White point SHIFT

(Charles Boileau) #1

Hi to all,

I came into a little “issue” with the way ACES shifts the WP in P3-DCI.

I’ll try to make this simple…

I graded a movie with IDT: ALEXA and ODT: DCI-P3. As we all know the DCI-P3 WP is greenish and ACES shifts it towards something closer to regular D65 (I presume). This shift in WP is rendered onto the imsges once we “exit” ACES.

There was a concern towards the end of the online process that asked that certain rendered shots be double checked in my grading project. We proceeded to import the images on a separate track to verify if the grade was identical. Something that we used to do often to make sure a master was A1.

But even when I flag DCI-P3 as IDT, ACES shifts the white point. Causing the new images to be too pink.

Any thoughts on this? I tried with the P3-D65 and it’s not a solution.

Should ACES consider adding a ACES P3 IDT for this purpose?

Thank you very much!

(Jim Houston) #2

Hi Charles, I wonder about the description you are using. ACES renders to a D60 white point
and in the case of the DCI-P3 ODT uses codes that will look correct on a P3-DCI calibrated
projector. But the value of grey if it is not clipping to the edge of the gamut will be a D60 white point. So ACES itself should not be ‘shifting the white’, but it is making unequal RGB values to get to the D60 white on a DCI projector. There is some roll-off and scale at the top end of the
ODT for P3DCI to prevent harsh clipping.

Not sure which software you are using, so I don’t have a real comment to make. It does
sound like an interaction of the track color setups between two different tracks. An IDT would
not work here I think, an inverse-ODT for P3DCI could perhaps.

I hope these clues are useful… at least they are quickly given.

I’m sure it can be narrowed down with a little more detail on the color setup for the tracks.


(Alex Forsythe) #3


What I think you’re seeing here is that that ODT.Academy.P3DCI_48nits.a1.0.3 reproduces neutral (R=G=B) ACES values with the chromaticity x=0.32168 y=0.33767 (aka D60) even though the projector is calibrated so that equal RGB projector code values yield a chromaticity of x=0.314 y=0.351.

In other ODTs we call this “D60 sim”. Unfortunately, it’s not clear from the name of this particular ODT that’s what’s going on. The only hint is in the header comments of the CTL.

// Output Device Transform - P3DCI (D60 Simulation)

// Summary :
//  This transform is intended for mapping OCES onto a P3 digital cinema 
//  projector that is calibrated to a DCI white point at 48 cd/m^2. The assumed 
//  observer adapted white is D60, and the viewing environment is that of a dark
//  theater. 
// Device Primaries : 
//  CIE 1931 chromaticities:  x         y         Y
//              Red:          0.68      0.32
//              Green:        0.265     0.69
//              Blue:         0.15      0.06
//              White:        0.314     0.351     48 cd/m^2
// Display EOTF :
//  Gamma: 2.6
// Assumed observer adapted white point:
//         CIE 1931 chromaticities:    x            y
//                                     0.32168      0.33767
// Viewing Environment:
//  Environment specified in SMPTE RP 431-2-2007

It’s pretty easy to make one that does what you want but which color grading system are you using so we can make sure we can put it into your pipeline?

(Charles Boileau) #4

Hi Guys,

Thanks for the feedback. From what I’m reading it seems that you are understanding the problem. I’m not a hyper technical person but I think I get it.

So to be more clear. I’m working with Resolve 12.5.5. So I guess I would need a custom DCTL? And what you’re saying is that I would need a DCI-P3 (D60 sim) ODT (which does not exist)? And what does the “sim” really mean? Is it not “rendered” in the image?

But this opens a more global question: Are we currently doing the right thing?

We have a DCI projector (Christie CP2220) calibrated to DCI white x=0.314 y=0.351. Are we doing the right thing by using the P3-DCI ODT?

And I was flipping thru the ODTs and saw that there was a DCDM ODT… Aren’t most (if not all) DCDM in DCI-P3?

I’m a bit confused…

Thanks for the help!

(Alex Forsythe) #5


I think you’re doing what one would assume is correct based on your goals and our component names. The problem is the transforms doesn’t do what exactly what you’d think it would based on the name.

Our P3-DCI ODT has a D60 sim in it, but that’s not noted in the name. We haven’t provided a P3-DCI ODT without a D60 sim in it, but it sounds like that’s what you want. You’re not the first to get tripped up by this.

