Help. Banding Pattern printing on Epson 3880

[alexbo]

Please forgive me if the following problem has been brought up on this forum already.
I've tried searching for the solution, but came up empty handed so far.

I'm new to digital negative world, and for the past 5 or 6 days I've been working nonstop trying to build a QTR profile using all 8 inks on Epson 3880.
While I would like to think that I got pretty close to the ideal, there is one problem I was not able to resolve.

I get a pattern of banding strips in the shadows (the thinnest portion of negative).
It is mostly constrained to the 80-100% range.
I use all 4 light inks (LK, LLK, LM and LC) to cover that part of the curve, but even using just a single LK ink still displays the pattern.
The lines are about 1/16 of an inch wide, running perpendicular to the direction of the page.

Here is a screenshot of the portion of the scan that displays the banding:

I've tried adjusting the Gray Curve, tried head cleaning, manual head alignment, but nothing helps.
That said, printing the step tablet using native Epson printer shows no sign of banding.

So, I'm completely out of ideas at this point, and desperately need help from the more experienced members here.

Thank you,
Alex.

 

[jeffreyg]

I'm far from an expert when it comes to digital especially the technical aspects. Is it possible that your printer rollers need to be cleaned? I had a problem at one time with my 2200 and bought cleaning sheets from Epson that did the trick.

HOME PAGE

 

[alexbo]

Thank you for your input Jeff, I will look into it.
But I suspect it might be something else, since the pattern only shows up consistently over the small range of tones.
Plus the printer is brand new.

I'm far from an expert when it comes to digital especially the technical aspects. Is it possible that your printer rollers need to be cleaned? I had a problem at one time with my 2200 and bought cleaning sheets from Epson that did the trick.

HOME PAGE

 

[amphoto]

That the banding doesn't appear with the Epson driver probably means it isn't a cleaning or head alignment problem, or that the media isn't being moved precisely. I've heard only good things about the 3880 when it comes to the banding issue that plagued previous Epson printers (such as the 2100/2200). Given the 1/16" repeating nature of the banding, parallel to the direction that the printhead travels (if I've understood you correctly), my guess would be that the issue stems from the way that one pass of the printhead slightly overlaps the previous one, combined with the inks you're using in that part of the negative, the dithering algorithm and media settings (speed, dpi, platten gap etc.)

Do you get the same results when you change the dither algorithm (adaptive hybrid -> ordered -> fast -> very fast) / resolution (720 -> 1440 -> 2880) / speed (Bi-directional vs Uni-directional)?

Could you attach a larger version of the image showing the banding? Also, if you have a scanner, scan that section of the negative at a high native resolution (i.e. non-interpolated) and have a close look at the way the ink droplets have been laid down on the banding boundaries. Sometimes that can be very informative.

 

[alexbo]

Angus,

I will try these options as soon as I get home from work today.
Yes, the banding is parallel to the direction of the printhead.

I started thinking last night that I should be looking into adjusting the custom profile for the Pictorico, and the platten gap as a part of it. You are confirming my suspicions now.
Once I try it, I'll report back with the results.

I'm curious if anyone ever standardized and published these setting for Pictorico film on 3880 (or 3800), given how often this media is used.

Thank you.

That the banding doesn't appear with the Epson driver probably means it isn't a cleaning or head alignment problem, or that the media isn't being moved precisely. I've heard only good things about the 3880 when it comes to the banding issue that plagued previous Epson printers (such as the 2100/2200). Given the 1/16" repeating nature of the banding, parallel to the direction that the printhead travels (if I've understood you correctly), my guess would be that the issue stems from the way that one pass of the printhead slightly overlaps the previous one, combined with the inks you're using in that part of the negative, the dithering algorithm and media settings (speed, dpi, platten gap etc.)

Do you get the same results when you change the dither algorithm (adaptive hybrid -> ordered -> fast -> very fast) / resolution (720 -> 1440 -> 2880) / speed (Bi-directional vs Uni-directional)?

Could you attach a larger version of the image showing the banding? Also, if you have a scanner, scan that section of the negative at a high native resolution (i.e. non-interpolated) and have a close look at the way the ink droplets have been laid down on the banding boundaries. Sometimes that can be very informative.

