or Connect
AVS › AVS Forum › Display Devices › Display Calibration › How power law gamma calibration can lead to crushed blacks
New Posts  All Forums:Forum Nav:

How power law gamma calibration can lead to crushed blacks - Page 3

post #61 of 230
Quote:
Originally Posted by PlasmaPZ80U View Post

When CalMAN 5 comes out, will it make sense to calibrate to this new standard or is it really there for future use? Does all current program material like HDTV programming and BD movies use power law gamma?

Since the purpose of the formula is to emulate the behavior of a CRT in a studio I think it would be appropriate to use immediately. Since no two studio environments are the same using the power formula is as likely to be close the mastering environment as anything else.
post #62 of 230
Thread Starter 
My thinking as well. There is no way to know what the EOTF was for producing any particular source, raw CRT or pure power law or something in between. The new recommendation (actually not so new, it was published over a year ago) is designed to be used now.
post #63 of 230
I'm not sure this will really help, but LightSpace CMS is one of the main calibration systems in use within the professional film and TV industry, and we (Light Illusion) also provide a lot of direct calibration for many of the major players - be it in training clients to use LightSpace, or as direct calibration for those who prefer to have it done by an external operation.

I can state that the majority (the majority, not all!) of TV based calibration for mastering in our industry uses a Gamma of 2.2, measured by normalising pre-set black/white points.

This means setting black to be 'just' visible, using the Cal Image on the Light Illusion website, and then setting white to be approx 82 cdm2 (23 FtL).

100 or above is not used as most displays distort or clips colour channels at those levels.

The Contrast Cal Image is used to check for clipping.

The room illumination for TV work is set to approx a 'standard' living room environment... what ever that means! But, never a 'dark' room.

(For film grading things are rather different...)

Gamma (and Gamut) is then checked by direct measurement, with 2.2 Gamma as the target, normalising the black and white points the display is capable of.

You can see some of this here: http://www.lightillusion.com/display_calibration.html including the images used as described above.

Hope this helps.
post #64 of 230
Thread Starter 
So I take it that the Light Illusion systems will not be adopting the ITU recommendation BT.1886?
post #65 of 230
Quote:
Originally Posted by Light Illusion View Post

This means setting black to be 'just' visible, using the Cal Image on the Light Illusion website, and then setting white to be approx 82 cdm2 (23 FtL).

100 or above is not used as most displays distort or clips colour channels at those levels.

Hi, do you have test patterns that show this? I ask because your clipping patterns are for white only. Also, SMPTE and EBU have always differed on this issue. SMPTE recommends 120 cd/m2 and EBU recommends 80 cd/m2, presumably for the reason you cite. Finally, I do have color clipping patterns here and they show no clipping at least up to 240 on a monitor calibrated to 120 cd/m2.
post #66 of 230
Hi Tom, nope - never bothered with that, but it wouldn't be hard to generate if really needed.

I know SMPTE recommendation is 120 Nits, but that has never been used in the real world. The EBU setting has become more the standard, but usually slightly higher, as I said.

If people want to go to 120 nits, that's fine, but that's not the value used in the professional world as a 'standard' at the moment. It may change, but not yet.

Zoyd, we will probably add BT.1886, but am in no rush to do so as it is not being used at the moment.
post #67 of 230
Using the method in the first post, trying to flatten out the gamma curve @ 2.65 standard offset camera gamma target, do I still need to check the flashing low APL pattern on AVS disk and correct the brightness to make bar 16 stay black?

Because when I rise the 10 point WB @ 10% 20% 30% levels the screen turns a little brighter and the 14 15 16 bars on low APL pattern start to flash.

Thank you!
post #68 of 230
Thread Starter 
yes, anytime you adjust gamma you need to go back and reset brightness after you are done. I'd recommend using the BT.1886 function in the spreadsheet on the previous page instead of an elevated camera gamma.
post #69 of 230
Thank you! smile.gif
post #70 of 230
So, I wade into this thread at significant risk to my blood pressure level ... smile.gif

My question or comment is this:

My understanding of the BT.1886 recommendation is that the primary EOTF method is "The Standard" and the alternate method is *not* "the standard," but simply a way to emulate a CRT (which presumably will no longer be used in production/editing/mastering booths.) Therefore, if we wish to calibrate to "The Standard," we should just stick to the primary EOTF method, as anything else would just be trying to match an "unknown quantity." .... True or False?
post #71 of 230
Thread Starter 
Correct, while both emulate CRTs the secondary one would only be used in specialized conditions, probably if you already know that function is closer to what you need and are trying to match some previously rendered material.
post #72 of 230
Quote:
Originally Posted by HDTVChallenged View Post

