier here. In reviewing a display, a goal should to be remove as many variables as possible. If that means that every display should use a target of BT. 1886 for gamma instead of the display they come set for, then that will be the choice.
It also means that more display vendors might start to either offer BT. 1886 as a choice or the controls necessary to use BT. 1886 on a display. That option only benefits people down the road as they can use it when it is desired. If a display ships with a preset gamma of 2.2 or 2.4 then I will also mention that, but my pre- and post-calibration numbers will rely on a BT. 1886 gamma even if that causes the display to be worse.
There is no incorrect choice for what gamma one should use when calibrating a display, or reviewing one. Both readers and reviewers need to be aware of the reasoning behind a gamma choice, and the pros and cons of it.