R3D HDRx artifacts on reflections and specular lights

@Shebbe can you please try to select ‘Log’ in Color Management → Apply resize transformations
and verify the glicth? this was crucial for me.

Aces 1.3 with ‘reference gamut compress’ option helped a lot to reduce the artifacts but was not enough to definitely solve the issue. It seems this option operates in a different way from ‘Gamut compress’ DCTL. Infact the DCTL was not able to produce the same clean result.

The Magic Motion slider does not produce any effect yet.

This was always set to log in my testings. Now with ACES 1.3 if I apply the reference gamut compression in project settings the issue seems to disappear with image scaling set to Smoother. A few other scaling algorithms still give issues however but much less severe than without the compression.

It could be that gamut compression project wide happens before any image scaling which would explain the differences between that and the DCTL version perhaps.

So we know it’s some negative values issue that happens with scaling whilst combining X/A with Magic Motion. I don’t know how Magic Motion works but it has to be it’s math in conjunction with conversion in and out of ACES. It’s interesting that DaVinci Managed with something like DaVinci Wide Gamut/Intermediate doesn’t suffer from this and even the slider just works as expected then.

If you want me to test anything I’m happy to help but trying to understand/solve it is a bit outside of my knowledge at this point :slight_smile:

Thanks a lot @Shebbe
No, the slider does not work yet, but the A/X blending option do.

  1. original footage

  2. original footage close-up

  3. LayerA grade (about -7 stops)

  4. same grade above but with LayerX merged with Magic Motion (slider to 0.0)

    I shot LayerX 4 stop under.
    So we can create master in ACES OpenExr sequences that are able to keep all those informations in the sky (or other highlights) for further mastering when display technology could be able to show them.

Ah cool so it does blend just the bias doesn’t work. That part was hard to judge on the earlier files as everything was static now I can clearly see the second layer.

this is the frame above