PR: ACES 2.0

[KelSolaar]

Starting a PR for the ACES 2.0 work, I haven't even tried to build the configs but the classifying code is working, so did a solid good pass at updating the spreadsheets:

• Reference: https://docs.google.com/spreadsheets/d/1z3xsy3sF0I-8AN_tkMOEjHlAs13ba7VAVhrE8v4WIyo/edit?usp=sharing
• CG & Studio: https://docs.google.com/spreadsheets/d/1PXjTzBVYonVFIceGkLDaqcEJvKR6OI63DwZX0aajl3A/edit?usp=sharing

Current build artifacts can be found here:

Reference

• reference-config-v2.3.0_aces-v2.0_ocio-v2.4.ocio.txt

CG

• cg-config-v2.3.0_aces-v2.0_ocio-v2.4.ocio.txt
• cg-config-d60-views-v2.3.0_aces-v2.0_ocio-v2.4.ocio.txt
• cg-config-all-views-v2.3.0_aces-v2.0_ocio-v2.4.ocio.txt

Studio

• studio-config-v2.3.0_aces-v2.0_ocio-v2.4.ocio.txt
• studio-config-d60-views-v2.3.0_aces-v2.0_ocio-v2.4.ocio.txt
• studio-config-all-views-v2.3.0_aces-v2.0_ocio-v2.4.ocio.txt

[KelSolaar]

First artifacts are here: https://github.com/AcademySoftwareFoundation/OpenColorIO-Config-ACES/actions/runs/10648206791

I haven't checked anything yet.

[carolalynn]

Amazing Thomas - just want to note a few quick things:

CG Config:

• We had a few more views / displays in previous versions:
  • PQ P3 1000 nit Rec. 2020 lim
  • PQ P3 1000 nit P3 lim
  • SDR Cinema

Studio config:

• Name of look transforms seem off, the gamut compression is named "ACES2065-1"
• Did we decide to split the D65 / D60 into two configs?

General:

• talk about "extended" displays for Display P3 etc? Might want these in both configs.
• will do a more detailed naming/alias check later once CIF decides on names

[doug-walker]

I added three of the missing display built-in transforms to your spreadsheet (only one still missing).

Looking at the color spaces, I'm noticing both "Apple Log" and "AppleLog BT2020", despite the latter not being in the spreadsheet.

[KelSolaar]

I implemented support for:

• Passing the encoding as a suffix to the DisplayColorspace name
• Excluding the D60 views although we need to discuss how we want to name the Configs

[doug-walker]

@KelSolaar , thanks for adding the ability to have an SDR and HDR version of DisplayP3! Here are some comments looking at the most recent Studio config artifact:

• I like your use of the "SDR - Display" and "HDR - Display" suffixes on DisplayP3. However, I'm wondering if some people will complain that some display names indicate range and others don't. An alternative would be to leave the SDR displays as "Display" and change all the HDR displays to "HDR Display". But on the other hand, indicating the range on just those displays that have both could be considered a benefit. Minor point, but do we need the second dash, perhaps "Display P3 - HDR Display" would be sufficient?
• If we do use the proposed "Display P3 - SDR - Display" name, we should add the previous name and "displayp3_display" aliases for backwards compatibility.
• It looks like for all the HDR displays, we are include the SDR ACES views, to have them available for soft-proofing. So it's expected to see the SDR displays in the HDR version. However, please remove the HDR views from the SDR display.
• This config continues to have both the D65 and D60 views (I guess the new option to remove D60 was not in effect?). However, the Display P3 views are a strange and incomplete mix of D60 and D65.
• Both display color spaces have the same set of ACEStransformIDs. Not sure what to do here. Probably we should only include the SDR views with the SDR display, and HDR views with the HDR display. But just wondering if there would be scenarios where it's safer to have all of them?

[doug-walker]

I had an action-item from the last meeting to check the ACES v2 config to see if we have all the color spaces from the ACES v1.3 configs. Here are my findings:

• For the named_transforms, "ARRI LogC3 - Curve (EI800)" is missing.
• All scene-referred color spaces are present.
• For the display-referred color spaces, the displays that are not included in the ACES v2 set are missing, these are: "Rec.1886 Rec.2020 - Display", "P3-D60 - Display", and "P3-DCI - Display".
• There are a lot of aliases that are missing. I guess we should discuss whether we want to include all of them, or use this as an opportunity to prune some of the older ones.

Not sure what to do about the displays. On the one hand, they are not very relevant and I'm open to dropping them. On the other hand, I'm just wondering if anyone will complain about missing displays when moving from ACES v1 to v2.

[KelSolaar]

Hello,

Thanks for the thorough review!

1

Minor point, but do we need the second dash, perhaps "Display P3 - HDR Display" would be sufficient?

The second dash is coming from the - Display suffix that we consistently add to all the display colourspaces. It feels a bit arbitrary to add that to Display P3 when none of the other HDR display has it. What about Display P3 SDR - Display and Display P3 HDR - Display?

2

If we do use the proposed "Display P3 - SDR - Display" name, we should add the previous name and "displayp3_display" aliases for backwards compatibility.

Yes, good catch!

3

It looks like for all the HDR displays, we are include the SDR ACES views, to have them available for soft-proofing. So it's expected to see the SDR displays in the HDR version. However, please remove the HDR views from the SDR display.

Where did you see this? The only instances I saw are occurring for Display P3 which is "expected" because we use the same builtin for both HDR and SDR. It would be much cleaner to have a dedicated Display P3 HDR builtin as suggested in previous meetings:

• I would not need to use that new gross encoding column that we added for Display P3
• The generator would work perfectly fine as it was intended

I'm happy to make a PR for this on the OCIO side.

4

I guess the new option to remove D60 was not in effect?

Yes, I will check what happened for Display P3, maybe the quick filterer I wrote is not "smart" enough or there are yet another problem because we introduced that special case as mentioned earlier.

5

Both display color spaces have the same set of ACEStransformIDs. Not sure what to do here.

It might be time to source this AMF data from the ACES side and stop generating it on our end.

6

For the named_transforms, "ARRI LogC3 - Curve (EI800)" is missing.

I will check, it might not even be in the spreadsheet.

7

the displays that are not included in the ACES v2 set are missing

If they are not in ACES 2.0, they cannot be in the config at the moment.

8

There are a lot of aliases that are missing.

If you have a list we could talk about it yep!

[doug-walker]

@KelSolaar , I looked again at the alias differences I reported above and found a bug in my comparison script. The changes needed are actually very minimal. Please simply add "lin_ap1_scene" to ACEScg.

[doug-walker]

@KelSolaar , if you would like to add the additional DisplayP3 HDR built-in, you would make a second call to registry.addBuiltin here.

Suggested name: "DISPLAY - CIE-XYZ-D65_to_DisplayP3-HDR", but I'm flexible.

[KelSolaar]

OP updated with the latest config artifacts!