I am at my wits end with this problem I am having. It has been happening since I originally got this printer (epson 960). I had a profile made by Dry Creek Photo. We had some problems making the profile and I had to send the targets in changing settings a couple of times to try and nuetralize the darker tones until I finally got a OK profile that I could live with. However it never corrected the weird paintbrush effect I have in the darker brown tones pictured below so I delve into the problem once again. I called up Chromix and I spoke with Steve about it and he said they would give it a try and was confident that it could be fixed and if I wasn’t happy there would be no risk other than a few dollars in postage. Cool, Done. I promptly recieved my new profile only to discover that not only was the profile still giving me the same problems but the soft proofing was not even close to the output! Called Chromix tech support as well as emailed I was told the only options were that I either printed the target incorrectly or the printer is having problems. I was pretty confident in my ability in going through the print dialog box so we determined it was a bad printer. SUPER! Finally solved my age old problem…or did we?
I promptly started shopping for a replacement and had my eye on the Epson R800. I had previously owned a 2200 and this was suppose to have a wider gamut. I decided to download the publicly available profile from epsons web site to do some soft proofing and compare color gamuts between the two printers. To my suprise it had the same problem that my 960 profiles had when soft proofing. Which on my 960 does show up in the final print so it is not a soft proof or monitor calibration issue. I then downloaded the 1800 and 2400 profiles for the same paper and found the problem gone!
So, my question is this: What is causing this problem? Is there a difference between the large format printers and the small format printers with regards to color gamut or print driver? Is there a problem with the profile that is causing it to misinterpret the colors and do this weird painting thing in the darker browns? Please help me get out of this nightmare so I can either get a good profile for my 960 or buy a R800.
When I’ve seen posterization issues such as this, it has usually been the result of poor or currupt linearization, less than ideal paper, or to a lesser degree, choice of rendering intent.
We had this issue on our HP 20s (when we still had them) and could not cure it until switching paper. Posterization has cropped up on our Epson 7600/9600/4000 printers before, but it was either the result of not enough ink limiting during profile creation or a currupt linearization chart being measured (the result of a linearization chart printed with clogged nozzles).
“When I’ve seen posterization issues such as this, it has usually been the result of poor or currupt linearization, less than ideal paper, or to a lesser degree, choice of rendering intent.”
Well I can rule out the paper (Epson Premium Luster) and the render intent because it happens regardless of the render intent.
“Posterization has cropped up on our Epson 7600/9600/4000 printers before, but it was either the result of not enough ink limiting during profile creation or a currupt linearization chart being measured (the result of a linearization chart printed with clogged nozzles).”
Well it has happened on my 960 since day one through two different profile manufacturers and several attempts to print profiles so I am not sure about the clogged print head being the culprit. I always do a thorough cleaning before I print the target. I also find it odd that the profile built for the epson R800 that is posted on Epson’s site is doing the same thing. I am sure they cleaned the print heads before they printed a target. This being the case is there any way to narrow down the problem? How are you limiting the ink during profile creation? Through the print driver?
Since you’re seeing it in the profiles provided by Epson, and I can confirm that, I don’t know what to tell you. Seems like if there was an easy way to make it work, Epson would have worked out the posterization of there own profiles.
Perhaps it’s just an inherent charateristic of that printer/ink/paper/driver combination. I do notice that the R800 gamut shape looks a little twisted as L* approaches zero, so maybe its not so easy to get a linear response in the darker areas? I’m assuming you have tried printing/profiling at the highest resolution/quality settings for the printer…have you tried, or are you using a Rip?
I’m thinking maybe gamut mapping plays a role here as well. For the profiles you outsourced, do you know what software package created the profile? I just created a profile of an Epson 4000 with both ProfileMaker 5.0.4 and Monaco Profiler 4.7.2 using the same measurement data. The Monaco profile has a much, much smoother gamut shape in the shadow area. Although posterization isn’t too much of a problem, I can see the beginings of this issue in certain color transitions in the PM5 profile where I don’t see it in the Monaco profile.
I agree unless they just weren’t aware of it? I am thinking it might be a driver issue because the R1800 does not exhibit this problem as and it has the same print mechanism as the R800.
I have not, I am printing at 1440 dpi because I do not have a need for anything higher. I would also assume that printing at a higher DPI would only worsen the problem due to even more ink being laid down. Might be something to look into.
I am not sure since I did outsource them. Maybe you would know better than I. One was from Dry Creek Photo and the other from Chromix.
they use a “combination of GretagMacbeth’s ProfileMaker Pro and our own spectral processing routines”
I believe ProfileMaker Pro as well…right guys?
