Epson 2200 even produces good results on plain paper?

[Greg]

I've just received my Epson 2200 printer, and I'm *very* pleased indeed!

One little thing which I've noticed is that the results on plain paper seem
to be quite a bit superior to my previous Epson 1270 - this is handy for
those of us who don't have a second printer for "run of the mill" type
stuff. (there are occasions where I just want to do a quick & dirty colour
print on plain paper). Anyone else notice this? It has been a very long
time since I've tried plain paper in the 1270 (and I really can't be
bothered re-connecting my 1270 now , but I recall thinking that for
photos, it was bordering on unusable, and that I'd have to resort to
printing out the photos on photo paper and sticking them in the document. On
the other hand, results from the 2200 are entirely usable. Perhaps the
pigmented inks of the 2200 are more compatible with plain photocopier type
paper than the dyes used by the 1270?

Greg.
p.s If anyone actually notices that my post is from Australia, and wonder
why I have the 2200 and not the 2100, it's because I got it mail order from
overseas.

 

[CWatters]

I've just received my Epson 2200 printer, and I'm *very* pleased indeed!

p.s If anyone actually notices that my post is from Australia, and wonder
why I have the 2200 and not the 2100, it's because I got it mail order from
overseas.

Out of interest did you get US model 2200 or the Asian model whose number
escapes me? The higher price 2100 normally comes with "free" grey scale
balancer, CD printing tray and the paper cutter but I've not used these yet
with my 2100

 

[Greg]

Out of interest did you get US model 2200 or the Asian model whose number
escapes me? The higher price 2100 normally comes with "free" grey scale
balancer, CD printing tray and the paper cutter but I've not used these yet
with my 2100

I got the 2200, and yes, I am a bit disappointed that I don't have the
ability to print on to CDs. I'll probably try to obtain the grey balancer at
some stage. I *did* get a paper cutter with my 2200 though, as standard.

It's 110V only, so I also had to pay AU$50.00 for a step-down transformer.

Greg.
p.s One reason I bought this printer mail order from the U.S was simply that
work gave me a gift voucher at Amazon, and Amazon Australia don't stock it.
Had to jump through hoops to have it shipped though, because Amazon only
ship books and videos to Australia - just lovely.

 

[Pard]

I got the 2200, and yes, I am a bit disappointed that I don't have the
ability to print on to CDs. I'll probably try to obtain the grey balancer at
some stage. I *did* get a paper cutter with my 2200 though, as standard.

It's 110V only, so I also had to pay AU$50.00 for a step-down transformer.

Greg.
p.s One reason I bought this printer mail order from the U.S was simply that
work gave me a gift voucher at Amazon, and Amazon Australia don't stock it.
Had to jump through hoops to have it shipped though, because Amazon only
ship books and videos to Australia - just lovely.

But you can get the grey balancer by downloading it from Epson's
Russian site. They also have the Epson CD-Printer software if you
need it.

 

[Greg]

But you can get the grey balancer by downloading it from Epson's
Russian site. They also have the Epson CD-Printer software if you
need it.

Thanks, but the grey balancer is not just software - there's a test chart
for it too.

FWIW, the greyscale of my 2200 already looks very neutral to me.

Greg.

 

[CWatters]

It's possible to use a Kodak chart. See here for more info...

http://www.luminous-landscape.com/reviews/printers/making-beer.shtml

 

[Greg]

It's possible to use a Kodak chart. See here for more info...

http://www.luminous-landscape.com/reviews/printers/making-beer.shtml

Ah, thanks for that

I suppose it'd be possible to get someone like Wolf Faust to make up a
custom chart to match Epson's exactly, to get around the 45% and 20%
patches being slightly off.

Greg.

 

[Greg]

I've just noticed something of concern in my attempts at linearising the raw
response of my 2200.
I printed out a test chart, consisting of seperate red, green, and blue
stepped wedges, consisting of 23 patches each.
The green wedge is almost flat from code 165 to 217 with an L* of about 32
(a* & b* are also relatively constant - no change of hue), with a step up to
about 35 at the next patch, which has a code of 230. This step happens to
occur right where the wedge wraps around on the page - the two patches are
at opposite ends of an A4 sheet of paper.
I've now printed those two patches again, but in close proximity on the
page. I included an extra patch with a code of 224 inbetween them.
The discontinuity is *not* present on this new set of patches - the
luminance values are 34.4, 35, and 35.2.

