Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Shadow tonality problem with Epson K3 and ICC  (Read 4053 times)
nihil
Jr. Member
**
Offline Offline

Posts: 52



WWW
« on: December 15, 2006, 12:02:46 PM »
ReplyReply

Hello! I am having a problem with the shadow transitions when using ICC profiles. I can't get it to print black to dark grey gradations smooth. (see the attachment)

I just replaced my Epson 4800 with a 7800. With the 4800 I had the same problem, so I decided to let "printer determine colors", and use Color Controls. The paper I'm using is the Innova F-Type Gloss. So even if the colors were not 100% predictable by color-pro-standards, I managed ok using the semigloss setting, and was satisfied as long as the shadows were excellent. (crucial for my images, which consist of large areas of black/shadow detail).

Now with the 7800 I find that Color Controls does not give me the same quality any longer, for colors that is. It won't bring out the gamut potential at all. But the nice shadow tonality is still as it used to.

I've come to understand I really need to get a grip on ICC-profiles. I've tried a lot of them, got a custom made for the 4800. But the shadows doesn't look nice at all. I just tried Innova's own 7800 profile, and it is the best I've seen as for colors, and has the smoothest shadows I've seen in a ICC profile, but still awful.

See the attached image for an illustration. This is a gradient from about RGB=0,0,0 to about 100,100,100. This is a very very overexposed image to demonstrate this problem only. It looks as if the lower images has lighter blacks, but it is only because the flash was closer to that side.

I'm printing from PS CS2. Tried Relative Colorimetric, Perceptual and so on.

Please help me out here  
Logged

regards,
Erlend Mørk
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #1 on: December 16, 2006, 05:10:15 AM »
ReplyReply

Post your Photoshop color management settings, your print driver color settings, as well as the color management settings in the PS print dialog you used to print. You're probably double profiling--applying the same profile in Photoshop and in the print driver, which is guaranteed to make your prints look like feces.
Logged

nihil
Jr. Member
**
Offline Offline

Posts: 52



WWW
« Reply #2 on: December 16, 2006, 08:57:57 AM »
ReplyReply

Ok. Using the ICC profile from Innovaart.com, which is the one I want to make work:

Let Photoshop determine colors
Printer profile: The correct one, yes.
Relative Colorimetric (tried Perceptual as well though)
Black Point Compensation ON

In the printer driver:
No color management
Paper: Tried Premium Glossy, Semigloss and Luster (same problem)
SuperPhoto 2880 dpi
High Speed OFF
In paper config: Paper thickness 4 (0,1mm)

And by the way. The prints don't look like feces, judging from the test images from Bill Atkinson's ICC website. I'm sure many would be really pleased. Only the gradients reveal the problem which also show in my own kind of shadowy images.

Let me know if there are any other settings you need to know. Thanks for your time :-)
Logged

regards,
Erlend Mørk
Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #3 on: December 16, 2006, 09:30:33 AM »
ReplyReply

The magenta in the gray gradients shows you have a bad profile, or a clogged print nozzle, or possibly a bad gradient to begin with.  

The banding in the gradient could be a number of things, including a nozzle clog, but most likely it has to do with how you made th egradient.  Did you use a 0 - 255 drag with the gradient tool?  8-bit or 16-bit?  RGB or LAB?  Did you save the file as a tiff or psd, or as a jpeg?

You did not mention it, but did you turn on any of the other options like finest detail or edge smoothing?

Did you try 1440 in addition to 2880?

Not sure why you are adjusting paper thickness beyond standard?
« Last Edit: December 16, 2006, 09:39:33 AM by Jack Flesher » Logged

nihil
Jr. Member
**
Offline Offline

Posts: 52



WWW
« Reply #4 on: December 16, 2006, 10:09:10 AM »
ReplyReply

I always perform a nozzle check before printing. The color variations looks very exaggerated on the overexposed photograph. The lowest gradient from the "color controls only" image seem pretty neutral to my eyes. The middle one, using ICC profile from Innova has slight variations, just not that bad.

I used a gradient found at Bill Atkinsons site. But I converted from LAB to RGB. 16-bit. Saved as PSD-file. 16bit. This is only a crop of the darkest part of it, although it looks like it includes the bright part as well. I've checked of course, and it is alright, neutral and smooth.

I've also made my own gradients in a new document just to check. Same thing.

I left all other options at default, Finest detail and edge smoothing OFF. I adjusted the paper thickness as I found 4 to be best with F-type Gloss, when I performed a paper thickness test with the Epson 4800.

I've not tried 1440. I found it had a very negative effect on the colors and resolution with the 4800. So I went for 2880.
Logged

regards,
Erlend Mørk
Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #5 on: December 16, 2006, 11:11:45 AM »
ReplyReply

Quote
I used a gradient found at Bill Atkinsons site. But I converted from LAB to RGB. 16-bit. Saved as PSD-file. 16bit.

Okay, Bill's LAB gradient is excellent as-is.  Try keeping it in LAB and re-printing it straight from LAB, using your best profile.  Print it at 1440 and 2880 and get back to us with the results.  Use RC with BPC in the print dialog.  

Also, a stupid point probably, but do an automatic head alignment using glossy paper before printing the gradients this time.
« Last Edit: December 16, 2006, 11:20:14 AM by Jack Flesher » Logged

Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #6 on: December 16, 2006, 11:50:21 AM »
ReplyReply

Quote
In the printer driver:
No color management
Paper: Tried Premium Glossy, Semigloss and Luster (same problem)
SuperPhoto 2880 dpi
High Speed OFF
In paper config: Paper thickness 4 (0,1mm)

You can only use a profile for one combination of printer, paper, and driver settings. A profile is only valid if the printer, paper, and all print driver settings (paper type and print quality especially) are exactly the same as those used when creating the profile. If you change anything, you will invalidate the profile and have problems with gradients and colors. You cannot use profiles made for one printer model with another printer model and expect good results. The only exception to this is where the print head and engine are identical, but one model has a wider carriage than the other, like the Epson 7600 and 9600. Find out what paper type and print quality settings were set when the profile was made (the instructions for the profile should tell you) and duplicate them exactly. If you're still having problems, the profile is bad.

You may simply need to bite the bullet and get some custom profiles made for your printer. Printers vary from unit to unit, and a profile made for one will not necessarily work well on another.
Logged

nihil
Jr. Member
**
Offline Offline

Posts: 52



WWW
« Reply #7 on: December 16, 2006, 12:34:06 PM »
ReplyReply

Jack: I did an automatic head alignment. It printed some patterns, but I don't know if there should be a confirmation of success or anything.

BUT, 1440 dpi cured it! I can't thank you enough! I didn't realize that the dpi setting could have such a huge impact on that kind of thing. The shadows are now smooth like they should be, and no color casts. However, now it is more grainy, and the colors seem to lose a bit because of that. And I have all the time in the world to wait for a higher quality, so I wish I had a profile for 2880 dpi.

I was actually lucky enough to get a Eye-One Photo package with the printer purchase, but they didn't have enough units to supply me just yet. It seems to take some time. I hope I will be able to make good profiles with that equipment. I don't suppose anyone got a 7800 profile for F-type Gloss 2880dpi around for sharing, until I can make my own custom ones?
Logged

regards,
Erlend Mørk
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #8 on: December 16, 2006, 06:19:01 PM »
ReplyReply

I have the Eye-One Photo bundle (Eye-One spectrophotometer, mini Color Checker, and scanner calibration chart, but not Color CHecker SG) and used properly, it will make excellent printer profiles. Use the charts with the most patches (900 or more) when profiling, and remember to let them dry for at least 24 hours before measuring them. Also, stack several blank sheets of paper stock under the chart you're measuring, and never measure charts on a colored surface like a wooden table. The colored surface will show through the paper otherwise, and screw up your measurements, resulting in a bad profile.
Logged

Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #9 on: December 16, 2006, 06:56:36 PM »
ReplyReply

Quote
Jack: I did an automatic head alignment. It printed some patterns, but I don't know if there should be a confirmation of success or anything.

BUT, 1440 dpi cured it! I can't thank you enough! I didn't realize that the dpi setting could have such a huge impact on that kind of thing. The shadows are now smooth like they should be, and no color casts. However, now it is more grainy, and the colors seem to lose a bit because of that. And I have all the time in the world to wait for a higher quality, so I wish I had a profile for 2880 dpi.

I was actually lucky enough to get a Eye-One Photo package with the printer purchase, but they didn't have enough units to supply me just yet. It seems to take some time. I hope I will be able to make good profiles with that equipment. I don't suppose anyone got a 7800 profile for F-type Gloss 2880dpi around for sharing, until I can make my own custom ones?
[a href=\"index.php?act=findpost&pid=90853\"][{POST_SNAPBACK}][/a]

My pleasure.  Neither the auto head cleaning or alignment routines do a hard confirmation -- they simply stop after the printer sees it's got it right and tell you it's done.   I am pretty sure you will get an error message if it can not sucessfully complete either for whatever reason, so no news is good news  

You might want to at least try 2880 now that you have re-aligned the heads as it may have improved output there as well.  (FWIW, and I say this cautiously as it goes against convention, but I have found the 7800 and 3800 printers so well zeroed that I can use the profiles I generated at 1440 for 2880 output and get excellent results with no visible profile issues.)

Cheers,
« Last Edit: December 16, 2006, 07:01:25 PM by Jack Flesher » Logged

nihil
Jr. Member
**
Offline Offline

Posts: 52



WWW
« Reply #10 on: December 17, 2006, 09:00:59 AM »
ReplyReply

Jonathan: Thank you for your kind advice.  

Jack: I did the head alignment before trying both 2880 and 1440. So no 2880dpi with this profile I'm afraid.
« Last Edit: December 17, 2006, 09:01:20 AM by nihil » Logged

regards,
Erlend Mørk
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #11 on: December 17, 2006, 11:19:37 AM »
ReplyReply

Quote
Jonathan: Thank you for your kind advice.

Thank you. Just as a clarification on my earlier comments, when measuring a patch chart, put several blank sheets of paper of the same type as the patch chart under the patch chart, to prevent color bleedthrough. So if you're measuring a chart printed on Moab Entrada, have enough sheets of blank Moab Entrada underneath the chart to prevent any bleedthrough of the color of the surface underneath.
Logged

Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #12 on: December 17, 2006, 11:47:54 AM »
ReplyReply

Quote
Thank you. Just as a clarification on my earlier comments, when measuring a patch chart, put several blank sheets of paper of the same type as the patch chart under the patch chart, to prevent color bleedthrough. So if you're measuring a chart printed on Moab Entrada, have enough sheets of blank Moab Entrada underneath the chart to prevent any bleedthrough of the color of the surface underneath.
[a href=\"index.php?act=findpost&pid=90985\"][{POST_SNAPBACK}][/a]

Another way to do this is simply place a gray sheet of matte board underneath the paper before reading it -- this is what I do.
Logged

Stephen Best
Guest
« Reply #13 on: December 17, 2006, 12:56:24 PM »
ReplyReply

Quote
Another way to do this is simply place a gray sheet of matte board underneath the paper before reading it -- this is what I do.
[a href=\"index.php?act=findpost&pid=90989\"][{POST_SNAPBACK}][/a]

I don't think this is a good idea. The intention of the backing is to set the measured paper response to something similar to how it will be viewed, eg. framed with foamcore backing. The same reason D50 is used. The ruler that comes with the Eye-One Pro Spectro (Rev. B and later) is fine for this purpose and available as an upgrade for Rev. A users. It also makes readings pretty easy.

I understand though that ColorBase (if you use it) requires black backing. They probably chose this for consistency. Given that your measurements will be compared against the reference, also presumably done with black backing, this should probably be followed.
Logged
colourperfect
Jr. Member
**
Offline Offline

Posts: 75


WWW
« Reply #14 on: December 18, 2006, 08:29:02 AM »
ReplyReply

Or you can use the Gretag improved strip reading kit, which is available as an upgrade for older i1pro's.

I think it comes as standard now

Ian

http://www.colourperfect.co.uk
Logged
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad