DCI White Point within DCPomatic

Anything and everything to do with DCP-o-matic.
Andy-rit
Posts: 4
Joined: Mon Feb 04, 2019 9:37 pm

DCI White Point within DCPomatic

Post by Andy-rit »

Hello! I'm trying to determine which white point is being used in the color transform of DCPomatic. When in the custom settings page, using input values of (0.33333, 0.33333) for the white point and R(1,0) G(0,1) B(0,0) results in a unity matrix (1,1,1 across the diagonal, i.e. no color conversion). I'm doing this because I have an input image with Macbeth color patches in linear XYZ that I am trying to author with a full bypass of the color transform, but including the 2.6 gamma conversion, to verify projector calibration.

If the above is true, I would assume DCPoMatic is adapting to a white point of (0.33333, 0.33333). Given the older standards and uncertainty/flexibility of the DCI white point, how can I assure I'm transforming to the correct white point if my projector is calibrated to (0.314, 0.351), assuming inputs of either X'Y'Z' 2.6 (our higher tier workflow with the digital source mater requiring no input color conversion in DCPomatic) and a REc709/sRGB workflow (color & gamma conversion required).
carl
Site Admin
Posts: 2548
Joined: Thu Nov 14, 2013 2:53 pm

Re: DCI White Point within DCPomatic

Post by carl »

Hi,
If the above is true, I would assume DCPoMatic is adapting to a white point of (0.33333, 0.33333)
I can't see how this follows from your previous statement... the matrix that is displayed is just the conversion from RGB to XYZ (discussed here in some detail). As you say, if you convert from the space you gave (red = (1,0), green = (0,1), blue = (0, 0), white point (0.3333, 0.3333) that transform comes out as unity but I'm not sure it means anything about the white point that DCP-o-matic is using. I could easily be missing something! ;)

If you arrange for that RGB -> XYZ conversion matrix to be unity and leave the "adjust white point" switched off there isn't so much processing left - just the input gamma, normalization and output gamma, so I think any assurances about white point need to be taken care of in the input files.

We could add a feature to disable the RGB -> XYZ conversion entirely, if it would be useful. What file format are you using to bring in your linear XYZ images?