Inconsistent ACES results

Discussions - Using ACES
21

HLG_ACEScct
This is the curve applied by the inverse HLG Output Transform. If you do no grading, and then simply apply the same forward transform, you will end up back where you started. But you can see that the highest HLG code values are very stretched out by the steepness of the top of the curve. Grading could make that stretching visible, creating undesirable artefacts.

This, on the other hand is the result of the Input Transform (to ACEScct) I proposed earlier:
HLG_IDT
(Exposure adjusted so 38% HLG maps to ACES 0.18, or ACEScct 0.413588)

22

Thank you for your explanation. I indeed experiment unexpected behavior to high values using Output HLG as IDT when grading.

How could I get or generate this HLG IDT? could I just edit config.ocio and create a new HLG item without the RRT or this is more complex to do?

23

You need to make an IDT based on the REC. 2100 OETF. The HLG ODT is based on the Rec. 2100 EOTF, which is not an inverse of the OETF. So it is more complex than just removing the RRT.

I have experimented with creating an IDT in DCTL for Resolve, which does what I described. It is untested on real footage, so not something I am happy to post publicly at the moment. Contact me directly if you are interested. Its result could also be baked into a LUT to be added to an OCIO config.

24

I have just added an HLG (camera) DCTL IDT to the New Products page on my website, if it’s useful to anybody. It includes an exposure adjustment, which by default maps mid grey to ACES 0.18.