Sorry if this has been covered already. I was having a look at ACES V2 this AM using some of the ACES test frames and came upon some unpleasing results on 2 frames.
Note: I tested this with the blink script and a 64 LUT that we created from the CTL.
About the second issue, I think I reported something similar in this thread with the blue gradient example.
This answer was given to me back then :
But I would just expand on it to say that we sacrifice having a nice image appearance for every possible source image “out of the box”, in order to hit the corners and be invertible. You can of course use an LMT which “rounds off the corners”, and will therefore smooth out the appearance for extreme images such as the pure blue example.
Noted. TBH this is the answer I was hoping/expecting to get. Very well aware that we can’t get it perfect on all situations. First image is certainly clipped but roll-off would need to be adjusted then.
I know this is not ideal. There were requirements in the design of ACES 2.0 that had to be taken in account and eventually led to compromises.
What I would be curious to know is if some LMTs are planned in the long course. I think they would be helpful and even necessary in some cases. I brought this topic a couple of times already.
Thanks !
I’m a DaVinci Resolve user, so I’m not sure if those are the same. I don’t have dedicated a blue highlight fixing LMT. Resolve has an ACES neon suppression LUT that seems to take the edge off really extreme colours.
The Resolve Neon LMT is the “blue highlight fix”. It was the initial fix intended for ACES 1 before the more rigid RGC was developed. It compresses quite harshly and tended to affect values in the more usable ranges too much. It is not really intended to be used anymore in an ACES 1.x pipeline and definitely not in ACES 2.0, neither is RGC from what I’ve understood, despite Resolve allowing this in their software.
For ACES 2.0 there currently aren’t any official additional tools to address certain appearance issues, but during it’s development @nick made a LMT JMh Blue Compress designed to operate in JMh which is the model ACES 2.0 uses in it’s rendering. It’s called blue compress but it has a hue slider so it can address any angle really. Might be worth giving a go.
Just wanted to clarify that the RGC remains compatible with ACES 2.0. There are no recommendations either for or against its continued use in ACES 2.
We think that ACES 2 “out of the box” will remove the need for the RGC in most cases because the gamut mapping that is a part of the ACES 2 transforms handles the same problematic colors (albeit in a slightly different way). ACES 2 also uses AP1 to define a number of boundary restrictions. Therefore, using the RGC to map very erratic scene colors into AP1 as it was originally designed for would probably not be a bad thing. I don’t think it would harm anything to continue using the RGC.
Either way, Resolve should certainly not remove the RGC as a potential tool for the user in scenarios where it could prove useful.
Many of the problematic images that previously required the RGC in order to look reasonable through ACES 1 transforms can also look reasonable through ACES 2 transforms without needing the RGC.