I am using a Colorsavvy CM2C colorimeter to make the measurements.

Has anyone else ever noticed this kind of thing? I'm hoping that it's just
that I did that initial test chart print without having let the printer warm
up sufficiently. But I don't think I've ever seen it stated that an inkjet
should be allowed to warm up before printing test charts. Or am I just
expecting too much repeatability from this printer?

I haven't looked closely to see whether any other response differences have
occured. I'm going to print out the entire test chart again and compare
every patch with the first print.

Greg.

 

[Greg]

Oops - I meant to put this reply in the thread "Why are Epson inkjet
printers so non-linear in raw mode?". Sorry about that.

 

[Flycaster]

Oops - I meant to put this reply in the thread "Why are Epson inkjet
printers so non-linear in raw mode?". Sorry about that.

Greg, I do a fair amount of profiling with this printer and I'm confused by
your "attempts at linearising the
raw response of my 2200." Just how do you propose to do this with the OEM
driver- or, are you using a RIP driver?

In any event, I can assure you that these printers are *anything* but linear
in their native (raw) output from the OEM RGB driver. Frankly, why they are
so far off from a neutral calibration state is a real puzzle to me (and
others), and that is the principle reason why they can be difficult to
accurately profile.

 

[Greg]

Greg, I do a fair amount of profiling with this printer and I'm confused by
your "attempts at linearising the
raw response of my 2200." Just how do you propose to do this with the OEM
driver- or, are you using a RIP driver?

I've got this idea from Bill Atkson. What Bill seems to do is linearise the
response before profiling.
It makes a lot of sense to me. If the response is perceptually linear, the
test chart from the
profiling tool will print with more perceptual uniformity, rather than the
patches being bunched up
in certain colours. So, it should allow the profiler to create a better
characterisation of the printer.
I apply the linearisation curve in Curves in Photoshop to the profililer's
test chart image data before printing.
The profiler has no knowledge that I have done this - once it has created
the ICC file, I incorporate the curve
into the ICC file, by putting the pre-linearisation curve into the output
lookup table of the B2A tags, and the inverse of the
curve into the input lookup table of the A2B tags. I understand some high
end profiling applications directly cater for
pre-linearisation - the profiler I'm currently using does not and at the
moment I'm using a hex editor to add these
curves into the ICC profile.

This link has a brief summary of what Bill's procedure:
http://www.imagingrevue.com/forums/?thread=1206 although I get the
impression Bill had to do a lot more than
just pre-linearise the response when he made the 7600/9600 profiles, though.

If the profiler test chart had enough patches, pre-linearisation would not
be required I don't think. My largest
test chart has about 700 patches, but Bill does the pre-linearisation even
when using 1440 patch charts.

In any event, I can assure you that these printers are *anything* but linear
in their native (raw) output from the OEM RGB driver. Frankly, why they are
so far off from a neutral calibration state is a real puzzle to me (and
others), and that is the principle reason why they can be difficult to
accurately profile.

It's one thing being non-linear, and it's another producing different
colours at different times under different
conditions. In fact, even this would bother me less if the *shape* of the
response remained the same.

Greg.

 

[Flycaster]

Greg:

I'll take a look at the link and see what Bill has to say. I have a lot of
respect for him, so it should be interesting. I currently use Gretag
Macbeth Photo to profile my 2200, which uses one of Bill's 918 patch test
targets.

"It's one thing being non-linear, and it's another producing different
colours at different times under different
conditions. In fact, even this would bother me less if the *shape* of the
response remained the same."

I have not noticed this with my 2200. My color ramps show remarkably little
drift, such that I've only had to re-profile this printer once in a little
over a year. My pricipal problem has been with the VFA paper - the RGB
driver does a horrible job with dark colors using this paper and ink
combination, and the profile has a very difficult time correcting for it.
Some images come out just fine, others require a change-over to Archival
Matte.

Other papers are great, especially the Premium Luster. In fact, I've never
had to re-profile it.

 

[Flycaster]

OK, I've read his comments, but now I have a question for you. At the end,
he (Bill?) says. "Then I use ColorSavvy's SMPKit software to build an
accurate ICC printer pro-file from the CIE-Lab color measurements and the
actual RGB input data that was sent to the printer after linearization."

All fine and dandy, but HOW does he then input the PS curve information into
the profile such that it is applied first, prior to the actual LUT table
being applied? Or, does he just apply the "linearization curve" via a PS
Action on each and every file prior to invoking the profile conversion?

(Either way, it would seem to just be whole lot easier, not to mention more
accurate, to use a RIP calibration/profiling package instead.)

 

[Greg]

I don't know how Bill does it, but as I said earlier, I add it manually,
using a hex editor.
The AMP file (which Photoshop creates) can be pasted in to the ICC file very
easily,
using a hex editor. The AMP data goes directly into the output lookup table
of the B2A
(connection space to device space) tags. One needs to put the inverted
version into the
*input* lookup table of the A2B (device to connection space) tags. I'm using
Excel to do
the inversion.

I read a longer report from Bill on what he did to profile the 7600/9600
printers. Bill wrote
some software of his own to assist with it, apparently. I'm still not sure
of the exact procedure
he used for the 7600/9600 - the pre-linearisation wasn't the only special
thing he had to do.

Or, does he just apply the "linearization curve" via a PS
Action on each and every file prior to invoking the profile conversion?

It can be done manually, however the linearization curve has to be applied
*after* the profile
conversion, not before. Remember, the curve is applied to the test chart
image before it's printed. Thus,
the profiling software *thinks* the printer has a response which *includes*
the
linearization curve.

Greg.

 

[Greg]

Actually, regarding RIPs, I'm not sure whether Bill tried RIPs with the
7600/9600 - it'd
be interesting to know this. I've read about how good the ImagePrint RIP is
with the Epsons.
Certainly, looking at the "raw" response of the ImagePrint RIP (at Ian
Lyon's site), it's a lot
cleaner than the raw mode of the standard Epson driver!

Greg.
p.s A few more hours and I'll measure my new test chart print.

 

[Flycaster]

Actually, regarding RIPs, I'm not sure whether Bill tried RIPs with the
7600/9600 - it'd
be interesting to know this. I've read about how good the ImagePrint RIP is
with the Epsons.
Certainly, looking at the "raw" response of the ImagePrint RIP (at Ian
Lyon's site), it's a lot
cleaner than the raw mode of the standard Epson driver!

Absolutely. Frankly I think this whole exercise is kind of a waste of
time - in essence all you're doing is slamming the file via even larger
profile moves, which is a very poor substitution for having a neutrally
calibrated printer.

Tell you what...you make one for Premium Luster and I'll trade you a Gretag
Macbeth RGB profile for yours and we'll compare. I know they won't be right
on the money due to the slight printer differences, but they should be very,
very close. I'm curious...whaddya think?

 

[Greg]

RIP

Absolutely. Frankly I think this whole exercise is kind of a waste of
time - in essence all you're doing is slamming the file via even larger
profile moves, which is a very poor substitution for having a neutrally
calibrated printer.

I disagree that it's a waste of time. As I said, if the profiler test chart
had enough patches, the pre-linearisation
wouldn't be required, because there'd be enough patches to characterise the
non-linearities
very well anyway. I've already tried it both with and without the
pre-linearisation, and a soft-proof
of the version without the pre-linearisation loses some subtle detail. This
problem doesn't occur when
I pre-linearise, however I have not done exhaustive testing. Having said all
this, I'd feel a lot more comfortable
in accepting a profile with 1440 patches - I can see that it *would*
characterise the non-linearities of my printer
pretty well. The fact that Bill Atkinson still goes to the trouble of
pre-linearising, even with 1440 patches,
surprises me a little.

