Color Management in VFX — ACES

There are already many many articles talking about ACES. Well, I’m not sure should I do the same thing on the same topic. But if I want to introduce other concepts I probably still need to do it anyway.

Most of the definitions, explanations and introductions of ACES I will just quote from ACES documents, other professional pages and experts, and rewrite a bit in a easier way to understand.

WHAT IS ACES

ACES is stands for the “The Academy Color Encoding System”, is a solution for managing color and digital files throughout the life cycle of a motion picture or television production from acquisition thru archive.

Making movies is a lengthy, complicated processes and needs many productions and studios involved, well, movie industry has developed for more than 100 years and it definitively figured out ways to move images between productions. Before ACES, there is no standard for high fidelity covering the pipeline because it requires to consider every detail in each stage which will cost a lot to push all the productions, studios and artists to move their traditional workflow to a new one.

The ACES team, which is under the auspices of the Academy of Motion Picture Arts and Sciences, had released ACES 1.0 on a presentation at Siggraph 2015 and now it becomes the global standard to provide consistent color so everyone in the pipeline from on-set crews to post-production artists is able to see the same color.

TERMINOLOGY

There are many technical terms that only relate to ACES. Before we go into this, here is the terminology you might need to know.

Personally speaking, I would say there are lots of technical terms and acronyms that only use in ACES and too abstruse to many artists to get their first step in the world of ACES.

ACES Input Transform: Known as Input Device Transform, or IDT, is designed to be unique and individual to each camera. Many input transforms are actually developed by the manufacturers that are able to move their unique camera data to ACES.

Look Modification Transform, LMT: LMT is a ACES to ACES transformation that systematically change the appearance of ACES-encoded data viewed through an ACES Output Transform.

Reference Rendering Transform, RRT: Coverts the scene-referred colorimetry to a large gamut and high dynamic range space, resembling traditional film image rendering with S-shaped curve. RRT usually combines with ODT.

ACES Output Transform: Known as Output Device Transform, or ODT. There are many output transforms which are written specifically for each potential target display.

ACES2065–1: It is the principal scene-referred color space in the ACES framework for storing images linearly with the primaries AP0 that confines a wide gamut. This is also the main color space for exchange, I prefer to say it is the host of the ACES color space family.

• ACEScg: A color space definition that is smaller than ACES2065–1 color space and linearly encoded with the primaries AP1 for improved use within computer graphic rendering.
• ACEScc: A color space definition that shares the same primaries AP1 with ACEScg and logarithmic transfer characteristics for improved use within grading.
• ACEScct: A color space that shares the same features with ACEScc except resembling a toe behavior of Cineon files.
• ACESproxy: A color space definition that shares the same primaries AP1 like ACEScg, ACEScc and ACEScct with logarithmic encoded similar to ACEScc. The encoding is exclusively designed for the digital devices that don’t support floating point arithmetics encodings. Mostly used on set.

THE BASIC

This is the basic ACES. The original scene will be captured by different cameras and encoded in unique color spaces designed by the manufacturers, and here is how ACES begins to play the game.

Like the definition of IDT above, each image that is recorded with its own logarithm also has specific IDT to transform into a large, uniform color space and share the same primaries — ACES2065–1. The advantage of this workflow is from the step after optical photons hit the camera chips and transferred into electronic signals will be in ACES which represents all visible colors to the human eye that keeps the maximum fidelity throughout the pipeline, to the final stage the electronic signals to human eye that needs to convert from ACES to the multiple color spaces to accentuate the difference between displays.

The benefits of ACES are:

• Standardizes color management
• Maintains color fidelity at the highest level with wide gamut and high dynamic range
• Simplifies camera matching in DI
• Improves color and workflow communications
• “Future Proofs” the source master

ACES in DETAIL

You already have a big picture of whole ACES workflow, now breakdown part by part and understand ACES works in different stages. The whole pipeline can be simply distinguished into four parts: On-Set, Editorial, Visual Effects and DI grading.

Well, I know this series is more about VFX, I’ll just focus on the role that ACES plays.

On-Set

The goal is to be able to convert images to ACESproxy for digital cameras and represent the the Output Transform to the on-set monitor, or live grade and pass the grading data including ASC CDL and LMT downstream to editorial, VFX and DI within the ACES framework to see the consistent color.

For the negative films, ACES also has explicit definition to encode with Academy Density Exchange(ADX) format either from the Academy Printing Density(APD) scanner or regular scanners with the Input Transforms that is able to transform into ACES.

Editorial

In the ACES-based workflow, the images that reach the edit room from various sources should look identical to what was viewed on-set and in VFX, and what will be seen at the final color grading stage.

Without much color adjustments or changes involved, while they shouldn’t. At this point, editors may not even be aware of the ACES process that was used to create the proxies that they receive. What they will notice is how easily the images cut together with a consistent look.

Visual Effects

When moving into VFX part, it is a bit more strict in the process because it requires some pre-preparations before CG artists begin their works.

Cameras may have slightly different settings during shooting and therefore the multiple shots need to be balanced with “neutral grade” in the same scene. That will make huge benefits for CG artists and compositors while doing any effects than manually match scene for each shot which has its own color. The inverse neutral grade must be applied after the visual effects were added then deliver to DI.

So where is ACES? It actually already starts from the beginning. Studios receive images with camera native log encoded, and transform them with input transforms to ACES-based images which is ACES2065–1 color space with EXR container for storage and interchange.

In CG world, ACES2065–1 is not for working because it is far from optimal for computer graphics rendering. ACEScg is a color space which uses AP1 primaries closer to the actual devices that has better performance than working in ACES2065–1 directly. For delivering, it has to be converted back to ACES2065–1.

DI grading

In this stage, colorists usually prefer to grade in log space to model the perceptual phenomena, therefore VFX departments need to deliver EXR with ACES2065–1 encoded to DI and colorists will convert ACES2065–1 to ACEScc or ACEScct color spaces to continue color grading.

The difference ACEScc and ACEScct is the darkest value, ACEScc is a pure log function, but ACEScct has a “toe” near black, to simulate the minimum density of photographic negative film, and the legacy DPX or Cineon log curve.