 

[Joe Lipka]

How are you creating your digital negatives? Is Adobe InDesign part of your process or are you printing directly from photoshop?

 

[alexbo]

Joe, I'm actually printing directly from QTRGui where I was creating profiles for silver printing.

How are you creating your digital negatives? Is Adobe InDesign part of your process or are you printing directly from photoshop?

 

[alexbo]

Argus,

I have tried everything I could think of, but nothing has changed.

Here is what I did:
1. Ran head cleaning and alignment, twice. Just to be sure
2. Printed the step tablet using ordered, adaptive hybrid, fast, very fast dither.
3. Printed using different speeds uni/bi-directional
4. Printed using all available resolutions 720, 1440 dpi, 1440 super, 2880 dpi
5. Tried adjusting platen gap using all available options.
6. Tried creating custom paper profile and adjusting Thickness Pattern, platen gap and Paper Feed Adjust A (raised it to 0.40%)

No effect whatsoever ...

I have scanned the negative, as you asked, but for some reason the pattern is not visible on the negative scan
Here it is, in case you need to take a look at it
Full view (300dpi)and detailed crop(6400 dpi): [url]http://imgur.com/a/sG07j#0 [/url]

The pattern is much better visible on the scan from a positive print. So I printed it directly on Epson luster paper and scanned that
Again, full view (300dpi) and detail(6400 dpi) [url]http://imgur.com/a/AVvve#0 [/url]


I'm at the loss what to do next .....

That the banding doesn't appear with the Epson driver probably means it isn't a cleaning or head alignment problem, or that the media isn't being moved precisely. I've heard only good things about the 3880 when it comes to the banding issue that plagued previous Epson printers (such as the 2100/2200). Given the 1/16" repeating nature of the banding, parallel to the direction that the printhead travels (if I've understood you correctly), my guess would be that the issue stems from the way that one pass of the printhead slightly overlaps the previous one, combined with the inks you're using in that part of the negative, the dithering algorithm and media settings (speed, dpi, platten gap etc.)

Do you get the same results when you change the dither algorithm (adaptive hybrid -> ordered -> fast -> very fast) / resolution (720 -> 1440 -> 2880) / speed (Bi-directional vs Uni-directional)?

Could you attach a larger version of the image showing the banding? Also, if you have a scanner, scan that section of the negative at a high native resolution (i.e. non-interpolated) and have a close look at the way the ink droplets have been laid down on the banding boundaries. Sometimes that can be very informative.

 

[pschwart]

- Defects in the negative will always be visible with an 8-10x loupe.
- The marks on the print look like they are made by star ("pizza") wheels, not banding. If the marks
are in the direction of the paper feed they are certainly not banding.

 

[jeffreyg]

Alex,

I as I mentioned above the technical aspect of all this is not my thing. For several years I have been using Dan Burkholder's plug-in for digital negatives with PhotoShop and Pictorico OHP Transparency Film on an Epson 2200 with Epson Ultrachrome inks. I do tweak the image from the plug-in transfer but very little. I just use the Epson profile for enhanced matte and get very good negatives for pt/pd. (In my simplified way) Have you tried printing in a "default" way such as I do to see if the same problem occurs? I would think that if the banding does not happen then the problem is with something you included or left out in your programming. If you can get proper results without knocking yourself making a custom profile don't bother and enjoy printing.

HOME PAGE

 

[alexbo]

Philip,
The pattern is visible with a naked eye.
We are probably talking about different things here.
Take a look at this image http://i.imgur.com/KQfx6.jpg, you can clearly see the banding between 95% and 100% marks on the smooth gradient portion of the tablet.

- Defects in the negative will always be visible with an 8-10x loupe.
- The marks on the print look like they are made by star ("pizza") wheels, not banding. If the marks
are in the direction of the paper feed they are certainly not banding.

 

[alexbo]

Jeff,

I'm leaning towards doing what you suggesting, as there seem to be no problem printing through Epson driver.
It's just it's sad that I wasted 50+ hours of my time, used up almost 3 packs of Pictorico and over 50 sheets of Ilford paper, 1/3 of the printer ink, got almost a perfect profile, all of that for nothing ...
I mean it works for other people, there must be a logical explanation why it doesn't work for me.

But if I don't find the answer soon, I might just go with the Epson driver.

Thank you.

