I’ve been doing some trials comparing printer profile results, here on a
regular luster/glossy style photo paper.
Targets are made with preconditioning using a profile for another glossy paper
on same printer (here a sheet fed Canon Pro-1000), no special settings (expect
-G for optimised points).
Measured with an i1Pro, and profiles made with high quality, no other special
settings for these tests.
Profile checks against the used targets look to be a tight fit, all around this
result sample (CIEDE2000):
Profile check complete, errors(CIEDE2000): max. = 1.020662, avg. = 0.237625,
RMS = 0.282584
A2B tables were evaluated for different profiles (using different
preconditioned patch counts) against a separate 256 patch measured target (same
for all comparisons, but not one of those used for making any of the profiles,
to get somewhat different device values that profiled with - made with same
preconditioning as when making profiles). Using profcheck to compare this
separate comparison target with profiles done with 256, 512 and 1024 patches.
The results for 512 patch profile is
Profile check complete, errors(CIEDE2000): max. = 5.940424, avg. = 2.283050,
RMS = 2.467218
and for 1024 patches, it’s
Profile check complete, errors(CIEDE2000): max. = 4.427050, avg. = 1.517033,
RMS = 1.653229
For 256, it look a lot like 512 (missed the specifics in my notes just now).
I’m a little surprised that 512 or 1024 patch profiles don’t make a better fit
than this, when comparing to another set of device values than used for the
profile generation (where we saw max. = 1.020662, avg. = 0.237625, RMS =
0.282584). Perhaps it’s perfectly normal? I just thought a 1024 preconditioned
patch set covered the gamut so well that CIEDE2000 numbers would do better on
comparison targets.
If I compare to another 256 patch target done without preconditioning, the
result against the 512 and 1024 patch profiles are respectively
Profile check complete, errors(CIEDE2000): max. = 2.849606, avg. = 0.654959,
RMS = 0.784418
Profile check complete, errors(CIEDE2000): max. = 2.374234, avg. = 0.748152,
RMS = 0.870289
which seem a lot better, and matching my hopes.
I don’t need super accurate profiles, and have normally settled for just 512
preconditioned patches, and had hoped that gave sub de2k of max 3 and avg
around 1-2, but as shown don’t achieve that - if comparing to another
preconditioned 256 target).
For reference, an OEM profile against the same 256 conditioned and 256 default
unconditioned patches look like
Profile check complete, errors(CIEDE2000): max. = 6.823465, avg. = 3.852290,
RMS = 4.046874
Profile check complete, errors(CIEDE2000): max. = 4.165762, avg. = 2.065982,
RMS = 2.208021
Any comments on this, what to expect?
Best regards
Henrik
