Copy of meeting announcement.

White Gamut...it's not just for Polar Bears!

Come and learn why you should care about the way in which "white" is displayed in the new Digital Cinema projectors.

We will attempt to provide answers to your burning questions; What is White? What is Calibration White? What is Virtual White? What is Mastering White? How do I know what "white" is going to show up on the screen at the local theatre? Can I pick my own white point?

Does this really matter?

We will present a summary of the findings of the SMPTE White Gamut Study Group. The result is a set of recommended revisions to the
DC28 Reference Projector document RP 431-2. Our presentation will explain the details of how colors will be displayed by a DC28 Reference Projector, and how our recommended revisions will improve the match between mastering and exhibition. We will have a Digital Cinema demonstration that clearly shows the potential clipping that can occur without our proposal. Also we will show the Pixar short 'Presto'
using the new projector settings.

In addition Dan Sherlock of the Academy of Motion Picture Arts and Sciences'
Science and Technology Council will demonstrate and explain the new SMPTE Digital Leader. This not only provides the critical verification of sync between sound and picture, but also provides a wealth of features to verify proper image presentation.

This overview will show how the new digital leader helps with setting screen masking, verifying projector color space and white point settings, evaluating the quality of projector rescaling and filtering and checking the stereoscopic phase.

The meeting will be held at the Linwood Dunn theater, 1313 N. Vine St.
Hollywood. Social hour 6:30pm, meeting start at 7:30pm
MARCH 24th

This meeting is open to all interested.

What is White? What is Calibration White? What is Virtual White? What is Mastering White? How do I know what "white" is going to show up on the screen at the local theatre? Can I pick my own white point?This is news to me. I'm working on the assumption that DCI white is x0.314, y0.351. Why would you have different white points for mastering and calibration? I have no idea what "virtual" white means.

Are you going to this meeting?

"Come and learn why you should care about the way in which "white" is displayed in the new Digital Cinema projectors"

IIRC the white point in the DCI spec was designed to favor lumens by using a color temp closer to the natural Xenon spectrum. I think it's somewhere around 5500K. This has always been a weakness to the DCI spec in my opinion because the wider color gamut and other features of DCI could allow it to have uncompromising color accuracy. I don't know if this is the purpose of this SMPTE discussion but it wouldn't surprise me.

This is news to me. I'm working on the assumption that DCI white is x0.314, y0.351. Why would you have different white points for mastering and calibration? I have no idea what "virtual" white means.

Are you going to this meeting?

see http://www.smpte.org/journal/wp-content/uploads/2008/01/07-12-digcinema4.pdf Tables 1-2. Agrees with you. Note: 100% white at 14 FL.

It would be very interesting to attend this. Isn't the different white coordinate point due to the lamphouse color temp? Film release prints are timed to accomodate this, too. Still, you would think that the measured/calibrated white colorimetry would be .313/.329. I will make a couple of calls around "Hollywood" and see what I can find out.

It would be very interesting to attend this. Isn't the different white coordinate point due to the lamphouse color temp? Film release prints are timed to accomodate this, too. Still, you would think that the measured/calibrated white colorimetry would be .313/.329. I will make a couple of calls around "Hollywood" and see what I can find out.The DCI green is 24% more saturated and the red is 17% more saturated than Rec. 709, while blue stays the same. That shifts the white point towards yellowish-green. Thus, x0.314, y0.351. That's a color temp (a truly useless metric IMHO) of approx. 6300k.

The DCI green is 24% more saturated and the red is 17% more saturated than Rec. 709, while blue stays the same. That shifts the white point towards yellowish-green. Thus, x0.314, y0.351. That's a color temp (a truly useless metric IMHO) of approx. 6300k.Thanks for that info. Interesting.

The white point for DCI is a much greener tint than what is used in HT. This is because it is a more efficient point for the Xenon bulb that is used and it also is a brighter point for white with those bulbs. I went to a demonstration recently with a newer display technology where they calibrated the display to DCI spec and showed Blu-ray material on it and it looked pretty horrid compared to a perfectly calibrated 709 display right next to it. The whites just looked very green by comparison. They even had a display calibrated to DCI color points but with a 709 white point and that looked better than the native DCI display.

But if the material is mastered with that white point in mind you shouldn't even notice it in theaters and I understand completely why they do it.