Alex,

I as I mentioned above the technical aspect of all this is not my thing. For several years I have been using Dan Burkholder's plug-in for digital negatives with PhotoShop and Pictorico OHP Transparency Film on an Epson 2200 with Epson Ultrachrome inks. I do tweak the image from the plug-in transfer but very little. I just use the Epson profile for enhanced matte and get very good negatives for pt/pd. (In my simplified way) Have you tried printing in a "default" way such as I do to see if the same problem occurs? I would think that if the banding does not happen then the problem is with something you included or left out in your programming. If you can get proper results without knocking yourself making a custom profile don't bother and enjoy printing.

HOME PAGE

 

[pschwart]

Got it. Those steps are very discrete and vary in density. This looks like posterization created by the QTR profile which would explain why you don't see this behavior with the Epson driver.

Philip,
The pattern is visible with a naked eye.
We are probably talking about different things here.
Take a look at this image http://i.imgur.com/KQfx6.jpg, you can clearly see the banding between 95% and 100% marks on the smooth gradient portion of the tablet.

 

[pschwart]

Yes, QTR can be a rabbit hole. You might try finding a working profile on the web and tweaking for your environment. You might already be doing this, but consider one curve for light inks and one for dark inks instead of creating a separate curve for each color. Some colors don't contribute much UV blocking so not using these channels can simplify your profile. Ron Reeder has documented a pretty simple workflow for creating digital negatives with QTR. You can also use the Epson driver Advanced B&W with a yellow boost to create negatives. You will still need to create a correction curve, but you will bypass all the profiling required by QTR. PDN and other methods of colorizing work well on the 38xx, too.

Jeff,

I'm leaning towards doing what you suggesting, as there seem to be no problem printing through Epson driver.
It's just it's sad that I wasted 50+ hours of my time, used up almost 3 packs of Pictorico and over 50 sheets of Ilford paper, 1/3 of the printer ink, got almost a perfect profile, all of that for nothing ...
I mean it works for other people, there must be a logical explanation why it doesn't work for me.

But if I don't find the answer soon, I might just go with the Epson driver.

Thank you.

 

[L Gebhardt]

That issue doesn't look like banding. It looks like random dark dots added to the dither pattern for that shade. My guess is it's an issue with the setup of QTR. Maybe an artifact of some of the curves.

 

[alexbo]

I actually tried printing the original profile sent to me by Ron Reeder that he himself built, and it displayed the same artifacts, while he had no problems with that profile.

I followed Ron's workflow, building my profile. It was simpler at first, but when I noticed this problem, I started adding mode inks to the mix, because I read in one of his articles that using all of the available inks could help with banding issue, as it makes transition from one tone to the other smoother. At least that was the way I understood it.
But then, when that didn't help, I started removing light inks one by one, until I ended up with just a single LK covering that part of the curve, and still, the banding was there!

Everything points towards some flaw in my environment. Somewhere between QTR driver and the printer something doesn't work the way it supposed to, and it has nothing to do with the profiles or the printer calibration. Maybe QTR driver on Windows 7 64bit doesn't work well ... I'm not sure.
Ron suggested I try printing from Mac which I'm planning to try tonight. If that works with the same profile, I will definitely know that my Windows-QTR environment is the problem.
And if it doesn't ... I'm going to start with ABW workflow and see where that takes me.

Thank you everyone for you generous input.

Yes, QTR can be a rabbit hole. You might try finding a working profile on the web and tweaking for your environment. You might already be doing this, but consider one curve for light inks and one for dark inks instead of creating a separate curve for each color. Some colors don't contribute much UV blocking so not using these channels can simplify your profile. Ron Reeder has documented a pretty simple workflow for creating digital negatives with QTR. You can also use the Epson driver Advanced B&W with a yellow boost to create negatives. You will still need to create a correction curve, but you will bypass all the profiling required by QTR. PDN and other methods of colorizing work well on the 38xx, too.

 

[amphoto]

That definitely looks like posterisation - though not in the image itself if it prints ok with the Epson drivers. Very odd - definitely try some different versions of QTRip, you never know. When I was trying to get to grips with QTRip, I had problems with one version (on Vista x64) that wouldn't print correctly. Rolling back to an earlier version worked. Have you seen Clay Harmon's guide to QTRip and digital negatives? Eminently readable - you may find that odd bit of something in there to help. Here's the download link.