So, I wade into this thread at significant risk to my blood pressure level ... smile.gif
My question or comment is this:
My understanding of the BT.1886 recommendation is that the primary EOTF method is "The Standard" and the alternate method is *not* "the standard," but simply a way to emulate a CRT (which presumably will no longer be used in production/editing/mastering booths.) Therefore, if we wish to calibrate to "The Standard," we should just stick to the primary EOTF method, as anything else would just be trying to match an "unknown quantity." .... True or False?

is that BT.1886 calculator you've been working on ready for primetime?
post #73 of 230
Quote:
Originally Posted by zoyd View Post

Correct, while both emulate CRTs the secondary one would only be used in specialized conditions, probably if you already know that function is closer to what you need and are trying to match some previously rendered material.

Thanks ... just trying to work through the all practical aspects. I thought perhaps there would have been some benefit to the alternate method wrt trying to get our "modern" LCD's with poor black levels to look more CRTish for older titles ... as opposed to just going with straight power law.
post #74 of 230
Quote:
Originally Posted by PlasmaPZ80U View Post

is that BT.1886 calculator you've been working on ready for primetime?

More or less ... honestly I think Zoyd's is probably better, at least in the sense that he managed to get the alternate method worked out. smile.gif

Mine might be slightly more useful for a calibration run. The .NET version was taking too long to get it to work how I wanted, so I wound up re-doing it as a spreadsheet. I hate spreadsheets. smile.gif

I wanted to be able to just copy/paste from the widget into HCFR. At the rate I'm going, the usual suspects will have already incorporated BT1886 into their latest releases ... so probably wasted effort.
post #75 of 230
Quote:
Originally Posted by HDTVChallenged View Post

More or less ... honestly I think Zoyd's is probably better, at least in the sense that he managed to get the alternate method worked out. smile.gif
Mine might be slightly more useful for a calibration run. The .NET version was taking too long to get it to work how I wanted, so I wound up re-doing it as a spreadsheet. I hate spreadsheets. smile.gif
I wanted to be able to just copy/paste from the widget into HCFR. At the rate I'm going, the usual suspects will have already incorporated BT1886 into their latest releases ... so probably wasted effort.

Yes BT.1886 will be in a released version of CalMAN very soon.
post #76 of 230
Quote:
Originally Posted by PlasmaPZ80U View Post

is that BT.1886 calculator you've been working on ready for primetime?
Just threw this together quickly if all you're wanting is an easy way to get the numbers required to calibrate your display:

http://www.filedropper.com/1886
post #77 of 230
Quote:
Originally Posted by PlasmaPZ80U View Post

is that BT.1886 calculator you've been working on ready for primetime?

Alrighty then ... here is the spreadsheet version(s) ... I originally used LibreOffice to create the spreadsheet, so the graphs may or may not work with Excel. There's a MS Works version also, in case the Excel format file balks.

BT1886Calc.zip 13k .zip file
post #78 of 230
^^^ The above tool only does the primary BT1886 function, but also compares BT1886 to "power law" and "power law with linear offset black level compensation."
post #79 of 230
Quote:
Originally Posted by HDTVChallenged View Post

Alrighty then ... here is the spreadsheet version(s) ... I originally used LibreOffice to create the spreadsheet, so the graphs may or may not work with Excel. There's a MS Works version also, in case the Excel format file balks.
BT1886Calc.zip 13k .zip file

nice job, I especially like the part where you compare point gamma values between BT.1886, power, and power with BLC

now the only thing I need to find out is when CalMANv5 is coming out so I can decide whether to just wait for it to streamline the process or whether I should take an early peek at what BT.1886 gamma looks like on my LG

it's a shame I cannot try BT.1886 on my other displays since the best they offer is 2-pt white balance and gamma presets
post #80 of 230
Quote:
Originally Posted by PlasmaPZ80U View Post

it's a shame I cannot try BT.1886 on my other displays since the best they offer is 2-pt white balance and gamma presets
What is your source device? If you have a HTPC, you can calibrate gamma through the madVR video renderer.
post #81 of 230
Quote:
Originally Posted by Chronoptimist View Post

