ACES ODT Candidates for ACES 2.0

[rand thompson]

Keeping on the car for a moment Blair, how's this?

You can open that in a new tab for a closer look. Didn't include C because it's just too out of whack. Only providing two of my options here because I provide a few things. This actually is some new stuff that I'd be curious to get feedback on. I made a custom highlight roll-off to the ACES curve to somewhat get the car and cone to render better.

I've now had a full day to look at everything on the ACES side. It to me is still a work in progress and I still can't recommend the previous workflow due to color artifacts. Hue related ones.

I can't stress enough. RED, Graeme really, put a great deal of time into doing this pretty right.

Thanks for doing this! I like the RED IPP2 Med/soft and both the ACES A candidates/with HRO. How Is A looking in the VECTORSCOPE with the additional custom HRO?

 

[rand thompson]

Here's a comparison between the ACEScct workflow using the ACES 2.0 Candidtate A, RED IPP2 MEDIUM/SOFT and Davinci Wide Gamut

ACEScct with ACES CANDIDATE A REC709 ODT

RED IPP2 MEDIUM/SOFT

DAVINCI WIDE GAMUT

IMHO,

I always thought that RED's IPP2 Medium/SOFT was a little "HOT" and Bloomy in the Midtones in relation to say Davinci Wide Gamut and Now even ACES with the ACES 2.0 CANDIDATE A REC709 ODT


ACES with ACES 2.0 Candidate A REC709 ODT

ACES with ACES 2.0 Candidate A REC709 ODT with +4 1/4 printer lights

RED IPP2 MEDIUM /SOFT

DAVINCI Wide Gamut

Plus, although RED's IPP2 MEDIUM/SOFT has a more controlled Highlight ROLL-OFF than the other two Workflows they seemed "Bloomy" as well.


ACES with ACES 2.0 Candidate A REC709 ODT

RED IPP2 MEDIUM /SOFT

DAVINCI Wide Gamut

 

[Blair S. Paulsen]

Car shots: FWIW, my first test is the light blue/cyan shirt worn by the woman on the left looking at the engine holding enough chroma without the red on the car being over saturated. My second test is whether the lower mid tones hold up, or drop into shadow too quick. I am particularly attuned to this issue due to the way those values are mapped in HDR vs SDR. In most cases HDR will open up those near shadow areas just fine, but the SDR version may get crushy in zone 3, or even 4.
If I had to pick from Phil's examples above, I'd pick B with custom HRO. Unless the car is actually a bit orange, none of the A versions get the red right. IMHO.

Cheers - #19

 

[rand thompson]

Blair,

You're still going to have to bring those REDs in CANDIDATE B back into the REC709 COLOR GAMUT. That's going to make it again look more orange unless you apply a HUE SHIFT to bring it more back to RED however it's still going to have to be less saturated to fit into the RED REC709 Color Gamut.


ORIGINAL IMAGE

Brought back into REC709 COLOR GAMUT without HUE SHIFT towards RED

Brought back into REC709 COLOR GAMUT with HUE SHIFT towards RED

 

[Phil Holland]

I should underline that the cone in that particular sample image is a very, very known quantity in terms of color. Pink isn't exactly what it should be.

 

[rand thompson]

I should underline that the cone in that particular sample image is a very, very known quantity in terms of color. Pink isn't exactly what it should be.

Phil.

But that's my point! There is always a trade-off somewhere. Candidate A gives you the best of both worlds. After correcting for the Out Of Gamut REDS in Candidate B, if you wanted the car to remain more red than orange, pink Cones where the side effect.

 

[rand thompson]

Candidate A was also a liitle out of gamut but a Correction didn't change the final image from the original in any real adverse way as it did with CANDIDATE B.

OUT OF GAMUT RED CORRECTED

 

[rand thompson]

Midtone Reduction to give the impression of being more saturated but without pushing REDS OUT OF GAMUT

ACES 2.0 Candidate A with REDS brought back into Gamut

ACES 2.0 Candidate A with REDS brought back into Gamut with Midtone Reduction

 

[Blair S. Paulsen]

Thanks for doing those examples Rand. Are all the ODT candidates aimed at turning a P3 master into as good a 709 version (without trimming) as possible? Or perhaps for a direct to 709 finish? What about rec 2020?

