Epson Dithered Output

[John R Smith]

Printer experts -

More guidance is required, this time on Epson printer output, rather than input. I’ve started a new topic for this, instead of burying it in our epic "Clarification on Print Resolution" thread.

I have been looking at the specifications for various photo printers on the Epson USA site (which is actually a lot better than our UK one). The maximum dpi dithered output for the R2880 and the new R3000 is 5760x1440 (the minimum droplet sizes are 3 and 2 pl respectively). For the 3880 and the 4880 (and presumably the other large format printers) the maximum dithered output is given as 2880x1440 (minimum droplet size 3.5pl).

What exactly does this mean in practice? Ignoring the droplet size for a moment and assuming that all of these will accept a 720 ppi input, then for a given print size does this mean that at the highest quality settings the smaller 13 inch printers will give a smoother, less granular rendition than the 17 inch machines? All of these have the same nozzle patterns, with 180 nozzles per colour.

Your input is awaited.

John

[John R Smith]

Let me put this in a slightly different way, then. For those of you who are fortunate enough to have access to a 13 inch printer (R2400, 2880, 3000) and a 17 inch 3800, 3880, or 4880 or whatever.

If you print the same file from the same software to both printers at 10x8 or 16x12 at the highest quality settings, can you see any difference between the two prints?

John

[Sven W]

Maybe with a loupe  
All my prints are made for at least a 25-50 inch viewing distance.

/Sven

[michael]

John,

I have a 3880 and an R3000 (review coming one of these days). Jeff Schewe and I did this comparison a few weeks ago when we were shooting our new Camera to Print video tutorial.

With a very high quality file, and looking very carefully (with a loupe) you can see a small resolution difference.

In this case it's not called Pixel Pepping, it call Dither Diving.

Michael

[Mark D Segal]

John,

I have a 3880 and an R3000 (review coming one of these days). Jeff Schewe and I did this comparison a few weeks ago when we were shooting our new Camera to Print video tutorial.

With a very high quality file, and looking very carefully (with a loupe) you can see a small resolution difference.

In this case it's not called Pixel Pepping, it call Dither Diving.

Michael

There should be a law against dither diving. The dither diver can be blinded by the eyewash in the marketing pool. :-)

[michael]

Good one Mark!

Michael

[Schewe]

I have a 3880 and an R3000 (review coming one of these days). Jeff Schewe and I did this comparison a few weeks ago when we were shooting our new Camera to Print video tutorial.

To be precise, what we tested was whether or not a native file resolution of 450 PPI (or so, don't remember the exact PPI now) on the R3000 could be improved upon by upsampling the image to 720PPI. And yes, with the right file it can...

We didn't actually compare the R3000 to either the 3880 or the 7900 with the same file at different resolutions.

What I would expect is for smaller prints at higher resolutions on glossy media, the R3000 would have a slight edge in the potential detail over the 3880 or 7900 but you would need to know what to look for and have high resolution files to start with...

[John R Smith]

Thanks for your replies, folks.

I tend to view my prints not framed and on a wall, but in my hand and sitting at normal reading distance. That's because almost all of my experience of photography has not been in galleries or museums, but from from printed books and prints passed around from hand to hand, so for me a photograph is generally around 10x8 maximum. For that reason print quality at small sizes is very important to me, and that is where inkjet prints still have a disadvantage compared with a darkroom silver B/W print.

The problem is that there is so little useful information available from the printer manufacturers about the capabilities of their products. I used my Epson R2400 for the first two years only at 360 ppi because I had no idea that it could accept and utilise 720 input. There is absolutely nothing in the user manual or other documentation to tell you this. In fact there is nothing in the manual about input resolution at all. There isn't much in the manual about output either, except in general descriptive terms - no hard facts on dpi and dither. When I got LR 3 last year and started experimenting with 720 ppi, it was a revelation. It seems ridiculous that for decent information about the hardware, you are better off here on the Luminous Landscape or using Eric Chan's excellent site than asking Epson themselves.

John

[Mark D Segal]

Thanks for your replies, folks.

I tend to view my prints not framed and on a wall, but in my hand and sitting at normal reading distance. That's because almost all of my experience of photography has not been in galleries or museums, but from from printed books and prints passed around from hand to hand, so for me a photograph is generally around 10x8 maximum. For that reason print quality at small sizes is very important to me, and that is where inkjet prints still have a disadvantage compared with a darkroom silver B/W print.

