Advanced 3x3 matrix camera matching in Resolve

Github link for python script: This has been updated to intake the source & target images, and output a working DCTL file all in a single step. Thanks to my friend Riley for the help! =========== Explanation: On high end productions typically the main camera (ie. Venice, RED, ARRI, etc.) will be chosen as the primary “observer“ and all other second unit, b-roll, VFX & extra assets will be matched to that camera, before going through the rest of the compositing, grading & image formation (DRT) pipeline. Often that camera’s working space (ie. , Arri Wide Gamut 3, etc.) is used during production. It is increasingly common to convert everything to a standard mezzanine working space (ie. ACEScct) for easier interchange, however the “observer“ still needs to be matched especially under unique lighting conditions. There’s no such thing as “accurate color“ in a sense, since not even individual humans will perceive the same colors, much less a camera’s sensor. We’re not capturing the entire spectrum, but rather a limited set of channels (such as RGB) so there will always be some implicit bias in that perception, especially as lighting conditions vary. Thus the goal is to attempt to match the bias of one observer to another. It’s common practice for folks working in Nuke, Flame, Maya, Unreal, etc. to do this type of matching with Macbeth/Colorcheckers all over the place, but somehow no-one has ever made a tutorial on observer matching in Resolve. So here you go. =========== Chapters: 0:00 What is it 2:51 How Hollywood does it 4:07 Analysis in Resolve 10:03 Resulting Workflow 14:00 Regarding narrowband sources =========== Further reading links: Forum post: Colour Science Python library: Colour Checker Detection library: DCTL code for 3x3 matrix template: Transforms/ Steve Yedlin’s Display Prep Demo: Great free / open source alternative DRT called 2499: BTS article on Avatar 2 color workflow: Article on the risks & limitations of using Color Checkers for pure calibration: Chris Brejon’s excellent writeup on display transforms: Great explainer from FilmLight about an important trait of using 3x3 matrices, which is that they preserve linearity & additive mixtures: