2.5 is display gamma, NOT 2.2

Interesting! I myself have been using the 2.22 display gamma (with black compensation) in HCFR.
Any references/sources which speaks in favour of 2.5?

Ya, I have read much in these forums regarding 2.2 and 2.5 gamma values. From what I recall the suggestion seams to be 2.5 for a light controlled room and 2.2 where light cannot be controlled.

I have chosen to use 2.5 gamma for my set up. I have a light controlled room and at 2.5 my system produces excellent contrast and deep colors.

I always aim for 2.5 but generally I'm happy with 2.3 and up. Think my HD1 comes in at about 2.4 and my panny plasma is 2.5 ( ironically the uncalibrated 2.5 setting in the user menu is more like 2.2).

An undercompensated inverse 2.2 rather than 1/2.5 gamma is applied at the camera to compensate for the gamma altering perceptual effects of viewing in dim rooms (dark surrounds). NTSC assumes most viewing is done in dim rooms. If the end-user is viewing in a light room, the accompanying surround effects on contrast perception would require a gamma close to original compensated gamma of 2.2 to percieve contrast similar to what was seen in the dim mastering room on a 2.5 gamma crt. Dim viewing rooms, however, would require a gamma closer to the 2.5 gamma exibited in the CRT display the scenes were mastered from. If your room/surround is darker/larger than the mastering room chances are you may require a gamma higher than 2.5.

Varying gamma from 2.5 during coding or encoding is a tool to compensate for viewing conditions. To perceptually reproduce the proper gamma curve observed in the mastering room, unless the perceptual effect of your viewing environment exactly matches the mastering room, it would be necessary to alter gamma from the 2.5 standard at the display. One size would rarely fit all.

That's my take on it.

Only if your environment also matches what is used in the post production facility. Otherwise, a change in gamma is not only acceptable, but it is desirable.


Now let me add some nuance here: measured how? If I have a 3.0 gamma at the low end and a 2.0 gamma at the high-end, my average may be 2.5, but I assure you that a gamma curve like this will lead to problems in the image (lost shadow detail, potentially crushed whites). You need to know quite a bit more about gamma than simply one number. It is akin to looking only at the on/off contrast ratio as a measure of display performance. Interesting, sure, but wholly insufficient if you are trying to change how the display performs.

Bill

That is incorrect. Professional Sony Trinitron and Ikegami direct view reference CRT monitors which are found in just about every production studio are engineered for the ideal (2.20) response, a characteristic made possible per custom (gamma) signal processing.

All I know is that

(1) Sony XBR CRT monitors do not have a native 2.5 gamma after calibrating white and black levels. It is very close to 2.2.

(2) Many digital displays (whose black level cannot come even close to what a CRT is capable of) could not be calibrated to a flat 2.5 gamma without losing a significant amount of shadow detail.

Well, I’m still confused. I have struggled with this for a long time. I thought the target should be 2.2 and here is why:
If measurements to determine gamma are done using a colorimeter (e.g. eyeone LT), then how could it compensate for a dim environment? The probe is right against the screen. I just figured a measured gamma of 2.2 would yield a 2.5 viewing experience as long as the room was dark/dim. Maybe it’s just me that’s a little dim.

Whenever I get confused on gamma, I just head over to Poynton's website and re-read his Gamma papers.

This usually resolves the confusion for a little while.

Am I correct to assume that when all mastering is performed with digital displays (which I presume have a linear transfer function) then the inverse of the display gamma will no longer need to be applied therefore not corrected at the display. Gamma correction for perceptional purposes could simply be labelled as +.1 to -.4 or so to account for "dark surround", "dim surround", "day surround" and "on the dock surround" with a little custom tweaking thrown in to account for differences in display on/off CR.

Because your pupils adjust for abmient light and if there is an excess of light your pupils will constrict, your retinias will recieve less light and your brain will percieve black crush.

the gamma curve is measured thing and not a percieved thing. Your colorimeter will measure 2.2 in a dark or a bright room if that what you are calibrated to. It's your eye's ability to distiguish details in dark areas w/ abmient light present that requires a lower gamma curve for dark detail to be preceptable.

The probe doesn't; the software does. Having customizable gamma curves is typically one of the features that distinguish a budget calibration package (e.g., DisplayLT) from the "full" version (e.g., Display 2). Likewise, when you look at home theater calibration packages, you tend to find not only the ability to change the exponent, but also the underlying curve formula as well (e.g., CalMAN has four different gamma curves, with flexibility in the exponent for each).

Personally, I find getting rid of spot issues in the gamma curve to be a more important point for calibration than worrying about the overall exponent, since I have measured/calibrated multiple digital displays that had gamma values above 3.0 at the low-end. Combine this with an elevated black level, and the loss of shadow detail on these sets was no mystery.

One additional thought, the higher the black level of the set, the more important it will be to understand what you are seeing in the gamma-related data. You may find that your direct-view LCD has great gamma "statistics", only to find that you used some form of black compensation in the calculation. This tends to resolve the mystery of why 20% luma signals are indistinguishable from black on some sets.

Bill

So, the target should be 2.5, right?
What's the best way to calculate it with a front projector?

Yes the target should be 2.5 in a light controlled room. You may want to deviate to 2.2 in a non-light controlled room. Personal preference may come into play here. I have a light controlled room and use 2.5 gamma.

You do need some kind of light reading device/color meter to measure gamma. In basic terms gamma is the measured change in light output from one point to the next.