The problem is that there is so little useful information available from the printer manufacturers about the capabilities of their products. I used my Epson R2400 for the first two years only at 360 ppi because I had no idea that it could accept and utilise 720 input. There is absolutely nothing in the user manual or other documentation to tell you this. In fact there is nothing in the manual about input resolution at all. There isn't much in the manual about output either, except in general descriptive terms - no hard facts on dpi and dither. When I got LR 3 last year and started experimenting with 720 ppi, it was a revelation. It seems ridiculous that for decent information about the hardware, you are better off here on the Luminous Landscape or using Eric Chan's excellent site than asking Epson themselves.

John

John,

First, I'd be very surprised that one could sustain a general proposition today about any superiority of silver-gelatin printing over the highest quality inkjet printing now available. I'm probably opening-up a hornet's nest of debate with this suggestion, but so be it. I'd want to discuss this proposition in terms of what's measurable, not "impressions", though I don't discount the latter. I simply have no time to do it right now, but one of these days I may well haul out some of my old silver gelatin prints and try to detect and measure their maximum black compared with what I get from Ilford Gold Fibre Silk or Canson Baryta Photographique from my Epson 4900. But if you or any one else reading this thread has already done that, I'd be very interested in reading the results. In terms of resolution or smoothness of tonal gradations or colour inconstancy (commonly and erroneously called "metamerism"), I think those horse-races are over with.

As for the information about resolution in Epson's manuals, the ones I've ever used do tell me what to use for which purposes and which settings will give you better photo quality. I agree it's not an erudite scientific discourse - indeed rather telegraphic; but for me that's all I need from them. The best guide for me is to try different settings and look at the results, as you have done.

[Farmer]

John - did you ever specifically ask these questions of Epson?

Please bear in mind that the vast majority of users of the products simply don't want to know and user guides are written for the majority of users.

[John R Smith]

First, I'd be very surprised that one could sustain a general proposition today about any superiority of silver-gelatin printing over the highest quality inkjet printing now available.

As far as I can see, there is no debate about inkjet v chemical if we are talking about colour, and/or large print sizes. Inkjet wins hands-down. But a fine B/W silver or platinum print at 10x8 or less is a different matter. Look at the highlight detail, and in particular the skies and the gradations in clouds. I am talking about chemical prints from MF or LF negatives, of course. I have a passion for fine small prints, as I said before.

John - did you ever specifically ask these questions of Epson?
Please bear in mind that the vast majority of users of the products simply don't want to know and user guides are written for the majority of users.

I did, last year on the phone to Epson UK. The gentleman on the other end didn't seem to know what I was talking about.

John

[Mark D Segal]

My old enlargements - the ones I've kept are from 120 roll film negatives (2 1/4 by 3 1/4) or 3 1/4 by 4 1/4 inch sheet films - goes way back, so I know what you mean. I'm interested in seeing through passion to real numbers. As for tonal gradations, in inkjet prints they only get better with smaller size and higher resolution, so I remain to be convinced.

[John R Smith]

As for tonal gradations, in inkjet prints they only get better with smaller size and higher resolution, so I remain to be convinced.

Sorry, Mark, but that's not true. Tonal gradations do get better with higher resolution, to the limit of input ppi - say 720 in our case. But once you print smaller than a one-for-one pixel representation on paper, and have to downsample, then
your detail and equally the subtleties of gradation will be degraded. So as you print smaller than one-for-one on paper, you begin to lose quality, very noticeably. This does not happen with a silver print, because of the analogue nature of the process - the silver grains in the emulsion are so small that you would need a microscope to see them, not a loupe. Until recently, this effect was of little concern to most digital work, because with sensors of only 12 or 15 MP we were almost always upsampling, even at small print sizes. But now, with 40, 60 and 80 MP sensors, this is no longer true, and when we print small we are downsampling and compromising resolution and tonal smoothness.

John

[Shane Webster]

Based on what I saw Friday between my 7880 and 4900, the answer is it depends.  I ran some tests Friday with the printer resolution file (and will post some of my observations about those tests on that thread), but I did discover something interesting regarding dither and super microweave ("SMW") when looking at the Column A output.  First, on my 7880 and with matte black ink there was no horizontal banding even when printing at 1440 (printing Column A at 360 ppi at 360).  On my 4900, I have horizontal banding when using matte black ink with the Column A test even when printing at 2880.  I was very surprised and not overly pleased that I have the horizontal banding at the highest print setting when using matte black ink with the 4900. 

