This is sort of a resolve question, but sort of just a camera format question.
So I’m working on a piece of a documentary that was shot with Arri, a Sony, and apparently 3 different generations of a Canon C300. I don’t have the original footage, just a ProRes4444 stringout of the clips. After A LOT of effort I have managed to get them to send me an XML that links to the OCNs (that I don’t have) so that I can at least figure out what clip is from which camera.
I have never worked with anything from a Canon C300, what format is thing shooting in would you guess? I have one primary in resolve (although honestly it looks like it could just be 709) and then four CanonLog choices? I can guess but I was just wondering if anyone has worked with this and actually knows.
Thanks!
I just graded something that was shot in 3 locations, all on Canon. Two were RAW and one was MXF.
First hurdle is that Flame doesn’t support Canon RAW, so I guess they helped you in that regard already.
And there are significant variations in the LUTs. Like the C300 is much darker than the C500 in it’s log curve. Plus you have the odd Tungsten/Daylight pivot. Though almost everyone these days shoots daylight, even interiors since most lights we use on set are no longer tungsten.
Ideally you would know exactly which flavor Canon they shot with and the camera settings, if they have a camera report. And if they mixed multiple versions, then you would need to know that by clip (though if they used proper naming, the should set themselves apart by the camera index in the filename).
And then it will come down to judgment. If you are pretty certain they exposed correctly, use whichever LUT/IDT looks right and reacts correctly to your grade inputs.
If you need to do any keying, also test with a keyer and see if they react favorably or struggle.
1 Like
This is just grading with no keying needed. I was told they used a C300 mk1, mk2, and mk3. However i can see from the file names that there was also (instead?) a C500 in the mix. There are no camera reports, and the editor insisted that according to Premiere all these files were Rec709 so I’m not holding my breath on any help from there.
Do you know if they all use the same primaries?
Based on this: aces-input-and-colorspaces/canon at main · ampas/aces-input-and-colorspaces · GitHub
CLog2 and CLog3 uses the same primaries and only differs in the gamma curve.
So the ProRes transcodes you got, do they look like log or are they already tone mapped to Rec709?
In my case I got a copy of the OCN and a copy of the Premiere file so I could see what they had done and how the master clip settings / raw module was configured.
Oh they’re log, that was the initial fun part was trying to explain why I needed all this info after being handed a single PR4444 file with (at least) four different log formats all cut together.
1 Like
Oh my. In that case I would also ask for the Premiere project. You can’t assume that the editor selected all the right color spaces in the master clip. They could have mixed and matched.
Was just looking at the files.
The ones that were .crm raw files, I can see the camera details in metadata in Resolve:
Since it’s raw, there is no log curve assigned yet. The editor set all of them to CLog2 in the source settings in Premiere:
Those are correct settings. Now when you look up the difference between CLog2 and CLog3 (Enhance your filmmaking with Canon Log - Canon Europe) it may depend on the shot which is the better choice for the grade.
There were a few mxf files as well, direct from camera. In Media Info I can see that they’re C300MIII as well. But the log curve in the Transfer Characteristics field is just a hex number and a Google search didn’t disclose an decoder ring, we would need a camera to test. I used to know these values for Sony Slog-2 vs. Slog-3.
So on the MXF files there’s no way of know which were CLog2 or CLog3 without a camera report, or have some way for a reasonable guess based on the look.
Ok, this is all super helpful.
Thank you!
As far as I know, Premiere still exports everything tagged as Rec709, even when it isn’t. If you could get your hands on at least one original file from each camera it would help a lot to figure out the correct settings for you to use.
Regarding the original C300, I know it shot Canon Log (the original, not 2 or 3), and the color gammut was basically Rec709. They could’ve shot Wide DR and then it’d be less log and more contrasty, but still with some play to do a slight grade.
Good luck over there!
I can confirm that Premiere tags an export of Log footage tagged as Rec709. Not that I would expect more from Premiere.
There is a Canon XU utility that may be able to inspect MXF metadata in more detail. However, it’s design only allows it to work with camera media, not individual clips on your hard drive 
Asked a DP I know well who is a Canon shooter, to look at his files. Here’s the media info decoder ring.
Which highlights the danger in my use case. For all the raw files the editor picked CLog2, yet the MXF files I got were CLog3.
Both mixed in a single flatpass render. I actually applied the wrong one to a few clips, which explains why they were harder to match.
