There are TRC tags from ICC profiles, however Pillow is currently unable to extract those data.
I'll find out if I can extract it from the raw byte data.
There are 2 different standards to be used:
ICC v2
Have TRC with a curve data values. Can be interpolated per channel single dimensionally with scipy. or
Have TRC with a gamma value (ICC spec, curveType count value = 1)
ICC v4
Have TRC with a curve function as described in ICC v4 specification (page 71). Should be easily parsed to transfer functions.
Extra notes:
Embedded ICC profiles also have primaries coordinate and white point / illuminant that can be accessed with Pillow. I have yet to find out if colour-science can make a 'synthetic' profile from those parameters, that can be displayed later on the colour plot.
If all of the points above are possible, it will improve the accuracy of the plot significantly. And I don't have to rely on string parsing for the automatic profile detection.
kampidh
commented
2 years ago
There are TRC tags from ICC profiles, however Pillow is currently unable to extract those data.
I'll find out if I can extract it from the raw byte data.
There are 2 different standards to be used:
ICC v2 Have TRC with a curve data values. Can be interpolated per channel single dimensionally with scipy. or Have TRC with a gamma value (ICC spec, curveType count value = 1)
ICC v4 Have TRC with a curve function as described in ICC v4 specification (page 71). Should be easily parsed to transfer functions.
Extra notes:
Embedded ICC profiles also have primaries coordinate and white point / illuminant that can be accessed with Pillow. I have yet to find out if colour-science can make a 'synthetic' profile from those parameters, that can be displayed later on the colour plot.
If all of the points above are possible, it will improve the accuracy of the plot significantly. And I don't have to rely on string parsing for the automatic profile detection.