Using a Sony PXW-Z90, I set PP1 as SDR REC709 manual knee, and PP2 as HDR HLG/709 auto knee (auto knee required for HLG). I then turned gain to max 33db, started pointing at things that would blow out and clip the highlights, and toggled between PP1 and PP2. I laid the clip on a standard SDR REC709 timeline, and exported without any grading or color space changes, as it was set to 709 instead of BT2020 under HLG color space selector. As you can see, the HLG1 (Sony 87%) PP just sees things that the REC709 doesn't. Exposure was locked down, so no exposure changes were happening when I toggled the PP's, you are seeing the HLG gamma doing some sort of magic...W.O.W.

https://youtu.be/IDGM11NSoFM

I've been thinking about HDR HLG and its use in REC709, your example is perfect.
I wonder if metadata in a recorded HDR HLG file then edited in a REC709 seq and exported as REC709 has any effect on a BT2020 TV?

Color space is encoded in the metadata, so you shouldn't get ant weirdness if you throw an HLG/709 or project/export/render REC709 file at a normal HLG BT2020 capable TV. The project (FCPX wording) color space determines the encoded metadata for colorspace.

Paul

Thanks Paul, that's what I thought, it's weird, I kinda like the look of HDR HLG as it is for some things.

Apparently it takes all the raw data that the KNEE just compresses and throws away, and somehow saves it as a form of log and magically keeps it visible in the REC709 codec?. :)

Paul