A year ago I pushed in bunch of tone-curve LMTs and I’ve now added couple more. I’ve also pushed an ACES1 “colorimetric” LMT for ACES2 which gives ACES1-like colors through ACES2 transform, as well as ACES1 emulation LMT which gives ACES1-look through ACES2. I’ve also added the ACES1 SSTS (HDR) tone curve as a look since someone wanted it.
As was the case the with the previous LMTs for ACES1, the match with the new LMTs is quite good for normal surface colors but won’t match for highly saturated and super bright colors. The LMTs are freely available as CLFs, DCTLs and LUTs.
Here’s an example of the full ACES1-look emulation:
I’m hoping to have some time to add new LMTs in the coming weeks for ACES2.
I noticed also that the new C-ACES1 LMT seems to work quite well when combined with LMTs originally developed for ACES1. Here’s an example of matching ARRI K1S1 emulation look developed for ACES1, through ACES2. I’m doing that by applying first the K1S1 tone-curve LMT for ACES2, and then the new C-ACES1 LMT for ACES2 and then the C-ARRI_Classic LMT developed for ACES1. Here’s couple examples of this compared to the M-K1S1_Classic emulation through ACES 1.3 and the real ARRI K1S1:
I am not satisfied yet but I am slowly getting there… It would be nice if you could run some of the most “problematic” examples (blue bar, red xmas…) through your best LMTs.
Here is what I am especially paying attention to:
Lack of smoothness in the reds (there is like a line from a clip possibly ?)
Lack of purity attenuation in the yellows (probably breaking g0 threshold)
Problematic area in the blues (weird hue shits and harsh line)
I have looked at this image a lot in the past months. I don´t know how it was generated but it has been very helpful to evaluate things (I call it the “wavelength” image).
You can see like a blue outline “flip” as exposure increases. I cannot believe I did not spot that in the past two years. It was right in front of my eyes.
The LMTs don’t specifically address any “issue” per-se. In fact, the outer edges of the LMTs might not have any “look” in them at all, and you might just see the pure underlying transform, ACES2 for example. The C-ACES1 LMT does add some of the red->orange skew. It doesn’t have much of blue-treatment at the moment.
I created that image as a stress test but I can’t claim I know how it should render. Hopefully smoothly of course. It’s a visualization of “dominant wavelengths” in 5nm steps in full ACEScct range (full half-float range starting from 0). Nuke node of it available at DominantWavelength.nk, but I think Nick made a better Nuke version of it, available somewhere. There’s also “Planckian Locus” Nuke node if you want to look at the orange skews, or lack of, PlanckianLocus.nk. There’s other stress test exrs there too…
I think we did look at similar issue like that with one of the neon sign images in the Academy image set during development that exhibited a dark blue band when underexposed (in frame #33 I think).
I seem to have another version of the dragon-scene image that looks different. I do see it in the version in the Academy image set.
