Hi everyone,
First of all. thanks steve for building this space, It has been almost a year i'm documenting over the web, trying to find answers, i just collected more and more uncertainties...

Ok, this is one of my first experiences, where i have to create a DCP after grading. I got a JVC DLA-X70R. white point is (quite close to) x 0.314 y 0.351 mesured with lightspace and a Hubble probe, as the DCI spec says.
Then went the check in a theater. Happy with the contrast, i just noticed that my picture was a bit warmer. So I mesured in this theater the with point, gamma and gammut with a bunch of patches.
Results : gamma 2.6 : correct, gammut correct and...white point : x 0.322 and y 0.362 which corresponds to approximatively 5900K whereas DCI white point is about 6300K. So, i was happy, this explains the wamer picture.

But...by chance, the theater was calibrated a few days before i came for my tests, and i was able to speak to the guy who was doing the calibration. He told me that the white point for the calibration was x 0.314 y 0.351 : 6300K
... .... ...
I have to tell, that this guy didn't seem to be unexperienced at all, i'm curently waiting for his input about this post. I'm sure he did that in purpose.

So, here is my theory :
In lightspace, there is 2 kind of DCI, the D55 and the D65 (i work with the D65, by the way, why do we call it DCI D65 since the white point is 6300K...). Here is my guess : Since there is this 2 type of standarts, 5900K is just in the middle of 5500K-6300K, this is the best way no to disturb anyone since there are states that use one of the standart instead of the other.

Here is my questions :
- does anyone experience this kind of white point shift in their reference theaters?
- does anyone agree with my theory?
- Is there areas (like PAL SECAM and NTSC) where DCI D55 is advocated instead of DCI D65?
- what about the third DCI template in lightspace : the DCI XYZ? I understood that it does the XYZ conversion, but what does become the white point ?

Thanks for sharing your experiences

Hi,

The correct DCI P3 white point is the one 'we call' D65 within LightSpace.
The D55 entry is a 'special' we added for a particular customer, and we use D65 and D55 as a way to define them.
And you are right, D65 is NOT the correct term...

But, DCI projection is XYZ, which has a different white point, as shown within LightSpace.
This is the correct calibration for a DCI operation using true DCI standards, but is not the same as P3.

Any help?

Steve

Thanks steve,

As i understand, the calibration performed by the operator in the theater using a 5900K white point is right so what is the point of giving a reference white x 0.314 y 0.351 6300K in the DCI spec? This would mean I'm wrong calibrating my projector in 6300K ?


I'm sure this kind of question has been asked a lot but you have to agree that this is hard to understand : grading is done in DCI P3 (white point 6300K), in theater DCI XYZ white point is x 0.3333 y 0.3333 (around 5500K), Does the XYZ conversion compensate this white point shift...? If DCI XYZ is 5500K, why a calibration at 5900K???

AAAAAAAAAAAAAaaaaaaaaaaaaaaaaaaaaahhhhhhhh!
Once again, i'm lost, but this happens so many times before all becomes clear again.
I'm feel i miss somethink about the XYZ theory...

The white points are different, as it the gamut.
Any material graded in P3 will need conversion into XYZ to be able to be shown correctly on a DCI compliant cinema projector.

So, you need to know what colour space and white point your material is set for - as well as the display/projector.
Make sure they are correctly matched.

P3 ans XYZ space are different - it's that simple

Steve

OK i think i begin to understand.

I'm ok with the P3 to XYZ conversion, I need it for my DCP. I understand that it does the conversion (gammut converted, white point from 6300K to 5500K, keep gamma at 2.6).
what i still don't understand is why my mesure in the theater is at 5900K while DCI XYZ is at 5500K?

I start to think that my observation of a warmer image in the theater (due to, i thought, my calibration at 6300K and the white point mesured at 5900Kin the theater) was an illusion...

jb

You should not really use 'K' figures to mean white point.
White point is a specific xy coordinate.
K values are temperature, and can be different white points.

But, regardless, it sounds like the calibration was wrong...

i think i see what you mean,
in CIE diagram, while following the line of theoric whites, we can say that the y axis have less influence on the equivalent color temperature than the x axis. Considering that white lights are a "zone" on CIE diagram, different xy coordinates could have the same equivalent in color temperature.
Am i right?

back to my mesures, DCI P3 is x 0.314 y0.351, mesure in theater is x0.323 y0.355(shift in the red/orange area), mesure in my grading room x 0.317 y 0.329 (shift in the magenta area) this would explain the warmer image...
And when we observe this green shift on XYZ tiff before encoding DCP this come from that DCI XYZ white point coordinate is x0.333 y0.333 (shift in the green compared to DCI P3 white point). Right?

(ok, sorry this is not the same values i gave at the begining of my post.....)

Yep, all correct to me

may i ask a last question...
this would mean that in order to be right for my DCP, i should use a lut (using the convert color space within lightspace) form the mesured gammut and white point of my projector direct to DCI XYZ preset. No need to apply a lut from "my projector" to DCI P3 plus export tiff XYZ within da vinci resolve?

No - use the pre-set matrices.
Your specific display is not relevant to the conversion process.

So, Rec709 to DCI-XYZ, or P3-D65 to DCI-XYZ

(Assuming your display is accurate after calibration - if it's not, ie. it can't do P3 accurately - that's a different story)

Steve

all right,
I have an appointment in the theater tomorrow morning, i'll keep you in touch with my tests.

thanks a lot

jb

Also you might want to consider that the Hubble is a colorimeter and not a spectro, so the profile used might have an influence and should take account of the light source, which is significantly different between a DCI projector (Xenon lamp) and a JVC (UHP lamp)

Cedric

Hi cedric,
thanks for your input! For me, this means that i have to go back (once again) into the theory of probes, i realize that, for now, i don't have enough knowledges on this subject.

I just come from my tests in the theater this morning with the DOP a the movie. I prepared 3 differents tests with 3 different luts created within the colorspace conversion of lightspace.

1. a simple DCI P3 to DCI XYZ LUT (i knew that with this one my image in the theater was a bit warmer than in my grading room)

2. a DCI P3 to DCI XYZ LUT but i shifted the white point of the DCI P3 to be the one i have in my grading room

3. a last one with the RGB coordinates of my projector and his white point to DCI XYZ (i know that compared to DCI P3, my red is quite good but blue and green are a bit under the DCI P3 colorspace, i don't achieve to bring them on the theoric DCI P3 values)

Conclusion, the 3rd LUT is the good one for me, it brings back the saturation and the warmth of the picture to what i saw in my grading room.
The 2nd one (with only the white point shift) didn't do what i expected. I didn't see any real differences compared to the first LUT (this is still a mystery for me)

OK I know that it would be fare better to grade on a proper calibrated projector or to bring back my projector in a rec 709 colorspace and work with it but my projector can do better than rec 709, so...i have to say, i'm quite happy with this 3rd lut...

Any way, thanks Steve for all your inputs, the great lesson for me was : "in calibration, never use Kelvin while speaking of white point"

thanks again!

If you are using a display that is not capable of meeting the specifications for a given colour space, all bets are off!
This is a very poor way to grade, and cannot be recommended at all.

If a display cannot meet P3 correctly, stick to Rec709 and convert from there.
Using a display that is not accurate is not acceptable for grading - ever

Steve