ACES .icc profile white-point tristimulus values

I’m doing some experimentation with ACES .exr renders in After Effects CC 2017, and I’m curious if anyone can help me figure out what’s going on here with the white point numbers from different available ACES .icc profiles.

The first one in the upper left is an ACEScg AP1 profile from Apple, and came with OSX 10.11.6 I think. The second is an ACES2065-1 AP0 profile from Fnord, the creator of the AE OCIO plugin.

The second two on the upper right are both ACEScg AP1 profiles. The first is Adobe’s and comes with After Effects, and the second is from “Elle Stone’s Well-Behaved ICC Profiles”.

The last two on the lower right are ACES2065-1 AP0 profiles. The first is Adobe’s and comes with After Effects, and the last is from Elle Stone.

The two texts on the lower right are from the AMPAS technical documents for ACES2065-1 and ACEScg, and seem to show white point values that match the two profiles on the upper left, but none of the others.

Can anyone confirm who is right here, at least in terms of this one value? I’m sure there’s some other surprises hiding in the rest of the profiles.

The [ 0.96429568, 1. , 0.8251046 ] whitepoint is actually D50, I have not checked but I’m fairly certain that Adobe & Elle .icc profiles have been chromatically adapted to D50 which is the default ICC PCS whitepoint: they are defining the same RGB colourspaces but using a different reference whitepoint. If you were to check the R,G,B primaries they should be different from the ACES specification ones.