[argyllcms] Re: Number of patches - well behaved printer?

  • From: Don Craig <dmc@xxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx, Terry Ritz <t.ritz@xxxxxxxxxxxxx>
  • Date: Tue, 22 Feb 2011 02:36:08 -0800

Hi Terry,

I have the same equipment (ipf8300 and ColorMunki),
but I consider myself an Argyll novice. I've just finished comparing
several profiles for Canson Baryta 310gsm from the following sources:

1. ColorMunki with X-Rite ColorMunki Photo software,
80 patches per 8.5 x 11 inch page, two pages.
(0.7739 square feet paper, 0.535 ml ink each per page)

So the good news here is you can print patch pages to your
heart's content, each one obtaining its color values from an
image you are eventually going to print. The bad news is
that at 80 patches per 8.5 x 11 inch print, it takes quite
a while to get a significant number of patches printed and
scanned. (Particularly if you wait a day between each
printing and scanning cycle - the ColorMunki software
wants to see the results of the previous scan before generating
the next target page.)  There's very little ability to vary
the number of patches or the page size.

I tried a profile made from two patch pages (160 patches) -
it wasn't bad but it wasn't good - it had a slight green cast,
and separation of dark tones could have been better.

2. I printed out the Bill Atkinson 1728 patch two-page A4 set
for the Eye-One I/O (Bill's files are at 
https://public.me.com/billatkinson )
and sent it off to Rob Reiter at lightroom.com.
(1.4068 square feet paper, 1.312 ml ink, total both pages)

Rob charges $35 to make a profile on his Eye-One I/O
($75 for three), and once he received the prints he emailed
me a profile within a few hours. This profile was pretty good,
dark tones nicely resolved, and neutral as best as I can tell
without going through the exercise of photographing the test
print beside a Munsell chart (or whatever they're called now)
and comparing color temperatures.

Definitely an improvement on the ColorMunki, with
the advantage that its done by an experienced guy
who has an ipf8300, so it's a good reference point.
You're supposed to wait a day or more before
measuring the patches anyway, which I used in this
case as transit time to Rob.

3. Using Argyll I printed out a 2048 half-size patch target
for the ColorMunki, 36 x 24 inches, 61 columns of 34 patches.
With a little practice and a foamcore straight edge I scanned
them all in about 20 minutes (after waiting a couple of days
watching pigment dry).

The ability to control the number and size of
patches and the size of the print made this process
a lot more practical than X-rite's. The disadvantage is
simply the amount of paper and ink being consumed
in one calibration.
(5.7448 square feet paper, 6.345 ml ink)

The results, as far as I can tell, are a wash with the
lightroom.com profile. Very slightly less separation
of the second darkest and darkest 16-step gray
tones, very slightly more separation between
a red and an orange red color patch. (Evaluation of
Bill Atkinson's test image under tungsten and under
California northlight. No reference illuminants here.)
Creating 3-D gamut plots showed slightly bigger gamut
for Argyll in orange and green, and plot solid wrinkles
in different places.

Frankly, if I didn't care about (really, eventually)
understanding the color management process
I wouldn't be messing with Argyll and I'd pay $35
for a fast turnaround profile when I needed it.
The stock X-Rite software that comes with the
C-Munki is pretty primitive - better than nothing,
but you better not get picky. Argyll is capable and
configurable to the point of being completely
opaque to the novice, but eventually repays all
that messing around.

And if I'm going to use a patch-greedy C-munki,
it's nice to be able to configure the patch layout
in Argyll. With the ipf8300 rolls are a lot more
convenient than sheets, and I tend to standardize
on 36", often printing two-up 18" wide. So a 36 x 24
target is convenient to make, and you can frame it
and call it art when you're done measuring ;-) .

So Terry, to answer your question, I'd suggest using
2048 patches. It's clear that 1728 patches also gives
good results, and I was not impressed with a 160
patch ColorMunki attempt. Looking at what Bill
Atkinson went through developing profiling tests
only 6 years ago we've obviously come a long way
with the new equipment, but I don't think a few extra
patches in a target is much of a problem. If you care
about ink/paper consumption don't use a ColorMunki.

Don Craig

PS: Commands for 2048 ipf8300 ColorMunki patches were:
targen -v -d 2 -G -f 2048 filename
printtarg -v -i  CM -h -p 900x600 -M 25 -T filename

PPS: The consumption numbers in () are from
           Canon's Accounting Manager

On 2/21/2011 11:34 PM, Pascal de Bruijn wrote:
> On Mon, Feb 21, 2011 at 11:19 PM, Terry Ritz <t.ritz@xxxxxxxxxxxxx> wrote:
>> I’m working with a ColorMunki, profiling papers for my new Canon 8300. The
>> 8300 has built in calibration technology and as such, it should be
>> reasonably well behaved. I’m printing on cotton rag matte papers, a fine art
>> baryta gloss paper and canvas.
>> How many patches would you suggest using in my targets?
>> Out of curiosity, how many patches would you use for a less well behaved
>> printer, say an older pigment ink printer?
> I've been able to make due with two sheets of A4 paper, in double
> density mode, so that's ~500 patches. But considering you're printing
> on high end paper, you might want to go for 1000 patches or so.
> I don't have such highend hardware. So do follow other folks' advice
> over mine :)
> Regards,
> Pascal de Bruijn

Other related posts: