New ACES Working Group - IDT Implementation

ACES Community,

Today we are announcing the formation of a new implementation working group to focus on IDT development within the ACES system. This group’s focus is to sort out some of the issues related to implementation of IDTs. This includes investigations of consistent 18% gray placement, suggested defaults for the IDT generation procedure, and IDT development tools. The group will be Chaired by Josh Pines of Technicolor.

We will be announcing the exact schedule for this working group very soon but in the meantime, we’d love to hear your thoughts and ideas on IDTs within the ACES system and the Working Group Proposal posted here.

This new group, as with all the recent groups will be conducted virtually using GoToMeeting for meetings, ACESCentral for discussions between meetings and the ACES Workspaces for exchanging documents, code, etc.

Thank you
ACES Leadership

Working group proposal : ACES_Working_Group_Proposal_IDT_Approved.pdf (53.3 KB)

IDT working group home : Working Group Home

7 Likes

Great to see that new Working Group!

1 Like

Count me in :slight_smile:

1 Like

Please try to get Sigma on board the IDT train with their fp camera! It shoots cDNG raw, but I don’t know if they’ve profiled the camera spectral sensitivities for a proper IDT yet.

The current process with using the raw files is to either convert it into something else so it looks good, or start by pushing saturation way up. It would be great to just load an fp raw file and have it look as good as a RED or Sony raw.

1 Like

Is the raw cinema DNG really an issue, talking about the Sigma FP, that is? The cDNG metadata contains the color matrix so shouldn’t it be possible to go directly to ACES without IDT? Would be nice to get a definitive answer to this. I know it’s been discussed a little bit before but I’m still not sure about it.

I’ve been under the impression that the ColorMatrix1 and ColorMatrix2 tags (for D55 and D65) in the DNG would be enough, and I think Resolve produces correct result, at least seem to (as long as the “Apply Pre Tone Curve” is unchecked in the raw settings).

Does someone know what metadata is actually needed in DNGs to be able to correctly transform to ACES?

Yeah, those matrices map from CIE XYZ (D50) to the camera “reference” colourspace, and I do use them for that purposes. I think that rawtoaces can also, I never used it with them. Note that they are never given for two close CIE D Series Illuminants as it would defeat their purpose, so what you will find is usually Tungsten and D65.

Cheers,

Thomas

Thanks. Yes the two Illuminants appear to be Standard Light A and D65 in every DNG from Sigma FP I’ve checked. Quickly testing also with RawTherapee vs Resolve (through ACES) I see no visual differences how the color or saturation appears. With Lightroom there is a clear difference, though. Lightroom shows images brighter, more contrasty and staturated. It clearly applies some tone curve, perhaps from the DNG metadata (huesatmap maybe).

Lightroom applies two more things at least:

  • ACR Tone Curve
  • Hue Twists LUT
1 Like

Okay. These would be sensible to apply after transforming to ACES, right? In other words, tone curves or saturation changes aren’t necessary to correctly transform DNGs to ACES (without camera specific IDT). Assuming of course the metadata in DNG is correct for that camera.

Not really for the former unless you really like the ACR rendering in which case you would apply the tone curve, available here: https://docs.google.com/spreadsheets/d/1JV2BzrfnpXgKsqEsw6cS2bGc8I2u8isKabt3nGLckWs/edit?usp=sharing, as an LMT followed the inverse RRT + ODT or OutputTransform you are using.

And for the latter, the hue twists are meant to bring the correction from the 3x3 matrix closer to the values of a reference ColorChecker, the problem is that they are not exposure invariant so you don’t really want to use them for large scale production, they are fine for personal photo-processing, but nothing that requires accurate reproducibility and invertibility.

I systematically remove the ACR Tone Curve and the Hue Twists table whenever I use Adobe stuff at work.

Cheers,

Thomas

Right, it’s kind of a “look” and just because that look isn’t applied when the DNG is transformed to ACES doesn’t mean the transform isn’t correct. If someone likes that look they can apply that later. I just wanted to confirm that strictly speaking IDT isn’t required with DNGs. Thanks Thomas for answering my questions.

It is indeed not a requirement, if the vendor has done a good job characterising the camera, i.e. the ColorMatrix1 and ColorMatrix2 matrices and the software using the DNG files is able to convert to ACES2065-1, then you are good!

You could think as those matrices as being pseudo-IDTs, the main difference is that they convert to CIE XYZ (D50) instead of ACES2065-1, so there is a need for an extra internal step by the software to bring them in an typical ACES workflow.

1 Like