With the continuing development of display tech like Quantum Dots, WOLED, mini-LED, etc what is the current utility of a standard like rec 709 that attempted to make the best image with the displays widely available at the time? Rec 601 was built for CRTs to suit their particular profile and 709 when LCDs and plasmas ruled.

My point is that we are likely to see displays with substantially larger color gamuts join the party. AFAIK, so far, at least in the consumer space, few displays can even do 98% of rec 2020. For this, and other reasons, most of the deliverables specs I see still ask for rec 709 for video release.

Bottom line: a color gamut that can't handle a pure primary color like saturated red is a fail in 2022.

Cheers - #19

 

[rand thompson]

Thanks for doing those examples Rand. Are all the ODT candidates aimed at turning a P3 master into as good a 709 version (without trimming) as possible? Or perhaps for a direct to 709 finish? What about rec 2020?

With the continuing development of display tech like Quantum Dots, WOLED, mini-LED, etc what is the current utility of a standard like rec 709 that attempted to make the best image with the displays widely available at the time? Rec 601 was built for CRTs to suit their particular profile and 709 when LCDs and plasmas ruled.

My point is that we are likely to see displays with substantially larger color gamuts join the party. AFAIK, so far, at least in the consumer space, few displays can even do 98% of rec 2020. For this, and other reasons, most of the deliverables specs I see still ask for rec 709 for video release.

Bottom line: a color gamut that can't handle a pure primary color like saturated red is a fail in 2022.

Cheers - #19

Blair,

You're welcome. I think they should treat a P3-D65 master, A REC 2100 HLG master, and even a REC2020 master in a way that's relative to their abilities or enabilities to be mapped from whatever camera color spaces into the ACES AP-1 Primaries then through the RRT and ODTs to each of the before mentioned Output transforms . They may need to be more aggressive with some and less aggressive with others.

I think Canon with their new HDR Monitor Claims to be able to reproduce a higher percentage of the Colors of REC2020 than any other Monitor out there, atleast that's their claim. The problem that most people whom are using ACES feels should be addressed is that the AP-1 Primary is so much smaller than almost any Camera Manufacturers's Log Color Spaces. The result is that ACES has to bring in those larger Colors Gamuts back into the smaller ACES AP-1 Primaries Color Space and then mapped to your choice of ODT. The inability to do that in most cases successfully results in these OUT Of Gamut Artifacts which these New ACES 2.0 ODT Candidates are trying to resolve.

So ACES 2.0 is supposed to be right around the Corner so they say so we'll have to see what the final decisions that have been made are and the results of those decisions in hopefully the not so distant future.

 

[Blair S. Paulsen]

Thanks Rand. One would hope ACES workflow would not act as a color constrictor. Fitting into rec 709 is bad enough. Any sense of how much current uptake there is for REC 2100 HLG as the hero master?

Cheers - #19

 

[rand thompson]

Thanks Rand. One would hope ACES workflow would not act as a color constrictor. Fitting into rec 709 is bad enough. Any sense of how much current uptake there is for REC 2100 HLG as the hero master?

Cheers - #19

They have ACES 2.0 Candidates A,B and C for REC2100 so it would seem, judging from the included ACES 2.0 REC2100 Candidates available, that it seems to be of some importance.

 

[rand thompson]

New ACES2 2.0 Candidate From the "ACES Output Transforms Architecture Virtual Working Group"

GitHub Link

GitHub - ampas/ACES_ODT_Candidates


The Older ACES2 Candidate A rev009 REC709 from the group including Candidates B and C









The Newer ACES2 Candidate CAMDRT rev024 REC709

The Newer Candidate looks more saturated but Colors are contained 'IN GAMUT" as well if not slightly better than the Older Candidate A for the "YELLOWS" but not quite as well for the "YELLOW to RED" area. Plus the "Highlights" seem more saturated than I would like.


With LUM vs SAT reduction in around the HIGHLIGHT Area.