ipf5000, information Options

[digitaldog]

Michael sent me two targets today he printed on two different Canon 5000's (same paper from the same box). Note this isn't really enough data to say anything conclusive but it is interesting. The targets contain 918 color patches of known value. I measured both on the i1 i0 and saved the resulting spectral data as text files. This allows me to load both into the ProfileMaker Pro MeasureTool module using what is called the compare mode. The two sets of 918 spectral data measurements are compared and depending on what formula I select, I get a deltaE value. Note I set MeasureTool to provide what I feel is the best way to calculate small changes in delta, that being deltaE 2000. I've uploaded the screen dump. Let me explain what is being shown here.

The worst 10% of the matches are shown with a yellow square around the color. This is useful to show you which solid colors are the farthest apart visually. You might notice one patch with a red border (it's a gray patch) and when I click on it, the deltaE is supplied below. So here's the skinny. The average deltaE is 2.81. The worst 10% are 6.99 which I think is pretty high. That one gray patch is 12.42 which isn't good at all (note that a deltaE of one or less isn't perceivable and a deltaE of 6 or less is consider "accpetalbe" matching although you can easily see the difference.

To give you an idea, I did the same test with an Epson 2200 however, I only printed one target, the other was from a customer so the papers where certainly not the same lot. This was Luster paper too. The worst 10% deltaE was 4.84, MUCH lower! The worst patch (a very dark gray) is a deltaE of 6.47.

I'd like to see far more samples from different printers as well as the same printer and paper being used over the course of days or even weeks to plot the stability. But this initial test does back up the one poster who said they build a profile and it wasn't all that usable later.

[John Hollenberg]

Andrew,

I would be curious to know the delta E 2000 between two Epson 4800, 7800 or 9800 targets you have measured on Epson Premium Luster. Presumably with calibration info burned into ROM at the factory these printers would be much closer together than even the Epson 2400. Since the Epson 4800 is the competing printer size wise, this information would be valuable.

--John

[digitaldog]

Andrew,

I would be curious to know the delta E 2000 between two Epson 4800, 7800 or 9800 targets you have measured on Epson Premium Luster.  Presumably with calibration info burned into ROM at the factory these printers would be much closer together than even the Epson 2400.  Since the Epson 4800 is the competing printer size wise, this information would be valuable.

--John

Me too but looking over all the saved data files I have, I can't find any two from the same printer and paper combo's. I have lots of 4800 or 9800 files but I'm having difficulty finding two from the same paper (go figure). That's why I used the 2200.

[michael]

Here's a question that occured to me a few days ago, and this seems like as good a time as any to bring it forward.

Epson claims that all "Pro" printers are individually calibrated and linearized at the factory. But, since these printers (4800, 7800, 9800) are deliverd with dry heads and ink lines how is this process done? For each printer / head combination to be calibrated means printing, yet if prints were made with that system, how do they then arrive dry and clean?

Michael

[John Hollenberg]

Here's a question that occured to me a few days ago, and this seems like as good a time as any to bring it forward.

Epson claims that all "Pro" printers are individually calibrated and linearized at the factory. But, since these printers (4800, 7800, 9800) are deliverd with dry heads and ink lines how is this process done? For each printer / head combination to be calibrated means printing, yet if prints were made with that system, how do they then arrive dry and clean?

Michael

When I bought my 9600 there were still some fainter colors in some of the lines. When I posted, I was assured by those on Epson Wide Format group that this was normal as printers are tested with ink before leaving the factory. I guess they just didn't do as good a job cleaning out the lines on mine.

--John

[digitaldog]

Can't get this darn screen dump to fly so to view it, click this link:

http://www.digitaldog.net/files/deltaE.jpg

[John Hollenberg]

Can't get this darn screen dump to fly so to view it, click this link:

http://www.digitaldog.net/files/deltaE.jpg

That doesn't exactly look like a random distribution of colors with the yellow boxes (worst 10%). The significant differences are confined to certain colors. Looks like you would definitely want a custom profile. The key question is whether a particular printer changes significantly over time.

--John

[digitaldog]

That doesn't exactly look like a random distribution of colors with the yellow boxes (worst 10%).  The significant differences are confined to certain colors.  Looks like you would definitely want a custom profile.  The key question is whether a particular printer changes significantly over time.

--John

From the measured data, a custom profile was generated.

There's no real rhyme or reason why some colors have the yellow squares. We're simply seeing the differences between two sets of measured data per color. I frankly don't know why those colors differ as much as they do. Interesting that the worst patch is a gray however. I suspect that's not good.

Yes, I'd love to measure targets from the same printer over time (something we will do). So again, this is just a small snippet and nothing to take to the bank other than these two printers are not behaving in a similar fashion as much as I'd like (and see from Epson).

[BernardLanguilli...]

It would appear that the typical Epson solution "buy Colorbyte and use their profiles as is" won't do with the Canon, right?...

With such a large variation from printer to printer, it wouldn't make too much sense to use pre-cooked profiles.

It looks like my Monaco might become useful again.

Cheers,
Bernard

[digitaldog]

It would appear that the typical Epson solution "buy Colorbyte and use their profiles as is" won't do with the Canon, right?...

Is there any plans from ColorByte to support this printer? I'd think that's a key question.

[BernardLanguilli...]

Is there any plans from ColorByte to support this printer? I'd think that's a key question.

Hi Andrew,

I thought I saw comments from Michael hinting that the were planning to support it within 3 months, or did I misunderstand?

Or are you saying that you results might lead them to reconsider the whole story?

