So far I always worked and rendered with Rec709 ODT both for television and DCPs. As I can only see a calibrated Rec709-Image when grading, I think that’s the best way for my workflow. Even for cinema presentation via a P3 capable projector that then of course will only use 709 colorspace…
Now I wonder what will happen when I grade with ODT Rec709 and then output with the DCDM ODT.
As far as I understood that will produce a X’Y’Z’ result. XYZ colorspace with a certain transfer function (that I don’t know. Maybe 2.4?).
So I wonder: If there are colors in my camera RAW footage, that lie outside the 709 primaries and the Rec709-ODT was somehow mapping them into gamut when working on the grade - they will again be present in the X’Y’Z’ output.
Because when switching to DCDM, the ODT part will get the real ACES linear colorspace, go through the same RRT as before (I guess) but then doesn’t have to map or clip or limit certain colors. So I get more, but I cannot see and control it.
And further: What will happen when the DCP is created. It gets X’Y’Z’ with (in extreme cases) colors even outside P3 and does … what? Clipping or some kind of desaturation or other things?
There are in fact two DCDM ODTs in the ACES repository – DCDM and DCDM_P3D60. The latter clips to the P3 gamut. The former is unclipped, and any gamut clipping is left to you.
An additional consideration is that of white point. The DCDM_P3D60, as the name implies, is intended for an “assumed observer adapted white” of D60. This means that it maps R=G=B in ACES to D60 white in X’Y’Z’. But if you were viewing Rec.709, with its D65 white, when grading, then you probably don’t want a D60 DCDM, unless you graded using the D60 sim version of the Rec.709 ODT so your eye/bran combination was adapted to D60 when you were grading. Some ACES implementations include alternate ODTs for different “creative white points” such as D65. This is because DCI does not mandate a white point, and allows you to master with any white between D55 and D65.
Though Nick mentions them, there are three things to note – conversion from a D65 master to X’Y’Z’ requires a different matrix that what is used, but if you are using D60sim, things should work well. Also, the Rec.709 ODT assumes a dim-surround correction, so if you are grading in the dark on a Rec.709 monitor, the DCDM version may appear too contrasty. Lastly, if you are not up close to the Rec.709 screen and thus match the viewing angle of a cinema, you may feel you have satisfactory contrast but then in cinema viewing, it will not have the contrast you want unless you are really far back in a theater.