The DCDM can contain movies that were mastered relative to any primaries with any white point. You’re correct that most movies are mastered with P3 primaries and the DCI white point. The thing is that DCI white point is very distinctly green. Most manually push the image to a white point that’s more pleasing via color grading, but it’s important to note that depending on you exact projector setup it’s highly likely it will always clip to that green white point when using a P3-DCI setup.

We recommend to people they calibrate the projector to P3 with the ACES white point (aka D60). We have a PCF for that. Then you can use the P3-D60 ODT. The P3-DCI ODT we included in ACES 1.0 will produce the same look as the P3-D60 ODT when each is used in combination with their respective projector calibrations. What you’re asking for different look on the P3-DCI calibrated projector. Very do-able, but you’re likely going to push the color in the direction of where the P3-D60 and P3-DCI ODTs place it by default. It’s worth noting that just as you can push the P3-DCI to a different creative white, you can push the P3-D60 look that comes out of the ACES 1.0 transforms to a different creative white too.

I hope I didn’t just make this more confusing …

(Charles Boileau) #6

No you’ve actually made it clearer!

We did notice that the WP changed when we first started to use ACES. We figured it was to “cleanup” the green WP. Which, personally, doesn’t bug me that much.

But, could you please specify what you mean by: “but it’s important to note that depending on your exact projector setup it’s highly likely it will always clip to that green white point when using a P3-DCI setup.”

Do you mean here that in color grading your simply “virtually” adjusting the WP? The WP will always remain “in place”?

If so, I don’t see this as something wrong. It’s just part of the process… Would it be correct to say that?

And, what would be the advantage of using the PCF file VS just using the DCI-P3 ODT? Isn’t the result just the same?

Finally, what I was asking for what a way to bring images that we’re graded in ACES and rendered with the DCI-P3 ODT back in to Resolve to compare them to the original source material that has grades on them. I would not alter these images…

And, please indulge me… I never found an explanation as to why the green WP in P3. And also, if this was defined this way why do people change their WP to D65 or D60. Why not just “deal with it”?


(Alex Forsythe) #7

If your projector is calibrated such that R=G=B produces x=0.314 y=0.351 than the way to get another creative white is to make objects in the content you want to appear neutral have non-equal display code values.

Let’s say you have a grayscale that you’ve deliberately graded to have non-equal display code values because you wan that grayscale to be something other than x=0.314 y=0.351.

At some point the grayscale is going to hit 1 in display code value space in one of the channels and the white point will begin to shift … Then it will clip in 2 channels … Eventually all 3 channels will hit 1. That means that point in the grayscale will be x=0.314 y=0.351 because that’s what the display calibration dictates.

I think the answer is yes.

I think so … assuming you’re aware of the white point is going to shift back to the display calibration white point once the display code values max out in each channel.

The result should be pretty darn close. The P3-DCI ODT “D60 Sim” operation has a very slight roll off at the top so you don’t get that drift I was talking about above, but it means your max luminance out of it will be slight less then the projector max luminance.

(Full disclosure, this is what I believe happened based on piecing together the story over years, but I wasn’t around in the early days of digital cinema to say for sure)

I think the reason we’re stuck with the green white point is a bit of an “accident” of history.

When they were first developing digital cinema projectors max luminance was an issue and a lot of work went into getting to 14 ftL on screen. Obviously, they measured the luminance at a chromaticity that yielded the most light. That just so happened to be x=0.314 y=0.351.

When the projectors shipped someone built a PCF that was called P3-DCI and used that white point so the maximum possible luminance on screen was achieved. Nowadays getting to 14 ftL isn’t an issue, even at a variety of white points but people just know “Hit the P3-DCI button”. Likewise, downstream hardware and software that make DCPs out of P3 files pretty much universally assume the white point of those files is DCI so there’s a whole ecosystem of tools built around that unfortunate white point.

If you know what you’re doing, you can easily modify your pipeline to deal with another white point, but it takes some thought and work.

(Dean Du C.S.I.) #8

Hi Charles
You use ACES ODT P3-DCI in Resolve 12.5.5
After the output found color and timeline, the color is inconsistent
This problem is not ACES itself is the problem, is Resolve12 version of the Bug.
This question has been fed back to Resolve’s engineers before, and engineers have been repaired in Resolve 14,
I have been tested in Resolve 14, P3-DCI discoloration problem has been fixed

(Charles Boileau) #9

I’m not sure I follow @Dean… Are you saying I wont see the same problem in 14?


(Charles Boileau) #10

Thank you @Alexander_Forsythe! Very informative.