Color Grading In Spectachrome Part 2 – Working With The Fuji XT-3

Happy to see part 2 for this series. Also, thrilled to see that the XT-3 isn’t too bad at compression artifacts. Forgot if you mentioned this in Part one but was this captured on an external recorder (like a Ninja V) or internally in camera using H264h265.

Do you find that cineon ever restricts your dynamic range? I love that workflow but I’ve heard that cineon is obsolete due to it being limited to ~12 stops and we’ve gone beyond that now with RED & Alexa cameras. I know the Fuji is still within the DR of cineon but curious about your thoughts for the high-end cameras as well. Look forward to Part 3!

I also love using ACES in this same way! I prefer to convert my LogC into ACEScct with an ACES transform node (within a DaVinci YRGB timeline), then do all my grading after that ACEScct node (as opposed to using an ACEScct project), and then I use another ACES transform node to go from ACEScct to Rec709. However, I’ve found that this method isn’t 100% consistent with a true ACEScct project for some reason. Any idea why? Like, if I grab a still of all the nodes that are in between the two ACES Transform nodes (the IDT and the ODT), and apply those same nodes on the same clip within a true ACEScct project, then the grade ends up quite a bit different, even if I have the same Rec709 ODT. The math is a bit different for some reason, which is odd to me. Because shouldn’t the two grades be the same if all the operations are being done in ACEScct?

Interesting! On my first tests I only made sure that my transforms were giving an identical result as ACEScct project. I’ll test out the nodes and let you know if I also have the same issue. Which version of resolve are you running?

Quick Question about this method as you showed it on the video, because I was tinkering with it when conforming a C200 source short film. In this instance of your video, why do you ignore Input Color Space and Output Color Space of your CST node and focus only on gamma? In my case, changing input and output color spaces to match source (Canon cinema gamut to Alexa) brought the saturation to a really good starting point. In addition, you turn the Source Gamma into Log C, but don’t change the ACES input transform of the ACES transform Node to anything (I think this might be because ACEScct is so similar to LogC?). It is strange to me that the transform is ACEScct to ACEScct. Obviously with Canon, I could probably do everything with just the ACES Transform node as well but was experimenting with the CST node before the ACES node as set out here in case I come across certain codecs or need to export things for VFX.

I chickened out because I had a client coming early next day so defaulted to use the standard RCM methods. Even though I really liked the starting point of the resulting color transforms with this method more than RCM, I didn’t want my naivete of the way these color sciences interact to bind me with a client in the room (plus I am a lot more used to how my controls on the surface react working in 709 timelines than Log).

Yeah that’s weird, I’d love to know why!