By the way, isn't there a possibility that the current discrepency results from differnt quality of head alignements with the 2 printers? Isn't this related to the graininess seen in some samples when the heads are pooly aligned? I mean, doesn't such a graineness endup affecting the color?

Cheers,
Bernard

[digitaldog]

Or are you saying that you results might lead them to reconsider the whole story?

By the way, isn't there a possibility that the current discrepency results from differnt quality of head alignements with the 2 printers? Isn't this related to the graininess seen in some samples when the heads are pooly aligned? I mean, doesn't such a graineness endup affecting the color?

Cheers,
Bernard

I can ping John P at ColorByte and ask him. Took an awfully long time just to write the 2400 driver but it's possible they are working on the Canon.

Can't answer you question about the alignment. I just measured the targets.

[PlatonicCaveman]

I can ping John P at ColorByte and ask him. Took an awfully long time just to write the 2400 driver but it's possible they are working on the Canon.

Can't answer you question about the alignment. I just measured the targets.

Just received an email from John today...his response was between 60 and 90 days.

[John Hollenberg]

Just received an email from John today...his response was between 60 and 90 days.

Translated into earth time, that would be 6-9 months. Or did I misread the track record of Colorbyte when it comes to release dates?

--John

[AWOL]

This may be obvious to everyone but me. But this does seem like the place to ask. I just got a ipf 5000 and am getting ready to profile it with monaco Profiler platinum. Is everyone making profiles as a RGB printer or as a seven color printer? Can the printer even recognize a 7 color profile or does everything have to be sent to it as a RGB 8 or 16 bit file? I don't have a roll feed yet so dealing with the top "tray" has been pretty much impossible with canvas.

AWOL

6. ICC Profiles. This question is for Andy from Digital Dog. I was told that the supplied Canon ICC profiles for theirm media are in 8 bit and that the Photoshop plug-in detects this and uses an algorithm to convert the the file to 8 bit so that it can be printed as the printer accepts only 8 bit files? I know that with the new Colorvision PrintFixPro that you can create 16 bit ICC profiles. Is it even worth using a 16 bit profile. Does any of what I say about the plug-in printer sound accurate? I thought I was well educated in colro mangement, but I am finding that there is a whole level above where I operate from the operational level to the scientific level.

[alanrew]

Michael sent me two targets today he printed on two different Canon 5000's (same paper from the same box). Note this isn't really enough data to say anything conclusive but it is interesting.
<snip>

One possibility that hasn't been mentioned yet is that the two printers may be using ink from different batches, and the underlying problem is that Canon haven't got their consistency tight enough with ink manufacture yet.

This may sound unlikely, but I recall a problem with Epson inks a few years ago when batches of their ink made in China was noticeably different from ink made in Japan.

So the printers might be well-calibrated (using a 'standard' ink in the factory - but what if that varies as well??) but the ink isn't.

So how about a test involving 2 printers but 1 set of cartridges? i.e. print target from one, transfer carts to other printer, print target from other printer, then compare.

This would help Canon to narrow the problem down to 'printer' or 'ink'.

Of course, _both_ ink and printers might vary :-)

Regards,

Alan

[digitaldog]

This may be obvious to everyone but me. But this does seem like the place to ask. I just got a ipf 5000 and am getting ready to profile it with monaco Profiler platinum. Is everyone making profiles as a RGB printer or as a seven color printer? Can the printer even recognize a 7 color profile or does everything have to be sent to it as a RGB  8 or 16 bit file? I don't have a roll feed yet so dealing with the top "tray" has been pretty much impossible with canvas.

AWOL

Profile it as an RGB device. The driver will do the conversion from RGB to the 12 color Seps. Quickdraw and GDI drivers only understand RGB anyway.

I'm not sure where all this 16-bit profile stuff is coming from but I suspect Canon has some documentation somewhere that is causing confusion. Profiles unlike images don't have a bit depth. Some products allow you to control the number of nodes used to build the profile but this isn't a "8-bit versus 16-bit" issue whatsoever. In fact, it is sometimes the case that a smaller node used to build the profile will produce a better result. So if you guys want to consider 16-bit files fine but don't worry about profiles. Get good ones and don't concern yourself with the node or "bit depth".

After you build the profile, you can use either an 8-bit or 16-bit file just like you can with ANY file you print out of Photoshop.

[digitaldog]

One possibility that hasn't been mentioned yet is that the two printers may be using ink from different batches, and the underlying problem is that Canon haven't got their consistency tight enough with ink manufacture yet.

Hardware or ink, doesn't really matter with respect to consistency between this product and the competition (at this point).

Again, with only two targets under our belt, this is far from a definitive statement however, the initial results are not comforting.

[JeffKohn]

Hardware or ink, doesn't really matter with respect to consistency between this product and the competition (at this point).

Again, with only two targets under our belt, this is far from a definitive statement however, the initial results are not comforting.

Personally I'm far less interested in consistency between two printers than consistency of one printer over time. I guess the former issue is of interest to those using canned profiles, but with custom profiles I just don't see it as something to be concerned about.

[JeffKohn]

One thing I'm wondering about is the 16-bit driver interface, and how "open" Canon will be with this. Are we only going to have the option of using the two Plug-ins Canon has provided so far? Will Canon give Mike Chaney the time of day so that he can add 16-bit support to QImage? What about Adobe Lightroom?

Right now I use QImage for my 2400 not so much for the interpolation/sharpening (in fact for large prints I disabled those and do my own), but for the page layout capabilities and the convenience of associating driver settings with ICC profiles. With a large-format printer the layout capabilities would be even more important; I'd hate to be stuck using "Print with Preview" for all my printing needs....