With PK active in the 4900 and printing the Column A image at 360 ppi output at its native resolution, there is obvious banding at 1440 across all four ppi sections of Column A without SMW on and no banding on the 180 or 240 ppi sections with barely (or am I imagining them) perceptible horizontal banding on the 360 and 720 ppi sections when printed at either 1440 or 2880 with SMW on.  The end result was when printing the Column A test on the 4900, its MK output did not match the 7880's output of the same input file outputted at either 1440 or 2880.  However, when printing PK with the 4900, its output at either 1440 (SMW on) or 2880 virtually matched the 1440 output of the 7880 (I say virtually because there is a difference in shading of the 720 ppi section due, I think, to either dot gain of printing matte black, or printing at 1440 since the printer is not using the smallest droplet available at 1440 or because I don't think the 7880 has as small a droplet as the 4900 leading to the columns appearing closer together making the section appear darker gray).

[Mark D Segal]

Sorry, Mark, but that's not true. Tonal gradations do get better with higher resolution, to the limit of input ppi - say 720 in our case. But once you print smaller than a one-for-one pixel representation on paper, and have to downsample, then
your detail and equally the subtleties of gradation will be degraded. So as you print smaller than one-for-one on paper, you begin to lose quality, very noticeably. This does not happen with a silver print, because of the analogue nature of the process - the silver grains in the emulsion are so small that you would need a microscope to see them, not a loupe. Until recently, this effect was of little concern to most digital work, because with sensors of only 12 or 15 MP we were almost always upsampling, even at small print sizes. But now, with 40, 60 and 80 MP sensors, this is no longer true, and when we print small we are downsampling and compromising resolution and tonal smoothness.

John

John, first, on a Phase One p40+ digital back for example, the large dimension is 7320 PPI, so printed at 720 it's a tad over 10 inches. The small dimension is 5484, so at 720 that's 7.6 inches. This seems to fit within your category of small prints. There is hardly downsampling to speak of in this context. In fact once one crops a bit, as normally happens, there need be none period. But even if there were, I've been doing a lot of testing of downsampling in another context recently, going from 720 to 360 (even 1440 to 360) and I'm impressed with how little difference it makes to anything.

[John R Smith]

John, first, on a Phase One p40+ digital back for example, the large dimension is 7320 PPI, so printed at 720 it's a tad over 10 inches. The small dimension is 5484, so at 720 that's 7.6 inches. This seems to fit within your category of small prints. There is hardly downsampling to speak of in this context. In fact once one crops a bit, as normally happens, there need be none period.

Mark, that is absolutely correct, and for 720 ppi on my 39 MP back the print size is 10x7.5 ins. Which is my "Digital Contact Print", and the size I have recently adopted for the majority of my work. Funnily enough, in my darkroom days we would not have considered this a "small print". Most of the work I did for magazines and newspapers was 7x5, and 10x8 would have been a bit special. I used to like to print my MF Rollei work as fine prints to 8x8 ins on 12x9.5 ins paper, with a large white border. That 12x9.5 paper was the largest I ever did print. I've pulled some of these out beside me here as I type, on the old Ilford Galerie, and they look gorgeous. There's certainly no problem with D-max!

John

[michael]

You're right Jeff, but since you left I have done this exact experiment between the two printers, and my findings are as mentioned.

I just confused the comparison you and I did with the one I did a few days later on my own.

Michael

[Mark D Segal]

Mark, that is absolutely correct, and for 720 ppi on my 39 MP back the print size is 10x7.5 ins. Which is my "Digital Contact Print", and the size I have recently adopted for the majority of my work. Funnily enough, in my darkroom days we would not have considered this a "small print". Most of the work I did for magazines and newspapers was 7x5, and 10x8 would have been a bit special. I used to like to print my MF Rollei work as fine prints to 8x8 ins on 12x9.5 ins paper, with a large white border. That 12x9.5 paper was the largest I ever did print. I've pulled some of these out beside me here as I type, on the old Ilford Galerie, and they look gorgeous. There's certainly no problem with D-max!

John

Yes, even without seeing your stuff, I have every confidence that visually there's no problem with DMax. Do you have a spectro? If so can you measure it? That would be interesting!

[John R Smith]

Yes, even without seeing your stuff, I have every confidence that visually there's no problem with DMax. Do you have a spectro? If so can you measure it? That would be interesting!

Sadly, Mark, I have no spectro. However, to satisfy our joint curiosity, I have just compared the blackest black on the old silver print Ilford Galerie to the blackest black from my Epson on Harman Gloss FB Al using the Smith Mark I eyeball in best daylight. An imperfect test I know, but I can see no difference at all.

John

[Mark D Segal]

Thanks for doing that John. That's pretty much the visual observation I expected. Eyeballs are probably where it should stop, but sometimes the numbers reveal interesting things. I'll see what I can manage.