ACES – What’s it all about, do I need to worry about it?

You may have heard the term “ACES”  in presentations or workflow discussions for a while now. You may know that it is the Academy Color Encoding System and is a workflow for post producing high end material, but what does it mean in simple terms?

This isn’t a guide on how to use or work with ACES, it’s hopefully an easy to understand explanation of the basics of what it does, why it does it and what advantages it brings.

One of the biggest problems in the world of video and cinema production today is the huge number of different standards in use for acquisition and viewing. There are different color spaces, different gamma curves, different encoding standards, different camera setups and different output requirements. All in all it’s a bit of a confusing mess. ACES aims to mitigate many of these issues at the same time as increasing the image quality beyond that which is normally allowed with existing workflows.

There are 3 different conversion processes within the ACES workflow, plus the actual post production and grading process. These conversions are called IDT, RRT and ODT. It all sounds very confusing but actually when you break it down it’s fairly straight forward, on paper at least!

The first stage is the IDT or Input Device Transformation. This process takes the footage from your camera and converts it to the ACES standard. The IDT must be matched specifically to the camera and codec you are using, you can’t use a Red IDT for a Sony F55 or Arri Alexa. You must use exactly the right IDT.  Using the IDT (a bit like a look up table) you convert your footage to an ACES OpenEXR file. OpenEXR is the file format, like .mov or DPX etc.

Unlike most conventional video cameras the ACES files do not have any gamma or other similar curves to mimic the way our eyesight or film responds to light. ACES is a linear format. The idea is to record and store the light coming from the scene as accurately as technically possible. This is referred to as “Scene Referenced” as you are capturing the light as it comes from the scene, not as you would show it on a monitor to make it look visually pleasing. Most traditional video systems are said to be “Display Referenced” as the are based on what looks nice on a monitor or cinema screen. This normally involves gamma compression which reduces the range of information captured in the highlights. We don’t want this if we are to maximise our grading and post production possibilities so ACES is Scene Referenced and this means a linear response to match the actual physical behaviour of light which is very different to the way we see light or film responds to light. That linear response means lots and lots of data in highlights and as a result large file sizes, there is no limit to the dynamic range ACES can handle. The other thing ACES has is an unrestricted color space. Most traditional systems (including film) have narrow or restricted color space in order to save space for transmission or distribution. If a TV screen can only show a certain range of colours, why capture more than this – this is “Display Referencing”. But ACES is designed to be able to store the full spectrum of the original scene, it is “Scene Referenced”.

In addition  by carefully matching the IDT to the camera, after converting to ACES all your source material should look the same, even if it was shot by different cameras. Now there will be differences due to differing dynamic ranges, colour accuracy, and noise etc, but the ACES material should be as close as technically possible to the original physical scene so a grade applied to one camera make or model should also work for a different camera model in exactly the same way.

Now this big color space may well currently be impossible to capture and display, but by deliberately not restricting the color space ACES has the ability to grow and output files using any existing color space.

So… using the IDT we have now converted our footage to ACES linear saving it as an OpenEXR file. Or as in the case of some grading packages like resolve we have told it to convert our material into ACES as part of the grading process. But how do we view it? ACES Linear looks all wrong on conventional monitors, we now need a way to convert back from ACES to conventional video so we can see what our finished production will look like. Well there are two stages to this. The first is called the RRT, the second the ODT, sometimes these are combined into a single process.

The RRT or Reference Rendering Transform is designed to convert the  ACES linear to an ultra high quality but slightly less complicated standardised intermediate reference format. From this standardised format you can then apply the ODT or Output Device Transformation to convert from that common RRT intermediate to whatever output standard you need. In practice no-one sees or works with the RRT, it is just there as a fixed starting point for the ODT and in most cases the RRT and ODT operations are combined into a single process a bit like adding a viewing LUT. The RRT transformation is incredibly complex while the ODT is a much simpler process. By doing the difficult maths with one single RRT and then keeping the ODT’s simpler it’s easier to create a large range of ODT’s for specific applications. So from one grading pass you can produce masters for broadcast TV, the web or Cinema DCP just by changing the ODT part of the calculations used for the final output.

If your using a conventional HD monitor then you will need to use an ODT for Rec-709 so that the ACES material get converted to Rec-709 for accurate monitoring. It should be noted now though that as you are monitoring in the restricted Rec-709 color space and gamma range that you are not seeing the full range of the ACES footage or RRT intermediate.

So, it all sounds very complicated. In practice what you have to do is convert your footage using the right IDT to an ACES OpenEXR file (or tell the grading application to convert to ACES on the fly). You set up your grading workspace to use ACES and then set your output RRT/ODT to output using the standard you are viewing with (typically Rec-709) and do your grading as you would normally. One limitation of ACES is that due to the large color space many conventional LookUp Tables won’t work as expected within the ACES environment. They are simply too small. You need at least a 64x64x64 LUT which is massive. At the end of the grade you then choose the ODT for your master render, this might be 709 for TV or sRGB for the web and render you master. If your taking your project for finishing elsewhere then you can output your files without the RRT/ODT as ACES OpenEXR.

The advantages of ACES are: Standardised workflow with standardised files, so any ACES OpenEXR file from any camera will look and behave just like an ACES OpenEXR file from any other camera (or at least as closely as technically possible).

Unlimited dynamic range and color space, so no matter what your final output you are getting the very best possible image. Of course limited by the capture capabilities of the camera or film stock, but the workflow and recording format itself is not a limiting factor.

Fast output to multiple standards by doing the difficult maths using a common high quality RRT (Reference Render transform) followed by a simpler ODT specific to the format required. Very often these two functions are combined into a single ODT process.

So is ACES for you? Maybe it is, maybe not. If you use a lot of LUT’s in your grade then perhaps ACES is not going to work for you. If your camera already shoots linear raw then your already a long way towards ACES anyway so you may not see any benefit from the extra stages. However if your shooting with different cameras and there are IDT’s available for all the cameras your using then ACES should help make everything consistent and easier to manage. ACES OpenEXR files will be large compared to conventional video, so that needs to be taken into account.

2 thoughts on “ACES – What’s it all about, do I need to worry about it?”

  1. Hi Alister,

    Is there an ACES IDT for the Sony F3 (in SLOG)? Or is it best to grade SLOGged F3 footage in the regular Resolve color modes?

    Cheers

    1. No, I have not seen an IDT for the F3 and Slog.

      ACES is really best for cameras that capture linear raw as this way you stay linear in camera and then linear in ACES. Your not getting as big a benefit if you start with Log or conventional gamma as much of the extra data that ACES is trying to retain has already been discarded.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.