OCIO Color settings for FusionConnect clips in ACES CC?

Hello everybody,

I was eager to try out the Fusion Connect feature in DaVinci Resolve, but I have been struggling with the color/gamma that came back from Fusion.

macOS 10.12.5
Fusion 8.2.1.6
Davinci 14 beta 5
Color science: ACES CC
IDT: Slog2 (individually applied)
ODT: Rec709

1) I edit my footage in Davinci and grade it.
2) Then I disable the ODT and the grades in Davinci.
3) On the edit page I chose "New Fusion connect clip", I select "EXR no compression" and "Open Fusion Connect Clip".
2) I do my magic in Fusion and render it out (float32).
3) Back in Davinci the colors and gamma of the Fusion Connect clip are just wrong.


Now here's what I found out so far:

4) I looked for the rendered Fusion file and put it into my media pool.
5) I select "ACES CC" as IDT, and voilà the colors and gamma are 100% correct.


It looks like DaVinci renders the FusionConnect clip out in ACES CC to start with, am I correct?

So how can I translate the ACES CC footage in Fusion to the correct color that DaVinci expects when it comes back? I already tried it with an OCIO color space node (including the newest config.ocio 1.0.3 but as soon as I choose anything but RAW as output-space, my OCIO and saver nodes turn red, and Fusion doesn't render out anything.

Use an OCIOColorspace node before Saver to transform the image from your working space to ACES2065-1, then write out as raw data (keep both gamut and curve). Or change the Saver output color space and curve to ACES2065-1, although it seems that OCIOColorspace node is more flexible, plus it makes it explicit, what transforms are going on.

Resolve for some reason does not let you select ACEScc as IDT although it happily writes it out, and in the end expects intput exr-s to be ACES2065-1.

Dear Hendrik,

thanks for your quick reply. I tried what you said, but it didn't change anything at all. In the meantime I tried it with DaVinci YRGB Colored managed with timeline color space set to linear. this seems to work. But I want to get it working with ACES CC.

Any ideas?

What is your working space in Fusion? If your input exrs are ACEScc, what transform is applied to them when you load them in? You must keep track of your working space and do transform from there to ACES2065-1. I think the combination of raw and ACES2065-1 gives a null operation because for OCIO the source is unknown. Change the Source space to ACEScc, ACEScg or whatever your working space is.

THat's what's really bugging me... I have been trying to convert my FusionConnect file which is in ACES CC, I guess, with the OCIO node to ACES2065-1. But it gives me red nodes...

Maybe your OCIO configuration is incomplete if it gives error? You can try and download the full configs from here:
https://github.com/imageworks/OpenColorIO-Configs

For me it works as I described.

Thats where I downloaded the config.ocio ...

I also build the packages on Mac OSX as described using Homebrew. Do I have to install the config.ocio somewhere? I just opened it in the node.

Sorry, I was very stupid. Thanks for your help, I didn't figure that I had to download the entire folder, not just the config.ocio ...

Still I think all of this would be a lot easier if Resolve would let you choose the IDT for a FusionConnect clip. And also choose "Generate Optimized Media" as the EXR-file is quite hungry.

Use an OCIOColorspace node before Saver to transform the image from your working space to ACES2065-1, then write out as raw data (keep both gamut and curve). Or change the Saver output color space and curve to ACES2065-1, although it seems that OCIOColorspace node is more flexible, plus it makes it explicit, what transforms are going on.

Resolve for some reason does not let you select ACEScc as IDT although it happily writes it out, and in the end expects intput exr-s to be ACES2065-1.

Just one more question: Why you would suggest to place the OCIO-node right before the saver node? I thought ACES CC is not linear: As Fusion works linear wouldn't it be better to place it right behind the loader node, so that all of my operations happen in ACES2065-1?

Because I suggested the node for explicit conversion from your working space to ACES2065-1. I don't know what your working space is, for ACES project it should usually be ACEScg, but it is up to you to set this up correctly. ACES2965-1 covers imaginary colors and can give you funky results if you forget that, so I wouldn't advise using that as working space.

In loader node or right after that you should move your image from ACEScc to working space, use separate node for this. All the comp work will happen between these two OCIOColorspace nodes. I think for ACES projects it can be beneficial to use explicit nodes, so that it is very clear that gamut transforms are done in addition to transfer function changes.

Btw, Fusion itself doesn't give a flying duck about whether the image is believed to be linear or not, it is up to the comper to make sure stuff gets transformed correctly and ops that need linear values actually get linear values.