@dubzeebass - The short answer is yes. I'll be in touch shortly.
It's been a big part of our work over the past couple of months but we are still a little way off from being able to give you exactly what you need - I will need a couple more things from you for this (I'll email you).
Actually, your C100 request and our subsequent work has exposed some very useful things about Canon DSLR and C-series internal color science that we suspected but didn't fully understand and this will eventually benefit ML - we may even be able to produce a true Canon log picture style at some point, not some psuedo, log-like facsimile.
I'm going to get technical again (sorry) and start talking about ACES and matching to other C-series cameras because it may help you and others to understand what we have to do.
The shots you provided were unfortunately over/under exposed for the exacting requirements of shooting Canon Log. Correct exposure is critical to preserve dynamic range and white balancing must be performed to a proper reference as you are baking this into the video file and don't have the luxury we have with raw images.
FYI - There is a very specific procedure for obtaining accurate exposure when shooting Canon log on any C-series camera, as explained here by Jem Schofield -
http://www.learn.usa.canon.com/galleries/galleries/tutorials/eos_c300_tutorials_gallery.shtml. If you follow this you will get perfect exposure every time, if you don't you will always have issues.
Unfortunately, we don't have a C100 MkII to obtain a controlled lab reference (I'm still trying to source a rental here in Latvia), I had to base all the calculations on the color and white balance coefficients already available for the Mk1 (as per the official Canon ACES IDT) and some correctly exposed and balanced C100 and C100 MkII original stock footage to fully understand the MKII's colorspace.
Fortunately, the difference in color appearance between the MkI (and C300 MkI) and MkII can be neutralized by scaling RGB channels (basically changing white balance offsets) to achieve visually identical color with low DeltaE - we can assume the remaining difference is mainly down to noise (improved sensor in the MkII) but you will see pretty much identical color - this solved the first part of the problem and allowed us to build a MkII ACES IDT - basically we did Canon's job for them.
The next part is where things get really complex - matching color and levels to ML raw footage.
At present, MLV DNG files (from Raw2CDNG and MLVFS) have their color assigned by forward matrices which are linear and target an Adobe reference (I want to change this to a common, easily obtainable reference such as a Colorchecker) - the color in the C series cameras have (as far as we can tell) a non-linear color profile applied somewhere in the signal path before the IRE scaled lin-to-log transform happens (the IRE scaling is what makes Canon log and Cineon technically incompatible BTW so we had to factor this into our work).
This color profile stage essentially means an unknown amount of color values are mapped individually to remove/lessen any out-of-gamut issues or for aesthetic reasons (possibly to make it look more film like/cinematic) - this helps with things like spikes in the spectral emissions of certain light sources (i.e. LEDs) so already, color matching these non-linear to linear assigned sets of RGB values is not a simple procedure - it requires a lut somewhere in the pipeline.
Without the keys to Canon's lab, it would be impossible to exactly invert the color mapping that happens in the C100 MkII to obtain a true scene-referred linear signal (as we can do with raw footage) so we need to match MLV footage to the MkII (or vice versa) with some 'educated' guess work based on what we know about Canon color science and available data (i.e. the C100 MkI ACES IDT).
So, using our MkII ACES IDT we can be reasonably confident that we can transform C-Log MkII footage to scene-referred linear RGB values and because the DNG files can easily be debayered to scene referred values we can then color match them. The significant color matching part should ideally be handled by a trivial 3x3 matrix (as this will not introduce artifacts or errors) but to do any of this we need an very accurate chart reference shot from the C100 that is as close to perfect white balance and exposure as possible (and preferably shot with a very good lens) to enable us to calculate the required RGB offsets. Any remaining inconsistencies in reference color values can then be handled by a 3D color correction lut containing float values before a final transform to an output referred or log colorspace i.e. a signal that can be rendered or graded.
That covers a 'theoretical' color matching procedure for 2 cameras but we have to remember that even though we can obtain scene-referred values from the MLV DNG files it will not necessarily be reference-accurate or look nice when viewed in a display colorspace (i.e. sRGB/Rec709 etc if we only use the XYZ matrix and a 1D transfer function). The DNG files should also be color matched to a known reference. Ideally this would be handled by the the forward matrix tags but no sensor is perfect and a forward matrix transform can never fully match reference levels - infact it may make things worse plus forward matrices do not form part of the CinemaDNG spec so you can't guarantee all apps will render the same values.
Just as the in-camera profile of the C-Series cameras 'tweak' color, we need to apply the same logic to raw files if we are to match them and this means matching both cameras to the same reference values (i.e. a color chart) and not match one camera to another. In practice this approach brings actual color matching of 2 different cameras 'perceptually' close to all but the untrained eye. The downside is that you will lose some of the camera's color aesthetic so it is always a trade-off.
