Notes on Digital Color Printing
Last Updated July 8, 2000.
This is now sufficienly out of date that I wouln't consider it to be a
reliable predictor of current results. In general, the quality
and uniformity of digital prints has improved, so perhaps this should
be seen as only of historical interest.
As anyone who has ever
in a darkroom will tell you, color is really hard to get
For ordinary purposes, something close to accurate is good enough, and
many cases, accuracy isn't really the goal anyway; the goal is to
colors you like.
who are mass producing color prints don't have the luxury, or goal, of
colors especially for your taste; they should be producing as accurate,
and neutral a rendition of color as possible. This ought to be especially
true of prints from digital originals, since any manipulation of color
be done digitally. "Bits is Bits", the results should always be the
at least from the same lab. Any deviation from consistency is
evidence of poor quality control. No other excuses are possible.
This page features the
Chart image from my test set, which is my main reference for color and
of "ordinary" images. The full set of images I use to evaluate
can be seen here.
Each of the images in the test challenges the printing process in a
With the above in
what's below and weep.
||This is a single shot, so although the lighting may not be
it is consistent. Click on the photo to study the full size
All the 4x6 prints are from the same digital original, so in theory,
should look alike, or at least all from the same lab should look
original is from a Kodak PhotoCD scan of a slide, which I judged to be
most accurate among many I've taken with various types of film and
Note the 50% gray card positioned behind the Macbeth chart in the
||The original Macbeth Chart
||Ofoto (on a blue day)
|Shutterfly. Note that all three shutterfly prints have
extremely weak yellow.
Some Tentative Remarks
- Any evaluation of a
based on a few samples is extremely problematic. Every one of the
from whom I have repeated samples is inconsistent, though on the
some are better than others.
- To be fair,
that this is a really tough test; one that ordinarily would not be
At least half of the prints above are part of batches whose color would
close enough for everyday use. On the other hand, the rest range
marginal to awful. Especially in the digital world, don't let
bastards get away with it. Complain and ask for reprints or
until they get it right.
Additional Reading: Matt Dittrich has done similar
and reported his results.
Sharing site Smugmug has done
an elaborate comparison
test with a large set of photos and print providers.