Output Transforms Architecture VWG - June 21st, 2023

The recording of meeting #107 is now available.

1 Like

Good morning @nick ,

you were of course right, I did not directly interpreted the AgX BT.1886 output as HLG.
I did the follwowing steps in Nuke before I wrote out the file. I am not 100% sure if this is the right way to do it. But in the comparison to the other images in the clip that I shared it looked right to me.



That makes sense, from the perspective of seeing a reasonable HDR image from the HLG, as you now have a Rec.2020 SDR image you are interpreting as HLG, and the log part of the HLG curve is similar to a highlight compression, so the HLG EOTF will expand that to the display. You’re basically doing “fake HDR” with inverse tone mapping.

Because it is an accidentally similar curve, rather than a deliberately matched one, it is possible that some values could get stretched out into the HDR range when they should not be. And I would be interested to know how much of a perceptual match you get between the SDR and HLG, as there is not anything in this approach which deliberately targets that.

I don’t know if this helps, this is a translations of troys latest Agx as a dctl:

this is my latest test (currently working on one that outputs rec2020 and p3d65 too)

And this is another dctl test by Troy that forms the image at wider gamuts.

1 Like

The notes from meeting #107 are now also available.