New ACES Studio Introduction from Älmhult Sweden

(Troy Wuelfing) #1

Hi ACES Central Users!

I wanted to make my first post here and introduce myself and the studio I am working at a bit as I expect we will become more active on this forum as we wind our way through the adventure of color managed production work. I hope a few people will be interested enough to read and perhaps have some insights to offer about our questions.
I am the post producer at IKEA communications, we produce the IKEA catalog here in house as well as all global media. This means we don’t make the TV commercials that air in specific markets. Through a quark of the business structure each market is responsible for that on their own mostly. We produce media that all markets can use, which generally means limited dialogue and on screen text as we have about 80 languages we have to support.
Until recently we have relied nearly exclusively on external production companies to support our “film” production, while maintaining an army of still photographers and 3D artists to produce the catalog. We have recently begun the adventure of standardizing our productions for both the media we produce in house and requiring our vendors to follow the same rules.
We have also begun to build out a very robust internal camera department with the addition of an Alexa SXT-W, a few Blackmagic Ursa mini pro 4.6k cameras, as well as some other various camera systems. We are currently growing rapidly and building production infrastructure.
What brings me here is that we have standardized and built our pipeline around ACES color management. As stated before, we are requiring any production companies we hire to also work in ACES. This has just started and is, as expected, a significant challenge. We have found that even the highest end production companies seem to be poorly informed or educated regarding color managed workflows and why they benefit everyone particularly when sharing work product.

Below are some questions we have not been able to solve, or have low confidence we understand well.

  1. How should VFX artist be rendering their work? I have some ideas about this workflow, but I would love some real documentation about how rendered pixels are handled.
  2. What is the preferred color space for texture libraries? Does it even matter if the source textures are ACES? We would love some clarity in general on the process of computer generated imagery and ACES.
  3. I am shopping around the idea of unifying our color management across video and print, as we deliver our print material for the catalog in an RGB space already and ACES seems like an ideal way to allow us to work free of delivery intent across all media formats. Have any other large scale operations used ACES to color manage print and video in the same studio?
  4. Does anyone have experience using express dailies, DaVinci Resolve, and premiere with EDLs to move between premiere for edit and DaVinci for grade?
  5. Has anyone been using any ACES support in the newest version of premiere? I believe it is claimed to support it but I have not had my hands on it yet to check it out.

(Margus Voll) #2


I have some experience with Resolve and ACES and the cameras you name.
Also some of the VFX tests and jobs we have done with it depending what needs to be done. Those other cameras you might have they usually are dependent on input transform a lot so i can not comment on those. Think also Fusion in that context asd send from Resolve to Fusion can be really good tool for some VFX and cgi.

Regarding Premiere keep in mind that Resolve itself is nice for editing also specially with UMP 4,6 k material. Now it would depend if you make and cgi or text there in Premiere or just edit.

Main thing that that comes to mind with print and such is if those apps can handle ACES or do you need to map it with some other transform.

I have some ideas about VFX getting to your pipeline but not anything documented yet as it is in my head still.

As i have to go color session i can not go in to super details now but i gladly will discuss any questions you might have. As i’m not very far from you (Estonia) we can do some Skype sessions as well in some point if that would help or if i ever will be in Sweden work related i could drop by also.

(Jim Houston) #3

I am not aware of this. The ODT for print would need a substantial reduction in contrast to work right (my guess is where the RRT is at about 1.5-1.7 overall to about 1.2-1.3 overall with an adjustment to get blacks at the densest your printer will produce. The RRT is designed around a roughly 2000:1 target for output, but prints are often are 100:1. So a little bit of testing would be needed. If you have a reflection densitometer, you can make a series of greys in ACES about 1/2 stop apart and then print them. If you measure the densities, the desired slope of the curve is in the region of 1.2-1.3, but some artistic judgement is needed to deal with compression of an image onto the print. – with attention needed on the shadow and highlight portions. Good luck with this part.

(Jim Houston) #4

I’ll take a stab at this because no answers have popped up yet. ACES is designed around working in scene linear light and saving that in the OpenEXR file. So the manner of output rendering and the preferred color space for texture libraries both have an answer of linear light. In addition though you have to consider the color space primaries, so put those both together and you have ACEScg which can easily be converted to/from ACES files. [though using ACES directly for textures yields even more colors]

The real answer of course is it depends on what your software supports if you get away from DaVinci, etc…
Sometimes it is necessary to work in a log space (like Cineon…) and there is ACEScc or ACEScct for that. But if you can work in linear that helps a lot. If there is no explicit primary support for ACES but you can work in linear, then conversion from a linear ‘ABC’ primary to ACES is easy.

It does not matter if the source textures are in ACES, but you have to make sure that the rendering software
‘knows’ what colors are expected from the texture map. In other word, you have to make sure that part of the pipeline is color consistent. ACES helps get rid of some of the uncertainty about color spaces in the library of textures, which is more needed today because of the increase of wide color gamut spaces.
As a result, using the wide gamut ACES primaries covers most of the existing color primaries and can extend the life of the texture library.

