If you’ll pardon a bit of diatribe…
IMHO, the ‘best practice’ is to find or measure your camera’s full spectral response and use that to make whatever profiles you need. Using a target shot just considers however many patches are in the target, and is limited to the illuminant temperature that shines on those patches.
For your camera, here’s a dcamprof-digestible dataset, courtesy of ImageEngineering.de:
{
"camera_name": "Sony Alpha 6600",
"ssf_bands": [380, 755, 5],
"red_ssf": [
0.0125712, 0.00678265, 0.000994118, 0.00072877, 0.000463422, 0.0107028,
0.0235683, 0.0305066, 0.0374449, 0.0385045, 0.035645,
0.03399, 0.032335, 0.0302935, 0.0286661, 0.0286953,
0.0287244, 0.0296471, 0.0305698, 0.035592, 0.0394131,
0.0384298, 0.0374466, 0.0395758, 0.041705, 0.0471601,
0.0534466, 0.0662119, 0.0784421, 0.0898697, 0.101297,
0.0838789, 0.0674355, 0.0548917, 0.042348, 0.0400682,
0.0377885, 0.0468381, 0.0587201, 0.188485, 0.336915,
0.469139, 0.601363, 0.580983, 0.560602, 0.521388,
0.453926, 0.417786, 0.389477, 0.344798, 0.300118,
0.274492, 0.248865, 0.218306, 0.186514, 0.158595,
0.131644, 0.104781, 0.0816282, 0.0640417, 0.0461679,
0.0278632, 0.0120434, 0.00616374, 0.00137972, 0.000978346,
0.000576972, 0.000340114, 0.000185717, 0.000155012, 0.000124307,
0.000125662, 0.000140115, 0.000174217, 0.000208319, 0.000183584
],
"green_ssf": [
0.0589112, 0.0347282, 0.0105452, 0.00833358, 0.00612197, 0.0203351,
0.0386544, 0.0610374, 0.0834205, 0.100768, 0.114758,
0.138551, 0.162344, 0.182548, 0.204652, 0.23436,
0.264068, 0.308756, 0.353443, 0.420315, 0.480168,
0.511947, 0.543725, 0.614117, 0.684508, 0.756464,
0.828812, 0.880352, 0.92564, 0.961551, 0.997462,
1, 0.997757, 0.976387, 0.955017, 0.921544,
0.88807, 0.825984, 0.756745, 0.690445, 0.623308,
0.553816, 0.484324, 0.411971, 0.339617, 0.264251,
0.184365, 0.14262, 0.110409, 0.0881126, 0.0658167,
0.0568173, 0.0478179, 0.0388734, 0.0299426, 0.0248718,
0.0207661, 0.0183143, 0.016055, 0.0140844, 0.0112758,
0.00721004, 0.00363546, 0.00202544, 0.000716353, 0.000610939,
0.000505525, 0.000380107, 0.00028616, 0.000239418, 0.000192676,
0.000218906, 0.000240041, 0.000253534, 0.000267027, 0.000251438
],
"blue_ssf": [
0.127183, 0.0648246, 0.00246642, 0.00234576, 0.00222511, 0.0777864,
0.172268, 0.273711, 0.375154, 0.443668, 0.490231,
0.556089, 0.621948, 0.666944, 0.705494, 0.718257,
0.73102, 0.727706, 0.724392, 0.707639, 0.685088,
0.639344, 0.5936, 0.53582, 0.478039, 0.409391,
0.338026, 0.280182, 0.228516, 0.186115, 0.143715,
0.121981, 0.101072, 0.0834588, 0.065846, 0.0522709,
0.0386958, 0.0299489, 0.0224091, 0.0183692, 0.0156467,
0.013588, 0.0115292, 0.00984765, 0.00816608, 0.00660499,
0.00522464, 0.00473266, 0.00446277, 0.00442838, 0.004394,
0.00459608, 0.00479816, 0.00523804, 0.00573737, 0.005856,
0.00587945, 0.00524482, 0.0044813, 0.00352444, 0.00249338,
0.00135102, 0.000399788, 0.000213103, 6.06555e-05, 4.51565e-05,
2.96575e-05, 3.27286e-05, 2.93274e-05, 1.62175e-05, 3.1077e-06,
2.40473e-05, 3.87158e-05, 4.39774e-05, 4.92391e-05, 5.58638e-05
]
}
pixls.us doen’t accept .json files as attachments, so you’ll have to copy-paste the above into a .json file.
With this, you can make matrix or LUT ICCs corresponding to any color temperature without having to fiddle with shooting targets without glare. Further, you can use domain-specific spectral references, e.g., skin tone datasets, to ‘train’ your profile. I use open-source dcamprof to do all this, but the commercial Lumariver is based on dcamprof, both written by the RawTherapee developer who incorporated DCP handling, one of about 3 softwares I’d consider buying.
Now, to put this all in perspective, I have such datasets for all three of my cameras, but I only use profiles from them when I run into extreme-hue problems like blue LED theatrical lighting. I have found the matrices that come with raw processors to be pretty good for most use.
—and it was underexposed. So next time, I’ll take a series of shots so that I can get close to L=96% for the white patch without any digital exposure compensation. … also the other tips concerning glare and aperture.
