Unexpected impact of black point on gamut mapping

The black point of the target (SDR) image has a significant impact on the saturation of the gamut after gamut mapping, which seems rather unintended.

Comparison: https://slow.pics/c/8zf3I8ex

The first image is with the default 1/1000 contrast (203 mcd), and the second with a lowered black point at 23 mcd, shot in plplay. The saturation on the first image seems too strong, especially in the reds, giving skin a bit of a too red tint - and of course its very obvious on the red dress.

Watching the same scene with BT.2020 HDR PQ output, while the red dress still "pops" a bit, the skin tones feel much more natural.

I believe this has been shown before on other samples, but I'm putting it here to document and track any potential changes made.

Sample: orientexpress.hevc