In my opinion, daVinci Resolve is a very powerful piece of software for color correction, primarily intended for editing and motion graphics, but it works just as well for stills. [It exists in two versions, daVinci Resolve, which is free to use, and the Studio version, for which there is a license fee.]
I colour matched a shot of an Xrite Colourchecker in daVinci, as well as in darktable
— and the result puzzles me.
According to daVinci, everything below 10% difference is fine,
while darktable marks “bad” matches with an X, and semi-bad
matches with a diagonal.
What puzzles me is that it is not the same patches that are considered
“less good” in daV and dt. dt says my moss-green (patch 4) is bad, while daV gave it
a 5% verdict.
Patch 20 (light grey) is fine according to dt, while daV says it is as bad as
the moss-green. Et cetera; see other diagonalled patches.
I wonder about this difference. Might it be that daV is more skilled in
rectifying the color patches?
this is a bit of a silly sport. but if you really care about exactly these patches you can use radial basis functions instead of a simple matrix fit, such as the colourlut module in dt did in the olden days, or such as vkdt uses in the colour module too.
consider this input image (cropped and rotated to match the angle brackets of the checker target):
then use the cc24.pst preset to initialise a pick module with the reference values and the coordinates of the spots of the colour checker 24, then connect it like so to the output of the colour module:
applying the colourcc.pst preset, importing the 01 picker in colour and switching to parametric mode will then make all the 24 patches match. the differences are indicated in the gui (but are merely an indication of how much “work” the mapping does, the output will be the reference value for each patch):
Using eyes instead of numbers and the delta E criteria used by DT I think the ones in the DR image that you shared that are visually different are generally marked about the same in DT so ignoring DR claims about some percentage and using a visual scale of can’t tell difference, slight difference and clearly different as the possible outcomes… they don’t line up that badly???
Is DaVinci using BT.709 and RT sRGB for the R’G’B’ to RGB conversion? That would give you slightly different values in the linear domain. There may also be an extra conversion to Y’CbCr in the video software.
