Using correct IDT but getting a mismatch to reference cut

I received this plate which is in LogC3 (I was on set for supervision to take note of this) but in post, I’m getting a mismatch between what I see in my After Effects window vs a reference cut the editor sent to me, which he mentioned has rec709 LUT on. It appears almost as if there’s a double gamma being applied. Below is a screenshot showing this issue; image 1 is my After Effects window with the view transform set to ACEScg to rec709, and image 2 is the reference cut.

What’s really strange is when I put the same raw plate into Premiere and used a LogC3 to rec709 lut there, it appears to match the reference cut… leading me to believe that something funky is happening in After Effects between the way it’s interpreting it and showing it in rec709 after putting it through ACEScg.

Any help would be greatly appreciated! :slight_smile:

Hello, as much as it’s still possible there is something misconfigured in After effect (with the few context we have), I actually don’t think there is any configuration issue and the difference you are seeing is the result of comparison 2 different image formations algorithm.
ACES v1 is known to have trouble handling bright saturated source of light which seems to be the case in the sample you sent.
The plate being from an ARRI camera and you mentioning a rec709 LUT I assume that the “reference” is probably encoded using the ARRI Reveal pipeline or its older version (with a LUT from this or this).

Here is a visual comparison using the Blue Bar image:

As you can see, blue lights get pretty nasty.

One of the first way to ““fix”” this with ACES is to use the “Reference Gamut Compression” look available in 1.3, which will slightly improve those highlights (no idea how to enable this in AfterEffect).

However if you really need to match the reference I would drop ACES and just use the ARRI pipeline, but this all depends on your goal and constraints you are given.

To further help with your choice of algorithm I have this website where I compare multiple image formations algorithm including ACES 1.3 and ARRI Reveal: