Epson Stylus 1500w - Missing details at 1440 ppi

[Gubenco]

I am working on a lenticular project since 2016.
This is what it should look like

I need to print at highest possible resolution regardless of speed or cost. I have already created the optical device and now I am looking to find a solution to print it. I prepare to do some tests with other printers especially older Canon with 9600x2400 dpi. "Older Canon" because they are not producing printers with such a high resolution anymore. I am also considering to build a printer, anyway I need an aqueous flatbed. Having said that I invite everyone has a Canon with 9600x2400 or 4800x2400dpi to help me printing these square and circles. I declare that my Epson is only 70% capable to print at 1440dpi. Even a 1200dpi image will be better than what I have now. I have attached two test images and I will be grateful to see some results. Thanks

 

[maximilian59]

I use 1440 ppi, Aliased image are generated without filtering and pixel interpolation is made using nearest neighbor method.

So you alter the image during the printing process. Did you ever check how the image looks after enlarging it, but not printed at 100% view. I wonder whether the problems are part of the enlargement.

 

[The Hat]

I need to print at highest possible resolution regardless of speed or cost

I simply just don’t get what your trying to do..

From what I can see, your pissing up a rope with your ideas of 720 DPI and 1400 Dpi, because none of these things matter a jot, if your file size is to Feckin small.

Your last two samples posted above are only 181 Kb, but what you need is Megabytes not Kilobytes… try increasing your file size…
I made the sample 63.5 cm x 40 cm, and used 236.22 pixels / cm. before printing..
But when I printed out your file and is was rubbish, and when I increase the file size it work fine..

P.S. Would it be your're trying to print Vector Graphics.. ?

 

[Gubenco]

@maximilian59
I do not enlarge the image. I create it at "native" resolution. Precisely with one pixel per dot.

Curently I need 1440 dpi. I need much more but this is the maximum Epson say they can print.
In may tests 1440 is the highest resolution that makes a difference. Of course it does't translate to print as well as 720 does. I am trying to understand this and to overcome some artifacts.

My printer says it puts Max 5760x1440 dots per inch and this is correct. If I look to my prints at microscope I see - 1440x1440 grid (not 5760x1440). 5760 as means 4 dots of different color can be overlapped (and they are very precise at doing this). There is 6 colors but pairs of them compete for the same posion in the 1440x1440 grid LM and M theoretically are never overlapping. I don't know how other printers but Epson 1500w doesn't put 5760dots on X axys.
What I do not see is "dot variation" - this is not on my pages. at least not as they say.

My printer succeed in many ways to transfer a 1440ppi on page with a lot of image detail I can recognize, especially colors.

What I am trying now is to take maximum from this printer or any other.
Some images look better printed slightly rotated, I do print a lot of patterns.

 

[Gubenco]

@The Hat
Thank you for printing my images. I knew you have Canons and I really hoped you'll print them. Thanks.
The images are very small on the page. One is 4.28 wide printed with Canon at 2400dpi and the second is 8.57 cm at 1200dpi. I am curious when you printed them without scaling, at their current resolution, did the lines appear continuous without missing dots?

 

[stratman]

I am confused in what it is you are trying to do and why you are trying to do it as described. I also do not claim to be able to answer a single question you have.

My gist is you are trying to print 3D holograms using an ordinary inkjet printer. If so then you will be legend for forgers.

As @The Hat asked, are you trying to print a Vector or Raster file? What format is the original file in? PNG files you posted are not vector files.

Are you taking an image and shrinking it or enlarging it when you make the file (in Photoshop?) or when you print the file (from Photoshop)?

Why are you using Nearest Neighbor interpolation? Have you tried other algorithms?

You talked about Moire effect. Are we seeing an interesting effect of alias jaggies that causes missing segments to fill in when the image is rotated?

Couple links that may or may not help. Amongst other areas, the first references Moire effect and the second link Qimage.

https://americaninkjetsystems.com/learning-center/what-is-anti-aliasing/

https://forum.luminous-landscape.com/index.php?topic=35890.0

 

[Ink stained Fingers]

The internal file properties of the 2400.png file are:

Dimensions 1200 x 756 24 bit color
Print size 318 x 200 mm
embedded resolution 96 x 96 dpi

318 mm = 12.51 inches x 96 dpi are 1200 dots , so the internal data match at this point.

But why should I print a graphics file with 96 dpi when we are discussing issues with 720 dpi or 1440 dpi resolution ? I'm lost and step out of the discussion.

 

[maximilian59]

I do not enlarge the image. I create it at "native" resolution. Precisely with one pixel per dot.

This means, that for every inch of the printed file you have 1440 pixels. Your provided pictures are very small. Printing them with 2400/1200 PPI to the printer gives with a 1 to 1 relation a 1“ wide print, as you say with 1 pixel per dot. Everything else is a recalculation of the file for the print.

And what you found out with the 1440x1440 square is nothing new. All tech papers from Epson are referring to a square. All they state is, that there are this many concrete positions where a ink dot can be put down, not per ink. The higher figures for x-direction are just marketing speech. It is not wrong, but isn’t correct either.
With the light colors it is also understandable, that they are used the one or the other. They have normally the same hue but different lightness. If needed darker, the standard is used, which can’t get lighter. For lighter tones the light color inks are used. Makes no sense to use both on the same dot one on the other.
Did you ever calculate the theoretical diameter of a ink dot? For 1.5 pl it is about 0,018 mm for a hemisphere. When I did the calculation correct. With a little bit of dot gain that is the maximum for 1440 DPI. So if you want to distinguish two dots from each other with the same color without overlapping, one dot has to be left out and you are back to 720 DPI printed. With 3 pl the theoretical diameter is 0,025, which is only 40% more and not doubled as some may think.
Maybe you are able to measure the real diameter of a ink dot with your microscope.

 

[Gubenco]

@stratman
Yes, it is a sort of holograms. very large ones (like 2x1.2 m). Forgers won't appreciate this.

I can not use vectors. I render my images/paterns I a 3d software. I am using Photoshjop only for prepress and some time for color corection (I print from Photoshop). I use Nearest neighbor because of the small pixel budget I have. At small scale aliasing kills all the detail reducing anything to gray.

Are we seeing an interesting effect of alias jaggies that causes missing segments to fill in when the image is rotated?

Yes. I was expecting only continuous lines. Rectangles and circles Like in my attached images.
Thank you for Qimage link and thank @Ink stained Fingers as well, I have found interesting things today using this software.

 

[Gubenco]

But why should I print a graphics file with 96 dpi when we are discussing issues with 720 dpi

Maybe the image was converted on the server. I have attached a .zip
It is the same image 4048x2549 once with 2400dpi and then 1200dpi for canon test.