(Walter Arrighetti) #5

Hi Troy and welcome to ACESCentral.
As far as VFX and the question about “linear” color spaces, I think Jim has exaustively told everything. As any VFX/animation software is confident working in linear-to-light encodings (sometimes also referred to as “photometrically-linear” or “gamma 1.0”), using ACES2065-1 or ACEScg color-spaces should be pretty straightforward for any workflow. As Jim said, bear in mind their color-primaries and be sure that footage is never “watched” directly encoded in an ACES color space without using an Output Transform (RRT+ODT) first, that maps images to some display-referred encoding

As for harmonizing ACES with a printing process, parametrized Output Device Transforms (ODTs) would be created, in principle, as depedning on different, separate factors, like:

  • the print-referred color-space encoding (RGB is still 3-dimensional; CMYK is a 4-dimensional one; using additional inks of specific pigments for, say, metallic or glossy colors, would add dimensionality), which employ a color-subtractive model;
  • the spectral response to a standard reference illuminant (D50?) for every used ink(s) above;
  • the paper composition and reflectance characteristics.

Not only, as Jim said, printed-paper dynamic range is far inferior to that of contemporary HDR displays; you could have very narrow gamuts using standard-ink color-spaces than you would, for example, by employing separate high-saturation ink(s) to the process.
Just think, as an example, to adding an azure-sky, a fluorescent green, and/or matte red-brick cartridge(s) for reaching cerain colors. The resulting ODT would map to a very different (and wider) gamut than that of most consumer CMYK inkjet printers.
The Output Transform made by the RRT followed by one of such ODTs would translate the ACES2065-1 colorimetry in a “print-referred” encoding.

There could be also second-order issues related to representing some of those printed-paper colors in SDR video encodings used by graphic designers in the desktop publishign phase: those Out-of-Gamut colors (OoG) would need be handled in, say, sRGB encoding of their monitors.

I think there is a part of ACESNext devoted to implementing ACES using ICC profiles, which would perhaps be ideal as related to the cross-profiling of monitors and printers. Maybe you can dig more as regards this part of ACES.

Siad that, I think ACES is perfect for harmonizing video, print and production processes, as it very agnostic and based on the physics of light. Besides, once you have profiled print-referred ODTs for your deliverables (including catalogs for web and PDFs), which might have a longer startup time, you would have less-to-none trouble directly-checking and making sure color is consistent across production, print and video.

(Troy Wuelfing) #6

hey walter,
thanks for your reply.
I appreciate the information.
we are not printing to inkjet, we print about 300,000,000 catalogs for IKEA every year, plus many smaller IKEA publications. We also deliver all our print work to press in RGB space today.
I am working to learn a bit more about this process here, but I suspect its already more aligned than it would initially seem. i like the idea of an out of gamut system, better than what Adobe does today which could be a bidirectional warning in some kind of false color to indicate OuG for video or OuG for print.
I don’t think it would be a seamless transition, but for us it would mean one color space, any delivery which would have massive value in being able to use media however we want without worrying.
This print space use for aces is a very long term possibility i see for us.

for now we are working to get good at using aces for our film production, and very much appreciate the community here!

(Jens Lindgren) #7

Hi Troy,

Nice to see you are going full ACES. More studios should do that.
We worked a lot with IKEA to get them started with 3D 10 years ago.
We are a post-production house that are fully working in ACES for the last two years, and the first major show that we used it on was the Swedish Julkalender 2016 called “Selmas Saga”. Since then, all our productions are in ACES.

How should VFX artist be rendering their work? I have some ideas about this workflow, but I would love some real documentation about how rendered pixels are handled.

All processing (rendering, compositing) should take place in the ACEScg colorspace. Brief history of why:
In 2014 a discussion about rendering in ACES started on the 3D-Pro mailing list. This led to people making a lot of tests, comparing rendering in different gamuts and comparing them to spectral renderings. Rec.2020 came out the winner with lowest delta error compared to spectral. The only problem with Rec.2020 is that it doesn’t contain all of P3 gamut for digital cinema. So ACEScg was created based on the primaries of Rec.2020 but was expanded slightly so that it could contain all of Rec.2020 and P3.
One of the people making the tests was Anders Langlands:

What is the preferred color space for texture libraries? Does it even matter if the source textures are ACES? We would love some clarity in general on the process of computer generated imagery and ACES.

It doesn’t really matter as long as they are ACEScg when you hit render. So you could store them as ACEScg (prefered way in my opinion) or store them in any other format but use proper color conversion shaders before they are plugged in to a CG material.

Hope this clears up some things and feel free to contact me if you would like more in depth discussions.

(Troy Wuelfing) #8

perhaps we can setup a phone call and see if there is anything we could collaborate on. it seems from this post you may have some knowledge we would find valuable.
if you want to chat shoot me an email
tack tack!