I went to a demonstration recently with a newer display technology where they calibrated the display to DCI spec and showed Blu-ray material on it and it looked pretty horrid compared to a perfectly calibrated 709 display right next to it. The whites just looked very green by comparison. They even had a display calibrated to DCI color points but with a 709 white point and that looked better than the native DCI display.

But if the material is mastered with that white point in mind you shouldn't even notice it in theaters and I understand completely why they do it.This isn't surprising. I wouldn't expect that viewing material mastered for Rec. 709 on a display calibrated to DCI specs would be a very pleasant experience. The reverse would also be true. Were there any DCI-mastered material available, viewing it on a display calibrated to Rec. 709 targets should also look bad.

Historically print film has been balanced for D.55.

The baseline cineon digital scanning and recording spec was also correspondingly D.55.

Monitors for film work were mainly hardware calibrated to D.55 and then software luted for gamma and gamut.

I used to calibrate CRT monitors accordingly for D.55.

D.55 is quite a tough target for most displays to manage, It really stresses and shortens the life of a CRT (burns out the red gun). Most LCDs can't even get close.

These days I usually spec a D.65 baseline and a film luminance and LUT the display accordingly.
I found a lot of LCDs magically pass calibration that otherwise choked trying to reach a D.55 calibration.

I also find CRTs that would ordinarily have been consigned to the scrap heap after a couple of years of D.55 usage become perfectly usable with a baseline D.65.

The D.65 baseline also has the advantage of leaving you with a display that can be used for film or video work merely by switching LUTs rather than recalibrating hardware.

A D.55 display usually can't be lutted back to D.65 type requirements.

Not entirely relevant but I offer it up anyway.

I don't see how standards or mastering makes a difference - if the lamp native white is greenish and the standard succeeds in using the lamp most efficiently, don't whites end up being greenish?

No problem with balance. The material is balanced accordingly for the desired whitepoint. Whites will only appear greenish if they are supposed to.

Each DCI projector has its own white point (which varies from one projector to the next), but the DCI content itself does not have a white point, because DCI content is encoded in XYZ color space.

Hypothetically, if your content were encoded in RGB color space, then it would make sense for you to specify the white point in XYZ color space. So, for example, if your RGB content has a pixel with a value of (R=100%,G=100%,B=100%), then that value will be mapped to the white point. However, DCI content has pixels with XYZ values, so if you want to store a particular color of white for a particular pixel, you simply store the XYZ value for that pixel. It is not the white point. It is just a color among many other colors. Ideally, every DCI projector should be able to show that color accurately.

Each DCI projector has its own white point (which varies from one projector to the next), but the DCI content itself does not have a white point, because DCI content is encoded in XYZ color space.

Hypothetically, if your content were encoded in RGB color space, then it would make sense for you to specify the white point in XYZ color space. So, for example, if your RGB content has a pixel with a value of (R=100%,G=100%,B=100%), then that value will be mapped to the white point. However, DCI content has pixels with XYZ values, so if you want to store a particular color of white for a particular pixel, you simply store the XYZ value for that pixel. It is not the white point. It is just a color among many other colors. Ideally, every DCI projector should be able to show that color accurately.It still comes down to the projector outputting RGB. XYZ will need to be transformed to RGB for the projector output the proper image. If the projectors RGB balance (white point) is off, the XYZ based image will be off.

Each DCI projector has its own white point (which varies from one projector to the next), but the DCI content itself does not have a white point, because DCI content is encoded in XYZ color space.Eric, don't quite know what you mean here, since DCI has a clearly defined white point in X'Y'Z' of 0.3794, 0.3960, 0.3890. Using the DCI gamma value of 2.6, this translates to x0.314, y0.351, the reference white point for DCI. This is all spelled out explicitly in SMPTE 431-1, 431-2, and 431-3.

It still comes down to the projector outputting RGB. XYZ will need to be transformed to RGB for the projector output the proper image.
Yes, the XYZ values will be mapped to each projector's native RGB gamut.If the projectors RGB balance (white point) is off, the XYZ based image will be off.
Two different DCI projectors might have different white points. If so, they should have different XYZ-to-RGB mapping functions that are customized for each projector's gamut, so that both projected images will appear accurate, even though both projectors still have different white points.

