Hi Nick,
Just curious.
I saw your post and the screen snaps, and see that you have an Epson 9880
I also see that the profile name for the paper is Canvas.  Just Canvas.icm
I don't have a 9880 driver haandy, but I do have a 7900/9900, and a 7800/9800 and the profiles for Canvas paper (Epson) are clearly defined as Epson 9900 Canvas_MK and another for Canvas 9900_PK. Signifying that they provide profiles for their canvas printing media for Photo Black and Matte black.
So that seems to beg the questions, Are you using Epson canvas media?   Have you chosen the matching profile for the inkset you are using?

I never used Canvas, so I have no experience.
Just seemed to be a couple of relevant questions to sort out the diagnosis.

Here's some more questions for you.
I just downloaded the 9880 profiles for Canvas... Satin and Prem Canvas Matte.

Both instruction sets say to select WATERCOLOR Paper, Radiant White!!

Each type of profile (one for MK ink and the other for PK ink) seems to have two ICM profiles, one at 720 ppi and the other at 1440.

So where are you regarding these settings?

One other thought.
Qimage remembers the driver settings whereas PS does not. So a mis selected media would be remembered when using Qimage and not remembered using PS.
Fred





Fred

Nick,

I've already proven that the problem is not Qimage by directly comparing the output of Qimage versus PhotoShop (CS5).  You may not believe that, but now I want you to see the output from both programs!  Take a look at the attached.  It is using the image you emailed to me and using your septembercanvas profile.  This is the raw data that is going to the driver with the septembercanvas profile set as the printer profile.  Both Qimage and CS5 are sending identical data to the printer driver.  As you can see, the posterization is caused by your profile.  I tried Perceptual Intent with BPC on and off and RC intent with BPC on and off.  Results of all those tests are identical and have the posterization, except for RC intent with BPC on which was actually better in Qimage.  With RC/BPC, CS5 produced the same posterization but Qimage actually got rid of the posterization under relative colorimetric w/BPC.  The only reason that RC/BPC was different in Qimage is that your septembercanvas profile has a discrepancy in it where the black point is not mapped properly.  I see a lot of that in this business unfortunately.  My suggestion would be to get a better program for creating profiles, or see if the manufacturer of your canvas paper has better profiles.  The one you are using has some internal conflicts and some very jaggy LUTs that cause that posterization.

Now, as to why you see a different result in Qimage and PhotoShop: we're down to workflow or program/driver settings.  To be sure there are no "stuck" settings in Qimage, I would suggest doing "Help", "Reset all program settings" in Qimage.  After that, as I told you in a prior email, it is possible that you have a new driver and it is not compatible with settings from the previous driver that was used when printer settings were saved by Qimage.  To correct this, you must recreate the driver settings from scratch.  After doing the "Help", 'Reset printer settings" and when you restart Qimage, set the driver settings manually (without recalling a printer setup or job).  Once you recreate the settings, you can save them for future use.  The problems occur when you load an old printer setup or job that has settings based on an old (and now incompatible) driver.

Regards,
Mike

You don't see the above because you don't see the data going to the printer: you've been seeing it only AFTER the printer/driver has printed it.  I can assure you, that's exactly what the data looks like when it is sent to the driver.  Maybe in the past, you printed with RC intent and you had BPC on.  In your screen shots, you have BPC off now.  Maybe that's the difference.  I can't answer why you had something different before than you have now.  I don't know what you changed in that time.  I can only tell you (and show you) what both programs are doing now.  And both show posterization: the cause is the profile.  The reason it doesn't look that bad on the actual prints is because of the printer's inability to produce that much separation in those gradients.  In other words, when printed, those colors are dulled down due to the color management being turned off in the driver.

Mike

Black point and incongruous LUT's can cause any number of (usually unpredictable) problems.  Results are unpredictable because with errant LUT's (look up tables), you're leaving interpretation up to the error correction built into the color management engine.  I use Profile Prism of course: http://www.ddisoftware.com/prism, but many tools will find profiling canvas quite a challenge.  One of the most important things is to be sure you have the paper type and ink load settings set optimally for the paper you are using.  Improper paper type can cause ink load issues that make profiling nearly impossible, so hopefully the manufacturer of your canvas has stated which paper type and settings you should use for your particular printer.  If not, you have to do a lot of trial and error testing yourself to determine which gives you the best range and coverage.  I've seen abrupt transitions like the ones in your profile before and the cause is usually mottling of the ink as a result of oversaturation of the printed target in some areas.  This reduces the printer's color range and causes hue shifts as the color management model tries to shift hue to compensate because it basically ran out of room in the correct hue and had to use a different hue to simulate something close.

Mike

That's not really an issue.  It's an anomaly.  There are two issues that I see:

(1) You don't get consistent results now.  If you use your septembercanvas profile with Perceptual intent with BPC on or off, or RC intent with BPC off, you should get the same results from Qimage and CS5.  If you don't, something is wrong on your end.

(2) The profile you created (septembercanvas, which obviously is not the one you used before), has internal inconsistencies that cause abrupt color shifts in some gradients.

Now, your June 16 canvas profile does not exhibit the abrupt gradient and in fact, looks entirely different (much darker) than your septembercanvas profile.  So you've created three profiles (june, july, and september) that are drastically different.  I really think you need to get a handle on your workflow and make sure you are printing your targets properly, profiling properly, and using the correct driver settings.

Mike

Above looks OK from what I can see, with this clarification on your step 3 for printing the target:

(a) When you say you turn off color management in Qimage, you should be setting "Prtr ICC" to "OFF".

(b) When you say you turn off color correction in the driver, you should be selecting "Custom" mode and then "no color adjustment".

And after step 5 when you print the target, use "File", "Save", click the "P" button and save what you did above as a printer setup.

Then later when it comes time to use the profile, click "File", "Load" and load the setup you saved above.  All settings including the driver and Qimage settings will be recalled.  You only have to drop down "Prtr ICC" and select the profile you created.  Once you do select that profile, click "File", "Save" again and click the "P" button to save a printer setup again.  Now give it a name like "septembercanvas" to go with the profile.  That way you can always load the same settings and you'll be certain they are the same as they were when you created the profile.

Now, in looking at the three profiles you sent, take a look at the attached.  See what I mean by the vast difference?  The difference indicates that in June, you were using an entirely different method to print your targets: one which apparently worked better with your spyder because it doesn't have that big gradient shift.  Then in July and September, you're either not printing the target the same way or you aren't profiling the same way.  Either way, you're doing something drastically different in June than you did in July and September.  Looks to me like the June profile is based on "color controls" or some other settings, not "no color adjustment".  Maybe that's why it worked better.  Some profiling tools can't handle the raw (no color adjustment) print when scanning/making profiles.

Mike

Hi Nick and Mike ,
If you haven't seen this, a very interesting and possibly? related quandry was posted by Ctein on TOP today.

http://theonlinephotographer.typepad.com/the_online_photographer/2011/09/kinda-interesting.html