Here is an overview of how ACES works in DaVinci Resolve (as I see it for now). Other platforms like Nucoda, Baselight, and Scratch all handle ACES in their own way, but the main approach is the same. This is a simplified approach and I think it will help you understand it better:
ACES generally is a linear color space with almost infinite colors and around 30 stops of latitude. Most modern cameras like Arri, Red and BlackMagic cameras utilize a wide gamut sensor, (i.e. producing more color information compared to lower 10 and 8-bit cameras), and ACES makes use of the wide gamut color data. Think of it as slide scanned at 2400 dpi versus regular images that are scanned at 100 dpi. The actual math is more precise, but you get the idea.
There is 3 main parts to ACES while grading:
- First is the input transform (IDT). The input transform converts camera color data to the ACES color space and because every camera manufacturer have crafted their own color secret sauce, ACES allows you to bring various camera images into a uniform working space.
- Secondly, ACEScc is the commonly used color space for grading, which is like a conversion of linear space to log (ACES2065 linear > ACEScc log), making it easier for a colorist to grade using log tools.
- Third is the Output Display Transform (ODT) which makes your ACES or ACEScc display properly on all devices. It is important to select an ODT based on your device capability, for example, REC709, P3, and 2020. Ideally, they all should look the same if calibrated to their native standard. When new devices will come to market with better color gamut they can be used on the same principle and not to worry about redoing all the work you have done in grading already.
The first step is to set your Resolve Color Management to ACEScc, then set your output display transform (ODT) to rec 709. I usually leave the input transform blank for my project settings and set the transform for individual clips on the color page or in the media pool clip parameters. Why? Often, a project contains original media from multiple different cameras like Arri, Red, and BlackMagic for example, and if you’ve been using Resolve 15 you may have noticed that software automatically transforms raw camera files with the correct IDT. I think big workflow pipelines would benefit from this a lot as every minute you shave off the process is gain that can be used elsewhere. However, if there are files in a project like ProRes or DNxHD that aren’t RAW from a camera, you can manually select the correct transform.
I grade with log tools as ACEScc behaves like log. Adjust offset for WB, adjust contrast – pivot and master offset (exposure in a way), Hue and also saturation or color boost if needed. Most of the time i even do not use rings or balls of my Elements panel to adjust and if i do need then midtone ball is the one that mostly use. Less and less i need to adjust highlight or shadow separately. So in a way i start with few nodes for normalizing master offset (exposure), contrast and WB. If it feels off i use hue to line up shots to mach more evenly. Later on i start to build look on top of it like usually either with curves or regular log tools.
I have heard keying seems to be issue for some users with Resolve in ACES. So what can you do to get around that? I find a good option is to do it after “normalizing” the image and work with HSL key. Also consider using the blur option in the key tool as needed.
I find that it’s really helpful to use scopes with the HML tool. It is the most valuable thing you can have in your studio that helps you get a clean starting point fast with tight, shot to shot consistency.
Recently I read a comment on a Facebook Group posted by Walter Volpatto about efficient workflows that he uses, and it turns out my approach is very similar.
- Start with the input transform (which in this example is done already before nodes either automatically in Resolve 15 or by manual selection)
- Normalize your shot
- Create the look on key shot as shared node or post group to use over scene
- Add grain, or FX like film damage also as shared node or even on timeline level
- Add finishing touches like vignetting and masks
What Walter described is very clever and efficient approach in my opinion. You set the look your director wants for a scene and then normalise everything as close as you can with the look already intact. This works the best with well-captured material that is equal in exposure and lighting and really speeds things up. There are times when I have to do that shot by shot in more detail, when they don’t match, but that’s more a production issue than ACES or grading.
I wanted to point out that for VFX, it’s always a good idea to use linear, wide gamut, 30 stop material with no gamma at its core. This simplifies the entire process and your team will be happy they don’t have to manage a gazillion luts just to display the images correctly. In this way, ACES is helpful for post houses of every size—big, medium and small.
To conclude my rant, here’s why ACES is my go to color system
- Fast, clean results
- Easy to manage in a VFX pipeline
- Easy to monitor
- Make the most out of your grading with wide-gamut media
- About 30 stops of latitude and no gamma issues
- Behaves more like film and less like video
I hope that helps you out and shows how you can approach your workflow using ACES.
If you have any questions please send them here in comments and I will do my best to help.