I wonder what the exact workflow setting I should follow to color grade 14-bit RAW cinemaDNGs in ACES and linear in After Effects primarily, if not in davinci.
This was my previous workflow:
1- Set AE to 32 bpc and working space to rec 709.
2- Import cinemaDNG in AE via adobe camera raw. Select cinelog-C and fix WB, camera profile, remove nosie and sharpen and select correct lens profile. Nothing else.
3- Apply colorIO plugin and set input to cinelog-C and output to arri logC film matrix.
4- adjust exposure, saturation, vibrance, IR pllution etc.
If I want to work with ACES what should be changed in each step of my workflow?
Ok… There’s some prettu complex things in your workflow that might prevent you from working in ACES.
When you convert to from RAW to CineDNG: what are your choices from gamma and gamut? I’m guessing that your probably staying in linear for gamma. But the gamut would be what? Canon Cinema Gamut?
The idea here would be to convert your footage to something that ACES “knows”. For example: could you convert to ACEScg using opencolorIO ou for that matter ACEScct? CINELOG-C is not an adequate IDT for ACES. Could you convert from cinelog-c to LOG-C or Slog3/cine? Then you could export from AE in Prores, DPX etc… and use that new file to be correctly identified in ACES.
The other option would be to go from linear/camera gamut to linear/ACEScg (via OCIO?). And then export to EXR. But that could make playback hard.
I’m attempting to grade my holiday snaps in ACES in Resolve. 5D MkIII CR2s, some of them Dual ISO. I converted them to DNGs with an app called Switch (which uses Adobe DNG Converter for the relevant camera matrix), as this seemed to help in the past when dealing with Magic Lantern Video. I’m using a custom LMT OFX DCTL as well. I’ll report back when they’re finished.
Hi,
I am bumping this thread up because I have the same question and no solution yet. I received DNGs from a camera that does not have an IDT in Resolve for ACES yet. I found an old video where it is shown that Resolve once had an CinemaDNG IDT:
Is there a new workflow that compensates for the lack of a DNG IDT? Thank you!
CinemaDNG is quite a tricky format to work with due to inconsistency of debayer through different software. We’ve found that the best starting point is to debayer as P3 D60/Linear and then render to OpenEXR. The the files can be imported in Resolve and the correspondent IDT can be applied.
I am sorry but what you are suggesting sounds deeply wrong, or at least not completely specified.
Also, @colorbycontrast asked the right question: what camera was used for shooting? This is where all shoudl be starting from.
Each IDT converts from a camera colorimetry into ACES, not from a file format to another – whereas you are all speaking about converting some file format (DNG) into another (EXR). Also: EXR is another file format: it may or may not encode ACES images, as well as DNG may encode any camera colorimetry.
Without knowing which camera IDT is needed, any suggested workflow may --at best-- work by pure chance only.
Unless the DNG files are already natively in P3, the color pipeline by @Pablo_Garcia seems particularly wrong, since it involves bringing a probably scene-referred camera colorimetry into an output-referred space (P3) and then, somehow, back into ACES (scene-referred). I doubt this would make the workflow any easier, or even facilitate integration with other colorimetries in the same project.
If you select ACES(cc/cct) colour science in the Colour Management panel of the Resolve settings, your DNG should be automatically converted (like any other RAW format, Resolve try to bring it to ACES).
This works to the extend that the colorimetric informations in your DNG file are accurate (see the specifications of DNG for more details).
Thanks @walter.arrighetti for your comment. Your notes, on paper are very much correct but I don’t think you have understood the workflow I proposed (that has been tested for several cameras that use DNG as a format and do not have an IDT).
As Chn3d has specified, the camera they used (what ever it is…) does not have an IDT therefore Resolve will debayer it by default as Lin/709. Have you ever tried to grade that? Not to mention in a timeline that contains cameras that do have IDT’s.
Changing to a format that is ACES compliant and encoded as something that ACES understands in this cases is the best workflow we’ve found to be able to use and grade this files in large ACES projects.
Don’t want to get in to an argument here that is not gonna help any of the users. We have tested it and works. That’s all i can say
I think the misunderstanding here is that @walter.arrighetti is assuming that @Pablo_Garcia is taking about converting the DNG to a display-referred, tone-mapped P3 image, and then applying an inverse Output Transform. If I understand correctly, Pablo’s approach is simply to use P3 primaries to create a scene-referred linear image and save that in an EXR.
But, as @remia notes, if the metadata in the DNG is sufficient and correct to decode the image to P3, it will also be sufficient for Resolve to decode directly to AP0. The choice of P3 primaries is arbitrary, and should be unnecessary.
For some DNGs Resolve may read an exposure normalisation gain factor from the metadata, whereas others may simply be have the range from the sensor black level to clipping mapped into the float 0-1 range. In the latter case it may be necessary to apply linear gain to set a reasonable exposure for the image, or use the exposure control in the raw decode parameters.
I still think It will be nice to know the camera. I mean for example there are some workarounds for the pocket 6k that not involve exporting to EXR etc…
All of that relies on proper debayering and metadata documenting, already a database somewhere available on combination of cameras and RAW converters that give proper DNGs?
I think I can pick up where Ibrahim left off. I have the same question as he/she.
Ibrahim shot 14bit raw video from a Canon 5D Mark III.
Magic Lantern is a firmware for Canon DSLR cameras that enables RAW video. The firmware is ported mainly for the Canon 5D Mark III (3) and Canon 5D Mark II (2.) Among other entry level Canon DSLRs.
Magic Lantern creates a file ending in “.MLV.” To my understanding, an .MLV file is simply a container that holds .DNG frames. (Albeit with a little extra metadata in the header.) .MLV is not usable by most programs, so a standard part of the process is using an app to extract the .DNG’s from the .MLV. Resulting in cDNG sequences.
As it is now, there is no IDT for Magic Lantern enabled Canon cameras. But for those of us using Magic Lantern, we are used to workarounds, as the firmware is open source and user-developed; so development is slow.
I too, would like an ACEScct workflow (or one that is really close) in Davinci Resolve for Magic Lantern RAW video.
As I mentioned above we Magic Lantern users are used to work-arounds. For me, inside Davinci Resolve (Color Management set to ACEScct, and output set to REC709) choosing ANY IDT for cDNG sequences from Magic Lantern, reveals an identical image. It seems the IDT doesn’t change anything for cDNG sequences from Magic Lantern. Is there an explanation to this? Is it because it is RAW data?
I’m interested about that too, may be is it because cDNG defines forward matrices ? So it’s easy to switch from XYZ D50 to ACES.
I’m wondering how light values should be scaled if I convert a RAW footage to EXR, though. Do we need to consider DSLR dynamic range ?