What is your source device? If you have a HTPC, you can calibrate gamma through the madVR video renderer.

one display is an old panasonic plasma, which has no gamma controls, only 2-pt white balance in the SM

source devices are a panasonic bd player and comcast HD DVR/STB
post #82 of 230
PlasmaPZ,
Calman5 is due out next month (Summer) was the last I heard off Joel
The anticipation for Calman 5 is killing me tongue.gif..
post #83 of 230
Quote:
Originally Posted by zoyd View Post

Just to have a look, here are the two BT.1886 transfer functions calculated for my display (peak white 100 cd/m^2, mll 0.05 cd/m^2). It's interesting that in order to reach a gamma=2.35 over the flat portion of the primary recommendation curve requires an mll of 0.005 cd/m^2


Some of this discussion is going over my head. However, I think if you're using % stimulus and the absolute measured luminance values on a display to compute the slope in log-log space on graphs like the one above, imo that would not represent "gamma". AFAIK, there is no such thing as "absolute" or "relative" gamma, as some other folks seem to suggest here and elsewhere.

My interpretation of Rec. 1886 is that the γ=2.40 exponent used in the reference EOTF equation is, in fact, the display gamma in this case, and the other parameters in Rec. 1886 are designed to ensure that a display conforms as closely as possible to that. This could be an oversimplification of how it's supposed to work though, because there are still aspects of the standard that I don't fully grasp.

If anyone disagrees with this, please feel free to school me on how you believe it's supposed to work.
Edited by ADU - 7/5/12 at 4:36pm
post #84 of 230
Quote:
Originally Posted by ADU View Post

My interpretation of Rec. 1886 is that the γ=2.40 exponent used in the reference EOTF equation is, in fact, the display gamma in this case, and the other parameters in Rec. 1886 are designed to ensure that a display conforms as closely as possible to that.

You are close. 2.40 is the display "gamma" assuming the display has a "perfect" black level of 0.000000Nits.

The other parameters of the primary function are basically black level compensation on steroids. If you have a less than perfect black level, then "gamma" in the traditional power law sense is no longer a constant.

Thus, if you read between the lines, the goal appears to be to make any display with an arbitrary non-zero black level *appear* to look like a display with a "perfect" black level of zero and a "gamma" of 2.40.

IOW, if you can't have "real blacks" ... fake it. smile.gif
post #85 of 230
Ok ... So PlasmaPZ80 asked a question about how I set up BT1886 on my display. This caused some ruminations which in turn led to a couple of tweaks and additions to my BT1886 calculator/comparator that I posted a few weeks ago. So without further delay, here's version 2(ish) of the spreadsheet. As before, you'll need software that understands Excel files. I omitted the MS-Works version this time. BT1886CalcV2.zip 8k .zip file
post #86 of 230
Great job, thanks! Playing around with the values in the spreadsheet is very helpful in getting a handle on the differences.

Would it be hard to add sRGB in there?
post #87 of 230
Quote:
Originally Posted by Chad B View Post

Would it be hard to add sRGB in there?

LOL ... define "hard". smile.gif I just took a peek at the sRGB "gamma" equation(s) this morning and all I got out of it was a nasty headache ... too early ... not enough caffeine ...
post #88 of 230
Quote:
Originally Posted by Chad B View Post

Great job, thanks! Playing around with the values in the spreadsheet is very helpful in getting a handle on the differences.
Would it be hard to add sRGB in there?

isn't sRGB essentially the same as 2.2 power law gamma?
post #89 of 230
Quote:
Originally Posted by PlasmaPZ80U View Post

isn't sRGB essentially the same as 2.2 power law gamma?

No it has a linear tail and an exponent of 2.4.
post #90 of 230
Quote:
Originally Posted by Chad B View Post

Would it be hard to add sRGB in there?

Thanks to an assist by sotti, it turns out to be not that hard at all ... I was temporarily befuddled by the inclusion of the RGB<->XYZ translations in the "official" definition of the sRGB gamma function.

So ... added the sRGB, but then I couldn't resist making a few more additions (namely Delta Luma calculations) to work more easily with copy/pastes from HCFR.

So rather than link up an already "obsolete" version, I'll hold off until I finish up the new-new stuff.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Display Calibration
AVS › AVS Forum › Display Devices › Display Calibration › How power law gamma calibration can lead to crushed blacks