Pages: [1]   Go Down

Author Topic: Inaccurate Profiling Results  (Read 11294 times)

artlane

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Inaccurate Profiling Results
« on: December 17, 2006, 09:03:06 pm »

I'm at my wits end.

Operating system: Win2K, 1GB RAM

Photoshop CS

Calibrated NEC2090UXi

Epson 7600 with Ultrachrome inks

Image mode: AdobeRGB

VIEWING

Soft-proof settings:
Profile: 9600 EnhMatte EM1 Std v2.icc (Atkinson for enhanced matte)
Relative Colormetric, Black Point Compensation
Proof colors on

At this point, the image displays just fine and looks great.

PRINTING

Print with Preview settings:

Color Management
Source space: Document Adobe RGB
Print space: 9600 EnhMatte EM1 Std v2.icc
Relative colormetric with black point compensation

Printer driver:

No Color Adjustment
Enhanced Matte Paper
High Speed Disabled
Super X ON

All images are printing overly dark/contrasty and oversaturated to the red.
They all display fine with soft-proof.
They all print just fine if the standard Epson drivers are enabled.
The problem arises only if I am trying to use ANY of Atkinson's profiles or any profiles that I have created.

Double-proofing looks like the culprit -- but I've ensured that the settings are correct -- I think. Perhpas I am overlooking something?

I thought I'd ask before reloading Photoshop from the ground up!!

Thank you very much.

Art Lane
http://susanlane-fineart.com
Logged

David White

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 272
    • http://
Inaccurate Profiling Results
« Reply #1 on: December 17, 2006, 09:59:27 pm »

Seems like this was just covered in the Printer, Papers and Inks section in the following thread:   Shadow tonality problem with Epson K3 and ICC
Logged
David White

artlane

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Inaccurate Profiling Results
« Reply #2 on: December 17, 2006, 10:47:47 pm »

Actually, no. Similar though.

In my case I am printing at 1440, as specified by Atkinson.

Yes, I am using the 7600-9600 profiles (not the 7800-9800 ones).

Any other suggestions would be appreciated.

Thanks,

Art

Quote
Seems like this was just covered in the Printer, Papers and Inks section in the following thread:  Shadow tonality problem with Epson K3 and ICC
[a href=\"index.php?act=findpost&pid=91074\"][{POST_SNAPBACK}][/a]
Logged

Paul Sumi

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1217
Inaccurate Profiling Results
« Reply #3 on: December 18, 2006, 12:58:50 am »

This is a dumb question to ask, but is Adobe's Gamma Loader disabled in your startup?  The effects SHOULD be available on-screen if it is, but it's still something to check.

Paul
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Inaccurate Profiling Results
« Reply #4 on: December 18, 2006, 08:14:06 am »

The 7600/9600 are not linearized to a common standard at the factory, and therefore are subject to more unit variation than the newer models. I tried the Atkinson profiles, and they worked pretty good as long as I had the ink usage to something like -25% in the advanced driver settings. Using the defaults gave me similar results to what you describe. But I eventually gave up and bought an Eye-One Photo spectrophotometer kit and made my own profiles.
Logged

artlane

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Inaccurate Profiling Results
« Reply #5 on: December 18, 2006, 09:10:23 am »

Hi Paul,

I triple-checked under my control panel (add/remove programs) and there is no entry for Adobe Gamma. I removed it a while ago.

Thanks,

Art

Quote
This is a dumb question to ask, but is Adobe's Gamma Loader disabled in your startup?  The effects SHOULD be available on-screen if it is, but it's still something to check.

Paul
[a href=\"index.php?act=findpost&pid=91096\"][{POST_SNAPBACK}][/a]
Logged

artlane

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Inaccurate Profiling Results
« Reply #6 on: December 18, 2006, 09:15:35 am »

I understand. However, I've rolled my own profiles using Colorvision's FixPrintPro Colorimeter, and have discovered the results are very similar to what I am obtaining with the Atkinson profiles. It almost seems as if something is stuck.

I should clarify that when soft-proofing the profiles, they look great on the screen, but if I select Print Preview in the driver, the preview matches what is going to be printed and it looks pretty ugly.

Seems I've tried about everything.

Thanks... Art

Quote
The 7600/9600 are not linearized to a common standard at the factory, and therefore are subject to more unit variation than the newer models. I tried the Atkinson profiles, and they worked pretty good as long as I had the ink usage to something like -25% in the advanced driver settings. Using the defaults gave me similar results to what you describe. But I eventually gave up and bought an Eye-One Photo spectrophotometer kit and made my own profiles.
[a href=\"index.php?act=findpost&pid=91146\"][{POST_SNAPBACK}][/a]
Logged

colourperfect

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 75
    • http://www.colourperfect.co.uk
Inaccurate Profiling Results
« Reply #7 on: December 18, 2006, 09:19:25 am »

Jonathan,

I love your generalizations ...... "the 7600/9600 are not linearized to a common standard therefore ......"

Can you quantify this ? My personal experience and that of many others is that they were well matched across production lots. Even when these printers are repaired in the field special firmware is loaded into them temporarily which calibrates them against a common standard. I have watched this process with my own eyes.

Your point relating to ink usage is however a point to bear in mind. If you look at Bill Atkinsons Ultra Smooth Fine Art profile, an Epson paper, then this recommends a -10% setting.

I have found that a setting approaching -18% is needed for some non-Epson papers to minimise pooling. It is however a fine line before too little ink results in effects similar to banding.

I would recommend printing some stepped gradients out to identify the best paper setting to use and the ideal ink setting prior to profiling.

Artlane, is your monitor profiled recently and are you lighting conditions sensible for print / monitor comparisons ?

Ian

http://www.colourperfect.co.uk



Quote
The 7600/9600 are not linearized to a common standard at the factory, and therefore are subject to more unit variation than the newer models. I tried the Atkinson profiles, and they worked pretty good as long as I had the ink usage to something like -25% in the advanced driver settings. Using the defaults gave me similar results to what you describe. But I eventually gave up and bought an Eye-One Photo spectrophotometer kit and made my own profiles.
[a href=\"index.php?act=findpost&pid=91146\"][{POST_SNAPBACK}][/a]
Logged

artlane

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Inaccurate Profiling Results
« Reply #8 on: December 18, 2006, 09:48:44 am »

Hello Ian,

I sincerely appreciate your expert advice here.

Yes, I've calibrated with a Colorvision Spyder2. All other prints, using Epson canned profiles that I assume are activated when selecting paper type in their driver, look to be pretty close, although with fine art reproductions there is always tweaking required.

I've tried reducing the color density setting under paper config from 0 all the way down to -50, and what is curious is that the image that pops up prior to printing (print preview) doesn't change at all.

I am using Epson's latest driver for Win2K, version 5.52. I don't want to get the discussion unfocused, but I noticed Epson has thrown in some new settings, e.g. Advanced Photo, Text/Graph, ICM. It appears that these may be ignored when choosing Off (No Color Adjustment), however.

I also want to statethat I have a very early production Epson7600, but I have kept the embedded firmware up-to-date: BN1452.

Thanks again,

Art



Quote
Jonathan,

I love your generalizations ...... "the 7600/9600 are not linearized to a common standard therefore ......"

Can you quantify this ? My personal experience and that of many others is that they were well matched across production lots. Even when these printers are repaired in the field special firmware is loaded into them temporarily which calibrates them against a common standard. I have watched this process with my own eyes.

Your point relating to ink usage is however a point to bear in mind. If you look at Bill Atkinsons Ultra Smooth Fine Art profile, an Epson paper, then this recommends a -10% setting.

I have found that a setting approaching -18% is needed for some non-Epson papers to minimise pooling. It is however a fine line before too little ink results in effects similar to banding.

I would recommend printing some stepped gradients out to identify the best paper setting to use and the ideal ink setting prior to profiling.

Artlane, is your monitor profiled recently and are you lighting conditions sensible for print / monitor comparisons ?

Ian

http://www.colourperfect.co.uk
[a href=\"index.php?act=findpost&pid=91158\"][{POST_SNAPBACK}][/a]
Logged

Paul Sumi

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1217
Inaccurate Profiling Results
« Reply #9 on: December 18, 2006, 09:53:49 am »

Quote
Hi Paul,

I triple-checked under my control panel (add/remove programs) and there is no entry for Adobe Gamma. I removed it a while ago.

Thanks,

Art
[a href=\"index.php?act=findpost&pid=91156\"][{POST_SNAPBACK}][/a]

Art,

Adobe Gamma does NOT appear in your control panel on add/remove programs.  When one installs PhotoShop I believe Adobe Gamma Loader is automatically installed (I may be mistaken).

You're using Win2K, so I'm not sure, but in XP I "run" msconfig (the system configuration utility).  Click on the startup tab and Adobe Gamma Loader is hopefully not there.  If it is, click off its check box, click okay and reboot.

Paul
« Last Edit: December 18, 2006, 09:54:16 am by PaulS »
Logged

artlane

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Inaccurate Profiling Results
« Reply #10 on: December 18, 2006, 10:26:54 am »

Paul,

AG was not selected in my startup. In fact, I get a window that states that my Spyder 2 monitor profile was successfully loaded into my video card when booting up.

Just to be sure, I scanned my drives and deleted all files related in any way to Adobe Gamma. Next, I deleted my legacy Photoshop 7.0.1 installation as well.

After rebooting, the overly saturated profiles still display in the print preview that I see just before printing.

Someday I'll get to the bottom of this! It's beginning to look like I will need to de-install CS and reload the system.

Thanks,

Art


Quote
Art,

Adobe Gamma does NOT appear in your control panel on add/remove programs.  When one installs PhotoShop I believe Adobe Gamma Loader is automatically installed (I may be mistaken).

You're using Win2K, so I'm not sure, but in XP I "run" msconfig (the system configuration utility).  Click on the startup tab and Adobe Gamma Loader is hopefully not there.  If it is, click off its check box, click okay and reboot.

Paul
[a href=\"index.php?act=findpost&pid=91174\"][{POST_SNAPBACK}][/a]
« Last Edit: December 18, 2006, 10:49:16 am by artlane »
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Inaccurate Profiling Results
« Reply #11 on: December 18, 2006, 10:46:08 am »

Quote
Jonathan,

I love your generalizations ...... "the 7600/9600 are not linearized to a common standard therefore ......"

Can you quantify this ? My personal experience and that of many others is that they were well matched across production lots. Even when these printers are repaired in the field special firmware is loaded into them temporarily which calibrates them against a common standard. I have watched this process with my own eyes.

Your point relating to ink usage is however a point to bear in mind. If you look at Bill Atkinsons Ultra Smooth Fine Art profile, an Epson paper, then this recommends a -10% setting.

The profile I used was for Enhanced Matte, and had no recommendation regarding altering the ink usage setting in any of the documentation that came with it. So as far as I know, the need for altering the ink usage setting was primarily due to my 7600 laying down more ink than the one Atkinson used to make his profile. As to you witnessing a tech loading something in firmware, that calibrates nothing, unless what was loaded into the firmware was spectrophotometer measurements from printed patch charts from that printer.

In contrast, the newer Epson models have the measurements from an actual spectro-measured ink linearizarion patch chart loaded into firmware, to make each unit's print characteristics identical within the measurement tolerances of the spectrophotometer doing the measuring, rather than the manufacturing tolerances of the print head. Epson's own marketing was rather clear about at this when they started that practice.
Logged

colourperfect

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 75
    • http://www.colourperfect.co.uk
Inaccurate Profiling Results
« Reply #12 on: December 18, 2006, 04:51:09 pm »

This would suggest that the tech will need to carry round a spectro in order to recalibrate when they do a major repair. Either that or send of the patches for measurements and wait a week or so.

I am not sure either is probable, will need to ask my contacts at Epson.

Its entirely possible that the firware loaded in contains head parameter's, measured a manufacture that are used in the f/w. The variables that cause colours to differ are head nozzle size and ink drop size ( a function of the piezo chamber volume)

I am aware that the firmware can be incorrectly loaded into 7600's. Mine had a problem with the f/w to drive the USB port. It resulted in the PC not recognizing it. So other f/w mismatches are possible

Ian

http://www.colourperfect.co.uk


Quote
The profile I used was for Enhanced Matte, and had no recommendation regarding altering the ink usage setting in any of the documentation that came with it. So as far as I know, the need for altering the ink usage setting was primarily due to my 7600 laying down more ink than the one Atkinson used to make his profile. As to you witnessing a tech loading something in firmware, that calibrates nothing, unless what was loaded into the firmware was spectrophotometer measurements from printed patch charts from that printer.

In contrast, the newer Epson models have the measurements from an actual spectro-measured ink linearizarion patch chart loaded into firmware, to make each unit's print characteristics identical within the measurement tolerances of the spectrophotometer doing the measuring, rather than the manufacturing tolerances of the print head. Epson's own marketing was rather clear about at this when they started that practice.
[a href=\"index.php?act=findpost&pid=91183\"][{POST_SNAPBACK}][/a]
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Inaccurate Profiling Results
« Reply #13 on: December 18, 2006, 06:05:42 pm »

Quote
This would suggest that the tech will need to carry round a spectro in order to recalibrate when they do a major repair. Either that or send of the patches for measurements and wait a week or so.

I'm skeptical that the data is based off spectro readings in the case of the 7600/9600 and earlier models. Epson didn't start PRing their standardized linearization process until after the models replacing the 7600/9600 were announced.

If you buy into the notion that the printers are calibrated to minimize unit-to-unit variation, the only way to do that accurately is linearize the printer to a standard state by printing some color patches and measuring the prints with a spectro. And the only way to do that validly is to test the printer as a complete unit. Putting a print head in a test bench printer, printing a patch chart, measuring the chart at the factory, and installing the measurement data with the print head at the remote site isn't going to keep the linearization measurement data valid unless the voltages driving the print head, ink feed pressure, etc. are extremely consistent between the test bench printer and the user's printer.

I don't doubt that some kind of data about the print head is loaded into the printer when the print head is changed or other major repairs are done, I'm just skeptical that doing so is going to ensure unit-to-unit uniformity unless the data are spectro measurements from prints made with the particular printer after the repair is made.
Logged

artlane

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Inaccurate Profiling Results
« Reply #14 on: December 19, 2006, 05:13:27 pm »

To Everyone,

I think I've discovered my problem. Because many of you will probably disagree with the solution, I didn't invent it.

Quoting from "Photoshop 6 Artistry, Mastering the Digital Image" by Barry Haynes and Wendy Crumpler, pgs. 146-147, "Remember that when you print an image using a custom profile, you first want to use Image/Mode/Convert to Profile to convert your master image from your LAB or RGB color space into the space for your printer created by this profile. Then in your print dialog, you want to turn all color management off... "

So while I was shutting down color management, I was not converting the image to use the printer profile, instead trying to use an image with the RGB profile assigned.

I tried this, and obtained a print that was much closer to my original RGB image, but cannot obtain a satisfactory softproof if I assign the printer profile to View. Although I haven't experimented enough to solve this viewing problem, I suspect that if I use Adobe RGB or Working RGB to softproof and convert the image to the printer profile, the problem will be resolved.

As a side note, the print still isn't as true as the canned Epson paper profiles, and probably requires further adjustment.

Thanks to everyone who tried to help me with this!

Merry Christmas,

Art Lane

Quote
Actually, no. Similar though.

In my case I am printing at 1440, as specified by Atkinson.

Yes, I am using the 7600-9600 profiles (not the 7800-9800 ones).

Any other suggestions would be appreciated.

Thanks,

Art
[a href=\"index.php?act=findpost&pid=91081\"][{POST_SNAPBACK}][/a]
Logged

Greg_E

  • Full Member
  • ***
  • Offline Offline
  • Posts: 113
Inaccurate Profiling Results
« Reply #15 on: December 19, 2006, 08:18:57 pm »

I really shouldn't get into this, but I'm bored so why not.

The 76/9600 are not calibrated to any specific standard. They are of much higher manufacturing tolerances than the desktop counterparts, which reduces the unit to unit differences.

The chamber volume has little to do with droplet size as long as the chamber is large enough to contain the maximum volume of ink. I can tell you for a fact that when you use the regular Epson driver, you are not even close to using the maximum amount of ink that the head can fire. The droplet size is determined by the voltage applied to the element, and the frequency of the signal used to fire the drop. There are some technical papers about this on the service/repair CDs that only Epson techs can get.

The only variability that can be changed to the replacement heads is in the ID string that is entered into the printer after the head has been changed. It tells the CPU what the data address for that head is, and how much drive voltage should be used. All Epson heads are tested at the factory to make sure that they are working, and to determine the proper voltages to produce the required drop sizes.

Now if you don't believe me about the amount of ink that an epson printer can spray, go download the demo of the Evolution RIP, learn to use it, set it for 100 % reduction (zero reduction depending on your point of view), no ink limits, and no clipping in the linearization. Print. Now watch the ink run down the paper and make a huge mess on the floor. Also watch your ink counter fly down to zero. I suggest using a matte paper to prevent the mess on the floor that will almost certainly happen with glossy. Hahnemuhle German Etching will almost take this load, but get out your spectro and measure the densities and you'll see pretty quickly that you are over inking by a large amount. Now set the reduction to somewhere between 50 and 60 percent, try again and see how much closer you will be to the amount the Epson driver puts down.
Logged

Nill Toulme

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 738
    • http://www.toulmephoto.com
Inaccurate Profiling Results
« Reply #16 on: December 19, 2006, 10:45:38 pm »

That's interesting, but I'm more interested in Art's quoting Haynes and Crumpler to the effect that the image should be converted to the printer profile.  From what little I know (which is little indeed), that sounds dead wrong to me, and I'm surprised that they would say such a thing.

Nill
~~
www.toulme.net
Logged

Greg_E

  • Full Member
  • ***
  • Offline Offline
  • Posts: 113
Inaccurate Profiling Results
« Reply #17 on: December 19, 2006, 11:30:48 pm »

Quote
That's interesting, but I'm more interested in Art's quoting Haynes and Crumpler to the effect that the image should be converted to the printer profile.  From what little I know (which is little indeed), that sounds dead wrong to me, and I'm surprised that they would say such a thing.

Nill
~~
www.toulme.net
[a href=\"index.php?act=findpost&pid=91513\"][{POST_SNAPBACK}][/a]


That may be left over from PS 5 since it wasn't the most color managment sophisticated.

Convert to profile with all colormanagement off should be the same as setting source space, and destination space as they are really doing the same thing. Both are converting to the destination profile and sending them to the printer just using slightly different ways of getting there.
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Inaccurate Profiling Results
« Reply #18 on: December 22, 2006, 11:47:57 am »

Quote
Quoting from "Photoshop 6 Artistry, Mastering the Digital Image" by Barry Haynes and Wendy Crumpler, pgs. 146-147, "Remember that when you print an image using a custom profile, you first want to use Image/Mode/Convert to Profile to convert your master image from your LAB or RGB color space into the space for your printer created by this profile. Then in your print dialog, you want to turn all color management off... "

So while I was shutting down color management, I was not converting the image to use the printer profile, instead trying to use an image with the RGB profile assigned.

You always want to run the print driver with color management/adjustment/correction turned off, but there's generally no good reason to convert an image to an output profile (I'll cover the exceptions to this later). You simply need to select the correct output profile in Photoshop's print dialog and the correct paper type in the print driver settings and you're good to go if the profiles are good.

The only time you should convert an image to a output device profile is when submitting a file to a print service that requires you to do so, or printing something that has a lot of colors that are on the edge of what the output device can print, and you need to tweak carefullt to maximize available gamut. But you should only convert a copy of the image to an output device profile; your master copy of the image should always be an RGB editing space like ProPhoto or Adobe RGB.
Logged
Pages: [1]   Go Up