Eric, don't quite know what you mean here, since DCI has a clearly defined white point in X'Y'Z' of 0.3794, 0.3960, 0.3890. Using the DCI gamma value of 2.6, this translates to x0.314, y0.351, the reference white point for DCI. This is all spelled out explicitly in SMPTE 431-1, 431-2, and 431-3.
As I understand it, the value (X'=3794, Y'=3960, Z'=3890) is specified for a white field test pattern, which is used to check that the projector is producing the standard brightness level of 48 cd/m^2 for that value. The projector is allowed to have a white point that has a different chromaticity and is brighter than 48 cd/m^2, as long as it displays the test pattern accurately. For example, a hypothetical projector that has a D65 white point at 60 cd/m^2 probably has enough headroom to accurately display (x=0.314, y=0.351) at 48 cd/m^2. Also, the DCI content itself may contain other X'Y'Z' values that should be projected at 48 cd/m^2, since the X'Y'Z' encoding for DCI includes some headroom as well. In other words, there are other whites just as bright.

As I understand it, the value (X'=3794, Y'=3960, Z'=3890) is specified for a white field test pattern, which is used to check that the projector is producing the standard brightness level of 48 cd/m^2 for that value. The projector is allowed to have a white point that has a different chromaticity and is brighter than 48 cd/m^2, as long as it displays the test pattern accurately. For example, a hypothetical projector that has a D65 white point at 60 cd/m^2 probably has enough headroom to accurately display (x=0.314, y=0.351) at 48 cd/m^2. Also, the DCI content itself may contain other X'Y'Z' values that should be projected at 48 cd/m^2, since the X'Y'Z' encoding for DCI includes some headroom as well. In other words, there are other whites just as bright.I don't think so. I really think that you are making this more complicated than it needs to be. DCI has a set of references for colorimetry, and it includes a gamut, white point, and gamma function.

Take a look at the document linked below

http://www.avsforum.com/avs-vb/showpost.php?p=15978924&postcount=4

This clearly defines 0.314, 0.351 as the white point for DCI throughout the entire grayscale (see tables 7 and 8 on p. 515) not just at reference level of 48 cd/m2 (14 fL).

This clearly defines 0.314, 0.351 as the white point for DCI throughout the entire grayscale (see tables 7 and 8 on p. 515) not just at reference level of 48 cd/m2 (14 fL).
(x=0.314,y=0.351) is specified for those test patterns. Those test patterns are used to check that the projector accurately displays (x=0.314,y=0.351) at various luminances. Neverthless, the projector's white point is not required to be (x=0.314,y=0.351). The projector can have a white point that is different from (x=0.314,y=0.351) and still accurately display those test patterns.

For example, suppose that a projector has a white point of (x=0.313,y=0.329) when it displays the RGB value (R=100%,G=100%,B=100%), i.e., the projected image is measured to have a chromaticity of (x=0.313,y=0.329) when all three of the projector's color panels are producing their maximum light output. Now, suppose that the projector is displaying (X'=3794, Y'=3960, Z'=3890) in a test pattern. To do so, the X'Y'Z' value must be mapped to an RGB value. Let's suppose that it is mapped to (R=90%,G=100%,B=85%), and the projected image is measured to have a chromaticity of (x=0.314,y=0.351) at 48 cd/m^2. Thus, the projector accurately displays the test pattern, even though the projector has a white point that is not (x=0.314,y=0.351).

If a DCI projector gets a XYZ source, the DCI projector must be calibrated to the correct white point for the XYZ >> RGB transition. It doesn't make any sense to think that a "different mapping" of XYZ to RGB is created for each DCI projector. There is a specific translation formula to go from XYZ to RGB, and if the white point of DCI XYZ source material is corelated/translated to a projector RGB white point of x0.314, y0.351, then all DCI projectors should be calibrated to x0.314, y0.351. This assumes that the XYZ white point translates to 100% RGB.

This assumes that the XYZ white point translates to 100% RGB.
(X'=3794, Y'=3960, Z'=3890) does not necessarily map to (R=100%,G=100%,B=100%) on the projector's panels. (X'=3794, Y'=3960, Z'=3890) should be mapped to an RGB value on the projector's panels that results in (x=0.314,y=0.351) at 48 cd/m^2 on the screen. That RGB value should vary from one projector to the next, and it should also vary over the lifetime of a projector.