Tell you what...you make one for Premium Luster and I'll trade you a Gretag
Macbeth RGB profile for yours and we'll compare. I know they won't be right
on the money due to the slight printer differences, but they should be very,
very close. I'm curious...whaddya think?

That sounds like an interesting test. Before we do that, do you want to
trade colorimetric
measurements of a given test chart, to see how our printers really do
compare?
(also, I'd prefer to use Premium Semigloss, because that's currently my
preferred media,
but if you insist, I will get some Premium Lustre)

Greg.

 

[Greg]

Flycaster:
This link mentions pre-linearization:
http://www.newsandtech.com/issues/2001/12-01/pt/12-01_kober.htm

Just thinking about this test - perhaps a more scientific way would be for
you to
create two profiles for your printer, using your software - one without
pre-linearization,
and one with. I could make the pre-linearization curves for your profile if
you wish, by
sending you the test chart that I am using for the pre-linearization. You
could send the
measurements back to me, and I'd then create the binary pre-linearization
data for you to
paste in to your ICC profile. (it is all a bit tedious, but I could walk you
through the steps)
Or perhaps your profiler already caters for pre-linearization?

Note that I really don't know whether I am going about the pre-linearization
the best way.
For example, I am linearizing to a gamma of 2.2, whereas it appears to me
that the ImagePrint
RIP uses a somewhat lower raw gamma. I'm also curious to know whether
linearizing the RGB
response will also produce a linear CMY response - I'm not entirely sure
that it will.

Greg.

 

[Flycaster]

[snip]

That sounds like an interesting test. Before we do that, do you want to
trade colorimetric
measurements of a given test chart, to see how our printers really do
compare?
(also, I'd prefer to use Premium Semigloss, because that's currently my
preferred media,
but if you insist, I will get some Premium Lustre)

Sure, we could trade color charts but I find looking at a profiled test
print to be a far better "proof" (not to mention more interesting) than
doing the old scratch and sniff with my photospectrometer on a bunch of
color swatches. If you'd like, I can also send you the TXT file so you can
actually see the ramp info, rather than cracking the profile.

No problemo with Premium Semi-Gloss: I've got a 2880 and a 1440 Relco,
whichever you want. FWIW, I don't use this paper much except in 4" rolls
(for high volume 4x6 work) - the bronzing is little touchy with this paper,
imo.

Personally, I'm curious to see if your raw, unmanaged output is a heavily
"cyan-green" as mine and everyone else's Epsons seem to be. THAT is the big
problem with these printers, at least according to my contacts at Gretag
Macbeth and Chromix: their OEM RGB driver output is so far off neutral that
the profile adjustments are friggin' huge.

Let me know, and I'll send you whatever you want in the next day or so and
we'll compare notes. (I hope you have cable or DSL...these files are not
small.)

 

[Greg]

Personally, I'm curious to see if your raw, unmanaged output is a heavily
"cyan-green" as mine and everyone else's Epsons seem to be. THAT is the big
problem with these printers, at least according to my contacts at Gretag
Macbeth and Chromix: their OEM RGB driver output is so far off neutral that
the profile adjustments are friggin' huge.

Yes, judging by the curves I'm seeing, the output will have excessive green.
My previous
printer (a 1270) had much the same characteristic. Also, prints done with
the provided Epson profile
for Premium Semigloss look entirely normal - this means that my printer is
at least in the right ballpark
regarding colour balance, so it's bound to produce raw prints which look the
same as yours and everyone
elses.

I've printed off my linearisation test chart again. This time I warmed the
printer up first by doing a dummy print first.
The green wedge is a lot better this time - it doesn't
have the strange levelling off and the sudden jump. And for the rest of the
curve, it follows the
first one very closely indeed. The red channel follows the first one very
closely for it's entirety. The blue
tracks quite well, but both the first and second responses are a bit
wiggly - I think it's
because the luminance values are considerably lower for blue, so the curves
aren't as smooth, due to
tolerances in various places in the system. (including the colorimeter
itself, perhaps).

I'm not sure what to think about the strange kink in the green response that
occurred on the
first print of the chart. If it was just a nozzle blockage, I think it must
have been a very subtle one.

Greg.