Epson p10000, p20000 dot placement / banding

[srialto]

Hello,
I have had a p20000 since January of this year and there are many things to like about the machine. The problem we are having is with its dot placement in the context of lenticular printing where we see a significant amount of banding on the finished piece. Lenticular printing involves interlacing multiple images together into a single file, printing, and then placing an optical plastic over top of the image. Its at this stage that we see banding, its really not visible prior to the optical overlay. Previous generations of Epsons did not have this problem, at least up to the 9880 which was the machine I owned prior to the p20000. As you probably know Epson has changed its stated output resolution from 2880x1440 to 2400x1200 for these printers, along with other changes to the print head design.

Now for some clarification on the banding and what we have tried.
First the banding is vertical as it comes out of the printer - thus not caused by the print feed mechanism. You can see an example in the picture.
We have used all combinations of settings and done over a hundred tests including printing through the driver using all resolution settings with bi-directional and uni-directional on and off, printed through mutliple rips using custom high resolution profiles (not epson contone), increased and decreased suction, performed head alignments. Our latest exploration involved sending different ppi files to the printer to investigate if the output dpi was perhaps off by some fractional difference from 300 or 600. There may still be some room here for testing, but we are grabbing at straws on where to go from here.

I don't expect many people on this forum to be familiar with lenticular printing but perhaps someone has experience with these new generation Epson print heads and how they are handling dot placement and what options there might yet be to explore. We are even ready to jump ship from Epson but the tests we did with the new Canon Pro printers did not give us better results.

Any thoughts?

[deanwork]

I haven't printed with either of the new Epson models and certainly not with the materials you are using.

When these new 10-20k printers were offered I was very interested because of the added gray channel.
I had hoped to buy one and linearize and print with it out of the latest Ergosoft rip for highly permanent fast bw.  I ordered a sample bw print from Epson and was not the least impressed. My Canon and Hp tritone prints looked much nicer without having to add color dots, not to mention piezography through any old Epson printer that is far superior in tonal precision to all of them.

I then made a call to Epson and they told me straight out that this new design was created primarily for one thing, speed. They told me the extra gray was added to prevent banding at these increased speeds. They are designed to crank out very  large prints in big volumes, and there is certainly a market for that just not what I was interested in.

If you have not tried the 9000 and 8000 models out off Ergosoft with your workflow that would be very useful for you to try. I realize you have invested in the k series, but they just may be the wrong tool for your needs, I don't know.

Hopefully someone who has had some hours with these printers can offer more useful suggestions of other things to try.

John

[Wayne Fox]

Are you printing the same files?  Those type of files don’t lend themselves to resizing. If they were created for a 360 dpi device perhaps the issue is in the re interpolation. I don’t know how they are created but maybe new files need to be generated for a 300 dpi device for the Epson p10/20000 or the canons

[srialto]

The image files being sent to the printer take into account the resolution change of the new epson print heads. In fact, hearing from another person that the actual dpi might not be precisely 300 I tried sending variations of ppi to the printer to see if there was a magic number. In the Epson white paper they give some specifications that suggest this might be the case, in fact they give inconsistent specifications leading to different numbers. But I can't tell if this is going down unfruitful rabbit holes and my tests so far have not proven anything.

[BrianWJH]

Have you eliminated the possibility that the overlay plastic is not causing the issue, is it the same batch of material you had success with on the 9880?

If you have access to an older x880 printer, maybe try the combination of images and material to make sure it's not the plastic overlay.

How is the overlay applied to the print, is the adhesive causing the issue?

Brian.

[Wayne Fox]

The image files being sent to the printer take into account the resolution change of the new epson print heads

How are the files sent to the printer?  I guess my logic was if the problem is the same on both the p20000 and the Canon, the problem is not the printer but in the workflow.

[srialto]

Have you eliminated the possibility that the overlay plastic is not causing the issue, is it the same batch of material you had success with on the 9880?
How is the overlay applied to the print, is the adhesive causing the issue?

There are several different plastics we use and it happens to various degrees on all of them, including plastics we used on the 9880.
The adhesive is applied in the same way it was previously as well.

How are the files sent to the printer?  I guess my logic was if the problem is the same on both the p20000 and the Canon, the problem is not the printer but in the workflow.

Its a reasonable assumption, I was surprised to get bad results on the Canon as well. The imagery for the canon was sent through the print driver on an entirely different computer however, making the file prep the only possible variable. The file prep, or interlacing, is done through a dedicated software program, and I have used several in an effort to see the effect. There is a small difference from interlacer to interlacer, but none that resolve the issue to a suitable degree.

[deanwork]

16 bit rgb?

One possibility is to try Q-Image which has excellent resampling capability when sent to all of these printers. You should be able to use the free trial demo and now works with Mac also I believe.

There are several different plastics we use and it happens to various degrees on all of them, including plastics we used on the 9880.
The adhesive is applied in the same way it was previously as well.
Its a reasonable assumption, I was surprised to get bad results on the Canon as well. The imagery for the canon was sent through the print driver on an entirely different computer however, making the file prep the only possible variable. The file prep, or interlacing, is done through a dedicated software program, and I have used several in an effort to see the effect. There is a small difference from interlacer to interlacer, but none that resolve the issue to a suitable degree.

[srialto]

16 bit rgb?

One possibility is to try Q-Image which has excellent resampling capability when sent to all of these printers. You should be able to use the free trial demo and now works with Mac also I believe.

24 bit rgb. I have also tested on Q-image which did not improve the results.