kk650 said:
I’ve always found what your colour matching tool can do vey impressive but unless it can match the colours and image dynamics 100%, there is a margin of error of the algorithm making up colours/changing image dynamics that I personally would not be able to tolerate. If I were to create a release using that tool, I would never release it unless it was 100% accurate or the difference was so negligable that I could not tell the difference looking with my own eyes on my 30 inch monitor.

This really makes no sense to me. The changes to the image dynamics by just applying a curves adjustment are orders of magnitude larger than the margin of error of an algorithm. The color grading of an official release is not done with curves adjustment, as such the underlying colors cannot be retrieved by curves adjustment. Applying simple curves adjustment will result in small margins of errors in some areas of the frame, but very large errors in other parts of the frame (like for example good looking skin tones with undesired color casts in the background). The entire basis for using an algorithm is to solve the limitations of curves adjustment, which is essentially curves adjustment in dozens to hundreds of color spaces.

To clarify, the simplest case of the algorithm in just one color space (the RGB color space) is just simple curves adjustment as you applied it to the bluray, and thus has the largest margin of error. The more color spaces you use, the smaller the margin of error.

A direct consequence is, that the algorithm can match your regrade with essentially zero margins of error.

kk650:

Bluray matched to kk650:

A non-zero margin of error in a color matching with this methodology result implies, it cannot be matched exactly with any method that uses color distribution functions (which includes curves adjustment, since it’s just the simplest subcase), but can only be approximated. The more color spaces that are used, the better the approximation.