IDT technical questions

Hi I started to code the IDTs into DCTLs like Paul did, for learning purposes. Checking around I have two questions:

-Now the Alexa LogC V3 have a 1DLut instead of a formula, why is that? to make the process more fast?

-When I try to swim into the deep waters of Canons IDT I found this ¿Polinomial? 19x3 matrices that escape my knowledge, could you point me where I can find more info about why Is that used?

Thanks in advance,

1 Like

I need to check those matrices, maybe it’s to get differenciated color matrixing per zone of brightness? That would solve some problems as in general 3x3 are not good enough (cf the old LED light fix)

Hi, first of all, thanks for the interest. I checked and the c300 has a more common aproach but the c100 remains the same(I read some stuff as it being too complex and that they simplified It). I also found some info about polinomials, but nothing else(not a right explanation). Here you have the IDT
Canon_EOS_C100_IDT_A_D55_Ver.1.0.ctl (3.8 KB)

This is probably too complex for some implementations, like in camera processing. It allows to have more refined gamut mapping than just 3x3 that can only do cross-color and saturation; would be interesting to plot that.

1 Like

Could be Finlayson et al. (2015) without the Root Polynomial Expansion:

>>> colour.characterisation.colour_correction_matrix_Finlayson2015(M_T, M_R, 4).shape
(3, 22)
>>> colour.characterisation.colour_correction_matrix_Finlayson2015(M_T, M_R, 3).shape
(3, 13)
>>> colour.characterisation.colour_correction_matrix_Finlayson2015(M_T, M_R, 4, False).shape
(3, 34)
>>> colour.characterisation.colour_correction_matrix_Finlayson2015(M_T, M_R, 3, False).shape
(3, 19)



1 Like

Yeah I did check at Finlayson in past, I was not sure I was on the right track, I will check again. Thanks Thomas and Cedric to point me in the right direction.