I should point out that I don’t think ProfileMaker profiles are inferior (well, ok I do in some cases , but not based soley on this particular issue), just that the Monaco profiles give a smoother gamut surface, perhaps at the expense of a slightly smaller volume.
I just got off the phone with Ethan over at Dry Creek and as usual he was very helpful and informed me that since I had originally had my profile made (04) that they have worked on this particular problem and found a solution. He was able to build a new profile for me that fixed the problem. He also confirmed it is a driver issue with the lower end epsons.
I can say is this is the best I have ever seen coming from the 960. I will also say it rivals the stuff I was getting from my 2200 with imageprint without the bronzing. It is safe to say they have worked out the issues with the epson printers that even epson has not worked out given the profile for the R800 they have posted on there site. And yes, per Ethan, it is a new algorithm they are using.
I too have had my time with the same posterization happening with my R800 and finally was able to get a reasonable custom profile made with ProfileMaker 5 set to “Logo Colorful”. The resulting profile is only usable with Perceptual rendering intent. This profile works for 95% of images and for the remaining ones that still exhibit posterization, I have to do localized contrast adjustment using USM or something like that. At least the problem is minimized to the point that most of the time it is not an issue.
As to the R1800 profiles, apparently these are the NEW, IMPROVED R800 profiles as well (according to several threads in other forums). I downloaded these new profiles for my R800 hoping for something even better than my custom profile, but alas, they exhibited a bit more posterization. They are a definate improvement over the old R800 profiles though.
As to paper stock making a difference: Go to moabpaper.com/index.asp and download a profile or two of theirs. Unfortunately, there are no profiles for the Epson dye-based printers, BUT their profiles for the R800/R1800 exhibit no posterization. I have read of several reports where some photographers have quit using Epson paper and switched to MOAB, due to better paper/profile performance. Again, not a solution for you, Larry, but just thought I’d throw this out as another scenario. I haven’t tested this yet, but have just received my first pack of Kokopelli Satin and will find out first-hand in the next few days. My problem is that I have to import this stuff at a higher cost as I live in Canada, so the performance increase would have to be totally remarkable to justify bringing it in on a continuous basis.
Amazingly, MOAB creates their profiles using Colorvision’s SpectroPRO2 Combo (at $809 from Chromix). MOAB’s staff told me that they are using a target with 730 patches. The only downside with this package is that you aren’t supposed to be creating profiles for resale (according to Colorvison) and part of the software (Doctor PRO) only works as a plug-in to Photoshop, making this combo more-or-less an internal solution and not a full-blown “make profiles for anyone” solution.
So why can’t vendors with Spectrolino/Spectroscans sporting polarizing filters come up with good working profiles for these printers if MOAB can do it with an $800 solution??? Surely the difference isn’t all in the paper, is it?
So, this is a driver issue, but an improved algorithm during profile creation improves it…sounds like the profile is at least in part responsible. Does your new profile have a much smoother gamut surface than the previous one? And what is it about the driver that gives rise to this issue…the CMM?
I wouldn’t think the measuring device would be at fault,…but maybe the software crunching the numbers.
My new profile (the one made with Logo Colorful setting) does render the image much as I see the image before I softproof it - that is without posterization. As to your second question, I’ve read that these desktop printers are exceedingly non-linear and I suspect that the way people have made profiles up until now is not quite sufficient to accomodate for this non-linearity (my guess, anyway). Now I must qualify myself in that I haven’t had the pleasure of creating any profiles yet and have relied on others to make them, but I am getting damned interested in doing so!
I’m afraid to buy something like a GM Eye-One Photo or an X-Rite Pulse just in case these devices don’t cut it in certain situations, although your last comment alluded to the fact that the software may have as much or more to do with making good profiles than the equipment. One thing about measurment equipment and the Gloss Optimizer used in the R800/1800: From all my reading, it appears that the targets should be printed with G.O. off, UNLESS you have a spectrophotometer fitted with a polarizer filter (Spectrolino). Otherwise the G.O. does not allow “lesser” spectros to read with predictability/repeatability.
If the above is true, what about the new K3 inks that have some form of Gloss Optimizer incorporated into most of the cartridges? Is this going to mess up non-polarized spectrophotometers too? I just about have to buy either one of the above to find out, but what can I say? What I really want, to cover me in every situation, is a Spectrolino/Spectroscan with both GM and Monaco software so I can pick and choose. It’s a slow machine, but hey, it is the ultimate in versatility (and unfortunately, price)!
, but not based soley on this particular issue), just that the Monaco profiles give a smoother gamut surface, perhaps at the expense of a slightly smaller volume.