(X'=3794, Y'=3960, Z'=3890) does not necessarily map to (R=100%,G=100%,B=100%) on the projector's panels. (X'=3794, Y'=3960, Z'=3890) should be mapped to an RGB value on the projector's panels that results in (x=0.314,y=0.351) at 48 cd/m^2 on the screen. That RGB value should vary from one projector to the next, and it should also vary over the lifetime of a projector.In that case, if (X'=3794, Y'=3960, Z'=3890) translates to (x=0.314,y=0.351) and does not equal (R=100%,G=100%,B=100%), then there is a corresponding white point for a projector at (R=100%,G=100%,B=100%). Once set to this white point an input of (X'=3794, Y'=3960, Z'=3890) should equal (x=0.314,y=0.351).

In any case, when the telecine process is completed to produce consumer DVDs, they are mastered to a white point (R=100%,G=100%,B=100%) = (x=0.3129,y=0.329) and for HD REC 709 color space is used.

Erik

What you've described just reads like very convoluted way of dscribing an correctional LUT on the display itself. It will remap the source material to correct for variance on the display ( no two displays are ever identical).

In any case, when the telecine process is completed to produce consumer DVDs, they are mastered to a white point (R=100%,G=100%,B=100%) = (x=0.3129,y=0.329) and for HD REC 709 color space is used.
Yes, the white point should be specified for DVD content, since it is stored as Y'CbCr values. Otherwise, if it were not specified, you would not know what chromaticity should appear when the signal is (Y'=235,Cb=128,Cr=128).

Which brings me back to my original point, that the DCI content itself does not have a white point, since DCI content is stored as X'Y'Z' values. If you want to project (x=0.314,y=0.351) at 48 cd/m^2, you simply store (X'=3794, Y'=3960, Z'=3890). Or, if you want to project (x=0.3129,y=0.329) at 48 cd/m^2, you simply store (X'=3884, Y'=3960, Z'=4091). The white point is not specified in either case, and it does not need to be specified. You just store the X'Y'Z' value that you want.

Yes, the white point should be specified for DVD content, since it is stored as Y'CbCr values. Otherwise, if it were not specified, you would not know what chromaticity should appear when the signal is (Y'=235,Cb=128,Cr=128).

Which brings me back to my original point, that the DCI content itself does not have a white point, since DCI content is stored as X'Y'Z' values. If you want to project (x=0.314,y=0.351) at 48 cd/m^2, you simply store (X'=3794, Y'=3960, Z'=3890). Or, if you want to project (x=0.3129,y=0.329) at 48 cd/m^2, you simply store (X'=3884, Y'=3960, Z'=4091). The white point is not specified in either case, and it does not need to be specified. You just store the X'Y'Z' value that you want.I understand what you are saying, however it seems that there needs to be a reference point so that when the projector gets a XYZ signal that it can reproduce that correctly. In addition when the projector is calibrated to that specification, there should be no corrective mapping of XYZ to RGB for the projector, and accordingly, whether it receives a XYZ, HD-SDI or RGB signal, it outputs the correct white points. Besides that, the DCI source material is mastered to a specific standard.

I understand what you are saying, however it seems that there needs to be a reference point so that when the projector gets a XYZ signal that it can reproduce that correctly.
There are test patterns that contain certain reference points, and they are useful for calibrating the projector, but I don't think that they establish the one-and-only white point for DCI content, as there is no need to establish one anyway.

The DCI spec states: "The peak luminance as shown in the transfer function equation is 52.37 cd/m2. The extra headroom is reserved to accommodate a range of white points including D55, D61 and D65, while still supporting the reference white luminance of 48 cd/m2 as specified in SMPTE 196E for Digital Cinema-Screen Luminance Level, Chromaticity and Uniformity."

It may be useful to know how these projectors are calibrated. The procedure is to measure R, G, B, and W. The chromaticity coordinates (x,y) for these four measurements are entered into the projector. This characterizes the native primaries and native white. After this is done, the applicable standard like Rec 709 is selected or specific coordinates for the required primaries and white are entered. This information is also part of a PCF (projector calibration file) that comes with each DCI release.

It may be useful to know how these projectors are calibrated. The procedure is to measure R, G, B, and W. The chromaticity coordinates (x,y) for these four measurements are entered into the projector. This characterizes the native primaries and native white. After this is done, the applicable standard like Rec 709 is selected or specific coordinates for the required primaries and white are entered. This information is also part of a PCF (projector calibration file) that comes with each DCI release.
I understand why you enter the projector's native white point, and I understand why you enter an additional white point for Y'CbCr-encoded material, but I don't see a reason why you would enter an additional white point for X'Y'Z'-encoded material such as DCI releases. Is there a reason?