Good luck, and let us all know how you get on.

Angus

 

[alexbo]

Angus, yes. It does seem to be some form of posterization.

I have contacted Roy Harrington, and he seems to think that the issue is in the data.
That the gradient is only 256 steps and doesn't have enough information to print smoothly in a spot that has a high tonal separation.

I am thinking now that It's possible that I don't see it on the Epson driver print because it is not formatted for the same exact tonal separation as my QTR profile.

I hope it's as simple as that.
I'm going to continue testing and report back.

Thank you

P.S. Yes, I read Clay's guide. It's a great tutorial.

That definitely looks like posterisation - though not in the image itself if it prints ok with the Epson drivers. Very odd - definitely try some different versions of QTRip, you never know. When I was trying to get to grips with QTRip, I had problems with one version (on Vista x64) that wouldn't print correctly. Rolling back to an earlier version worked. Have you seen Clay Harmon's guide to QTRip and digital negatives? Eminently readable - you may find that odd bit of something in there to help. Here's the download link.

Good luck, and let us all know how you get on.

Angus

 

[htswv]

I was going to suggest that perhaps the problem is with the image file data but it looks like you've already considered that possibility.

 

[alexbo]

So, I just wanted to wrap up this thread by saying that the artifacts seem to come from the digital image itself.
Apparently Photoshop gradient banding is a well known problem. It seems that it's almost impossible to create a completely smooth gradient in Photoshop regardless of the file resolution. At certain levels of contrast the banding becomes very apparent. Those who design with Photoshop have to reserve to trick like adding noise to the image in order to mask the banding issue. Which is of course unacceptable when it comes to digital negatives.
The reason I couldn't see the banding while printing through Epson driver was due to the fact that the contrast of the negative print was much lower then the one created with my custom QTR profile.
In real life situations, this issue should't really be a problem, since gradation of tone is handled differently by both, the digital cameras and film scans.

Thank you everyone for your help.

 

[clay]

This looks like posterization to me. It probably means the the curve adjustments in the 90-100% range are too rapid. The biggest reason this probably won't be a problem is that these values correspond to the deepest shadow areas of your negative, and it is unlikely your print process will be able to discriminate that bottom 5% of gray anyway.

 

[pschwart]

well-calibrated inkjets and carbon transfers can distinguish tones in this range. If there is posterization in the step tablets and ramps, then it may creep into your prints. Best to have ink profiles and correction curves that do not result in posterization, then no need to worry.

This looks like posterization to me. It probably means the the curve adjustments in the 90-100% range are too rapid. The biggest reason this probably won't be a problem is that these values correspond to the deepest shadow areas of your negative, and it is unlikely your print process will be able to discriminate that bottom 5% of gray anyway.

 

[Kenneth Lee]

One of the first steps using QTR, is to print out an "Ink Pattern Page". Looking at my printout (Epson R2400), I noticed distinct banding in the LLK channel - all along the density range. None of the other inks display that banding, but the LLK shows it profusely.

I'm not familiar with the internals of QTR, but I don't think the banding comes from Photoshop's inability to render a smooth gradient: it's just the printer itself, as far as I can tell. Which tells me it's better not to use that ink at all.

Being able to turn off that inkjet nozzle, strikes me as one of the (few) compelling advantages of QTR.

 

[pschwart]

Not an option when banding is in a channel you need -- time for a print head cleaning and realignment And be prepared to send a lot of ink straight to the waste tank. Do you see the banding on OHP, paper, or both?

One of the first steps using QTR, is to print out an "Ink Pattern Page". Looking at my printout (Epson R2400), I noticed distinct banding in the LLK channel - all along the density range. None of the other inks display that banding, but the LLK shows it profusely.

I'm not familiar with the internals of QTR, but I don't think the banding comes from Photoshop's inability to render a smooth gradient: it's just the printer itself, as far as I can tell. Which tells me it's better not to use that ink at all.

Being able to turn off that inkjet nozzle, strikes me as one of the (few) compelling advantages of QTR.

 

[Kenneth Lee]

So far, I have seen the banding on the OHP. I'll do a cycle of head-cleaning and see if that removes it. Thanks Philip !