What Is Gamma

… and why is it important. Many HDTVs will have a control, often tucked away under advanced settings in a picture settings menu, labelled gamma. User manuals are often unhelpful in explaining what the gamma control does and even the better ones may be less than specific with terms such as “adjusts the balance between bright and dark areas of the picture”. To know how to adjust the gamma control we need to know what it does and why, so read on.

Gamma is actually a throw-back to the days of CRT televisions. Look at the graph below which plots incoming signal level versus light output of a typical CRT TV.

Gamma Graph

Notice how an input voltage of 50% produces only 18% of maximum brightness. This non-linear relationship between input voltage and output brightness is inherent in CRT technology.

To correct this “defect” an inverse gamma curve, like the one below, is applied to the source material.

Inverse Gamma

Combine the “corrected” input gamma black line and the “broken” CRT gamma red line and the result is a linear relationship between input and output (the yellow line).

Gamma Correction

So that was how it worked on a CRT TV!

Fast forward to flat screen displays and what do we find? Plasmas and LCD displays don’t behave like CRT screens – they are inherently linear devices – but the source material is all being corrected for the old CRT displays so there is a big problem. What we now need is to make the flat panel HDTVs behave as if they were CRT devices and this is what the gamma control should do. We need to turn the yellow line into the black curve.

A gamma setting of 2.2 is usually considered the ideal setting and though some manufacturers offer adjustments with choices such as 2.0, 2.2, 2.5 etc. selecting 2.2 doesn’t necessarily guarantee that 2.2 is what you get. The only way to be sure is to measure the display’s output with a video analyzer and doing so is an essential part of an ISF calibration.

In the absence of a video analyzer can it be done by eye? Possibly, but you need to know what to look for! If gamma is set too low the picture will appear to be “washed out” whereas if set too high shadow detail close to black will be lacking.

The user manual that advised that the gamma control “adjusts the balance between bright and dark areas of the picture” tells the truth but isn”t too helpful. One more graph may help:

Different gamma settings

Big differences are apparent on the graph in the mid range (30 – 70 % input voltage). What does this mean in practice? Here are three screen captures, with gamma set to low (e.g. 1.5), normal (e.g. 2.2) and high (e.g. 3.0).

Photo of low gamma
Example photo of low gamma (e.g. 1.5)
Photo of normal gamma
Example photo of normal gamma (e.g. 2.2)
Photo of high gamma
Example photo of high gamma (e.g. 3.0)

Look at the mid range first. Low gamma (e.g. 1.5) produces a picture that looks “milky”, it looks flat and lacks punch. High gamma (e.g. 3.0) has loads of punch but now look at detail in the black areas – it has gone. Normal gamma (e.g. 2.2) gets the black detail and a picture that is crisp.

Setting gamma correctly is just one part of accurately calibrating an HDTV but it is an important part.

Piers Clerk
ISF Calibrator
www.homecinemaengineering.com

© Home Cinema Engineering Ltd. 2008

15 comments

  1. Dr. Sheldon Tarre

    I would have expected that gamma 1.5 would have given the best results as it is closest to the diagonal. Why is this not the case? Thank you!

  2. Because the set display gamma in combination with the source gamma is what combined makes out the final gamma, see picture “FInal Display”.
    And the last picture only shows curves for the set display gamma, 1,5, 2,2 and 3,0. And if you put those curves into the final gamma-picture you´ll see that in combination with the source gamma curve the 2,2- curve will lead to the most diagonal final curve summed up.

  3. Nice, easy to understand, article about this often miss-understood subject.

    To put what Nicklas said in anorther way:
    Systemgamma (“final gamma”) = sourcegamma * displaygamma
    The source gamma is 0.45 (except near black), according to the tv-standards. With a displaygamma of 2.2 we get 0.45*2.2 = 1.0.

    The “1.0” suggest that the image we see is exactly the “same” as the original, hence 2.2 is regarded as correct by many. But not all, including me, and there are good reasons for that – but for now lets just say that 2.2 from black to white is ALOT better then almost ALL displays when thay are set as “out-of-the-box” ;).

    (“out-of-the-box” many displays confirm to a s-shaped gamma curve which in dark parts are more like 3.0 above and in light parts more like 1.5 to look much more “punchy” and more contrast-rich – the result is an image that look totaly unnatural that lacks detail just about everywhere)

    For readers who might want to now more about subject, I recommend:
    http://www.poynton.com/GammaFAQ.html

  4. this info presented here blows me away.
    the original crt display technology (electron beam impinging upon phosphor, then resulting in non-linear voltage:light output) being called a Gamma.

    and a significant aspect of the ‘color management’ solution involves Gamma.

    if in fact, lcd displays are linear voltage:light output devides, it blows me away that the chosen means to apply color management and its gamma aspect to lcd displays is to attempt to make the lcd display become non-linear (to try to approximate a crt response). this seems like an ass-backwards approach.

    it seems to me that if the lcd is intrinsicially a linear voltage:light output device, that the better (eg, more accurate) solution here would be to leave its response alone and use a linear gamma of 1.0 within the color management solution.

    to go further with this line of thought. crts exhibit a gamma response, lcd’s a linear response, plasma a different, and with newer technologies, any given display technology might exhibit any arbitrarily characterized voltage:light output reponse. given this, it seems like the smarter solution here would be to abandon ‘gamma’ as the one-size-fits all means of characterizing all display devices, and instead by empirical testing, characterize each devices voltage:light output characteristic curve, and apply a generic inverse-correction formula that is not bound to being only of a ‘gamma’ type. by using cubic splines one may model any arbitrary correction formula.

  5. A good easy to follow and well illustrated explaninationg of Gamma.

    “The only way to be sure is to measure the display’s output with a video analyzer and doing so is an essential part of an ISF calibration”

    Picking a preset maybe part of the ISF calibration, and a essestianl part. But ISF training on gamma is a few slides and a few minutes in a 2 day training course. They just pick a preset that is between 2.2 and 2.6. If the display has more complex gamma controls, you need to check what will be included in the calibration and what their level of knowledge is.

  6. Commenting on Greg Aiken and ‘abondoning gamma’, how does he suggest all broadcast cameras switch off their gamma at once, and what about all the thousands of hours of archive material that has been shot with gamma applied? It may be historical, but gamma probably needs to be continued with.

  7. With my UK Panasonic G20 I find that a Gamma of 1.8 (combined with taking down the brightness and upping the contrast) gives me the post accurate colours, especially skin tones and greens. Why should Gamma adjust the colour spectrum too?

  8. With my UK Panasonic G20 I find that a Gamma of 1.8 (combined with taking down the brightness and upping the contrast) gives me the most accurate colours, especially skin tones and greens. Why should Gamma adjust the colour spectrum too?

  9. Great Article!!

    I just returned from an ISF training class and have been trying to get all the information that I was presented all straignt in my mind. This article shed a lot of light on a hard subject to comprehend.

    Thanks

  10. Thanks Piers, for the down to earth explanation, and the screen shots, which
    really told the story. From an adjust my own perspective, I will proceed with this
    knowledge, and do as I always do, but with more knowledge of the subject, and will
    probably get better overall results. I like to make adjustments, document the
    setting changes, and watch the results for a day or two because the programming
    material is so inconsistent that one can only go for the best overall average, while
    using your favorite programming material or movies as the benchmark. Add to that,
    changes in room lighting depending on the time of day, or weather, and one really
    has to look at the overall results over a day or two to really see how the adjustments affected overall picture accuracy and viewing enjoyment.
    Thank’s again – allstar

  11. Coming to this discussion a bit late.
    A comment, the source gamma in the second graph looks very like the knee of an Image Orthicon transfer characteristic from the 1950s and 60s.
    The gamma pre-correction applied by later colour cameras with their linear-response Plumbicon Tubes was smoother and very close to gamma = 0.45, using oven-mounted diode/s; or used a level-switched stepped-response approximation which again was quite close to 0.45.

    A second comment, critical observation of on-screen pictures requires the ambient light, brightness and contrast to be near ideal and well controlled.

    And a third – pre-correction for CRT gamma raises the voltage of dark areas from the camera, compared to white, reducing contrast ratio.
    When the CRT’s low gain near black restores the original contrast ratio, it also lessens the effect of noise added between camera and receiver – a bit like the principle of DolBy NR for audio.

  12. Isn’t this backwards? Wouldn’t the HIGHEST gamma number show you the lightest dark grays? This would mean that the first image is the higher gamma?

  13. Should be lower gamma is darker, but the sun will come up tomorrow