I understand why you enter the projector's native white point, and I understand why you enter an additional white point for Y'CbCr-encoded material, but I don't see a reason why you would enter an additional white point for X'Y'Z'-encoded material such as DCI releases. Is there a reason?

I don't know. Also, I am not sure in which format the processing is done. It's probably in RGB, but I am not sure.

For example, suppose that a projector has a white point of (x=0.313,y=0.329) when it displays the RGB value (R=100%,G=100%,B=100%), i.e., the projected image is measured to have a chromaticity of (x=0.313,y=0.329) when all three of the projector's color panels are producing their maximum light output.Then in that case, the projector would not have the DCI gamut of:
R: x0.680, y0.320
G: x0.265, y0.690
B: x0.150, y0.060

The white point of this gamut? x0.314, y0.351

x0.3127, y0.329 is the white point we are all accustomed to because we are all accustomed to viewing material mastered according to the standards of the 3 widely-used gamuts (EBU, SMPTE-C, and Rec. 709) and they ALL use this white point. DCI doesn't.

The published SMPTE material on this seems pretty clear and unambiguous to me.

"SMPTE 431-1, “D-Cinema Exhibition Screen Luminance Level, Chromaticity, and Uniformity,” is a standard that specifies the reference values and tolerances for the screen luminance level, white point chromaticity, and luminance and chromaticity uniformity of the projected light for the presentation of digital motion pictures in review rooms and commercial cinemas. Because the perception of color and contrast is dependent on the absolute luminance of an image, the goal of 431-1 is to achieve the look in the projected image that will correspond to the look produced during the
mastering process."

The mastering process has certain standards. In order to look right, the display chain should adhere to those standards. Those standards are the gamut described above and a white point of x0.314, y0.351.

I have attached a page of one of the official SMPTE documents on DCI standards (431-2).

Then in that case, the projector would not have the DCI gamut of:
R: x0.680, y0.320
G: x0.265, y0.690
B: x0.150, y0.060

The white point of this gamut? x0.314, y0.351

x0.3127, y0.329 is the white point we are all accustomed to because we are all accustomed to viewing material mastered according to the standards of the 3 widely-used gamuts (EBU, SMPTE-C, and Rec. 709) and they ALL use this white point. DCI doesn't.

The published SMPTE material on this seems pretty clear and unambiguous to me.

"SMPTE 431-1, “D-Cinema Exhibition Screen Luminance Level, Chromaticity, and Uniformity,” is a standard that specifies the reference values and tolerances for the screen luminance level, white point chromaticity, and luminance and chromaticity uniformity of the projected light for the presentation of digital motion pictures in review rooms and commercial cinemas. Because the perception of color and contrast is dependent on the absolute luminance of an image, the goal of 431-1 is to achieve the look in the projected image that will correspond to the look produced during the
mastering process."

The mastering process has certain standards. In order to look right, the display chain should adhere to those standards. Those standards are the gamut described above and a white point of x0.314, y0.351.

I have attached a page of one of the official SMPTE documents on DCI standards (431-2).

The point of this meeting however is to review a possible amendment to this very standard based on field data. Remember it is an official SMPTE meeting and not some charlatan group running around.

Assuming one is blessed enough to have a DCI capable PJ and to be concerned about calibrating to the correct white point, how is the source content distributed to the average Joe in his home? What is the source for DCI content at the enthusiast level? I must be missing something.

Dave

x0.3127, y0.329 is the white point we are all accustomed to because we are all accustomed to viewing material mastered according to the standards of the 3 widely-used gamuts (EBU, SMPTE-C, and Rec. 709) and they ALL use this white point. DCI doesn't.
DCI uses X'Y'Z' encoding that accomodates many whites. For example, if you want D65 at 48 cd/m^2, then you use (X'=3883, Y'=3960, Z'=4092) for those pixels. Or, if you want D55 at 48 cd/m^2, then you use (X'=3893, Y'=3960, Z'=3837) for those pixels. I suppose that you could even use both of those whites within the same image, and some projectors could display both of them accurately at the same time.

The point of this meeting however is to review a possible amendment to this very standard based on field data. Remember it is an official SMPTE meeting and not some charlatan group running around.That's fine. I look forward to reading about the results of this meeting.

What is the source for DCI content at the enthusiast level? I must be missing something.There isn't much of anything, so far as I know. For now the DCI standard is used in professional applications only.

There was an announcement from Sim2 recently about providing a DCI-based system high-end home installations but I haven't heard anything more about it.