You should measure gamma from the lowest point that your meter will read up to 100 ire or 100% stimulus and in 10 ire/10% stimulus increments (5 ire/stimulus is even better). You should use window display patterns to measure it. If you own a CRT you may want to have lower gamma in the low ires (more brighter in the low ire end as front projector CRT's tend to come out of black very slowly) and a 2.5 gamma as you go into say 30 ire and above. For the most part I think you need a video processor like a Lumagen to do this kind of gamma adjustment. The Lumagen HDP has 11 gamma adjustment points that are somewhat parametric and this is well suited for 10 ire readings. Their Radiance is supposed to have 21 gamma adjustment points and is better suited for 5 ire adjustments. You also want your gamma curve to be as smooth a possible.

I think Bill made a great point about paying attention to each point and its output level rather than the gamma number. I don't look at the number at all until I am done, and try to get the best tracking at all points. The final number is less relevant than what the output is at each point along the curve. Some sets have some anomolies that can be improved by changing contrast, mode, drive, gamma, or backlight levels, depending on the design. I do what I can to get the best dynamic range and output curve and then review the numbers, moslty for reference.

On a display with only RGB cuts and gains you can adjust gamma by "ganging" the RGB controls at the lower (cut) and upper (gain) points of the grayscale. Some displays even have controls for this (all cuts /all gains).

Adding and taking away units if RGB also effects the overall brightness at each output point. The easiest way to do this is to get your RGB correct for your chromaticity aims and then add or take away units of RGB equally to reach you required luminance aim at that particular point of the intensity range for a given gamma scaled between your display black point and white point. In practice adjusting in equal RGB incriments will usually swing the colour aims off as well so you have to readjust for this as you increase or decrease luminance.

Essentially that gives you some way of effecting the curve ( gamma itself is a simplification) at these two points. Its rare for the display curve to have huge bumps between your adjustment points unless it has something wrong with it but obviously there will be certain limitations compared with a multiple point curve correction.

2.2 for mastering is really to help disclose the image a little more readily. I aim for 2.2 as a "technical" standard when creating imagery but I'll check it for 2.5 and sometimes even 2.0 depending on imagery.

2.5 I prefer for viewing with the idea being an end to end gamma somewhere over 1 ( 1.1-1.2). 2.2 is a little too flat for me in most viewing situations.

Digital video displays whilst linear devices usually apply correctional LUTS to give them a more traditional CRT type intensity response . This is really a prerequisite for a "video" display.

Adjusting "gamma" ( bear in mind the whole notion of video gamma is a simplification of the actual transfer curve) assuming you had a display that offered meaningful gamma adjustment , it would also need to state whether it applied numerically to the end gamma or the native gamma of the display.

Could you please explain 'end to end' gamma? Is that from 0 stimulus to 100 percent overall?....All of us in the short bus appreciate the explanation.

http://en.wikipedia.org/wiki/Gamma_correction
To compensate for this effect, the inverse transfer function (gamma correction) is sometimes applied to the video signal so that the end-to-end response is linear

aka end-to-end gamma of 1

By 'end-to-end' he means the entire chain from capturing the scene with a video camera, applying gamma correction roughly ^0.45 to correlate luminance to a voltage, that is to emulate the way the human eye sees luminance as perceived lightness, which is non-linear, so that the camera sees the scene the same way a human eye would...

... transforming the non-linear signal to linear intensity at the display using a power function of ^2.5 (gamma). Both are power functions so the constants multiply together: 0.45 * 2.5 ~ 1.1 (1.125).

Edit: Already answered above, sorry

Thank you both.

I'm still rather new to the "calibration world", so please forgive me (and let me know!) if my question about the above quote doesn't make much sense.

I was under the impression that changing "all gains" would be the equivalent to altering Contrast, while changing "all cuts" would be the same as altering Brightness (not sure about this one). So is this not true ? It would be great to be able to "fine-tune" both ends of gamma by playing with those two controls !

Thanks in advance for any comments.

Fernando

It is very hard to generalize, as different sets will respond differently. My experience is that the probability of making such "ganged" adjustments and maintaining decent gray scale tracking while effecting gamma changes is pretty remote. Most of these controls are not linear in their effect, and do not have exactly the same effect on all three CRTs either.

No think of the cut/bias control as adding or taking away individual units of RGB at the lower end of the grayscale however this doesn't just effect the color but the overall "luminence" as well (if you take away red for example the colour balance will slip towards blue and green but the overall luminance level will also decrease). Gain/drive essentially does the same thing at the upper end of the scale.

They interact with each other so in this sense they are similar to a brightness and contrast controls but its not that helpful to think of them in that way. Brightness and contrast are more about defining limit points.

Think of the grayscale as a tower of lego made up of bricks coloured R, G, B. The balance comes from the mix of the bricks , the "luminance" is the height of the tower. Less or more bricks effect the colour balance of the tower but they also effect the height .

Controlling the gamma in this way involves getting the tower the right height at that particular referenced point of the grayscale as well as the bricks in the right quantities.

I do say that you have to readjust for colour balance as you do it. Its not difficult but it means you might spend 30 mins on the grayscale calibration rather than 10. It also helps to realise that green incriments will make the largest overall adjustment to the luminence on most displays then usually red then blue.

I've done literally hundreds of displays including CRTs (mainly sony FW900 monitors) . The easiest way is to get the color balance correct first ( you will normally be slightly off down to one click plus or minus on one primary: I usually try to make it blue as normally adjusting blue then correcting the red and green is more predictable.

Unless you have a truly awful display you are not talking a massive readjustment to get the luminance aims correct. Its really not a big deal however it is about the only way you can have some control over gamma without resorting to prebuilt simplistic curve selections ( usually worse than useless) or some sort of multipoint 2d LUT type adjustment system.

2.5 would assume that the end user took it out of the box, and never used it or calibrated it. Durring the first 300 hours of its life it is 2.5, but it won't stay that for long.

Properly calibrated it should be 2.2, which also leaves room to compensate for aging, so that the display can be re-calibrated over time.

2.2 is the right number.