Native Printer Resolution - Fact or Myth?

[Euell]

Obviously, that cannot occur if the resolution adjustments are left to the printer. It follows that it is important to know the best resolution for uprezzing, which logically would seem to be the printer’s native resolution.

Not true. I write image-processing software, and have done so since the late 1970s for such classic hardware as Versatec printers and the original Canon CX print engine. These really had fixed pixel-pitch and the dots (more like squares) were full black or absent. At these dot densities the use of dithering was essential, and still looked awful by modern standards. Even OEM-tuned italic font looked bad. Photoprinters today like my Canon Pro-2000 have high native dot pitch, and fancy dithering algorithms backed by 12 carts and color science. Up close, the dither patterns are frankly amazing. Resolution of the incoming print job is completely irrelevant so long as the detail is sufficent for the artist's purpose.

Sharpening is eye candy. All sharpening is a falsification of the base image. Since the eye-brain system is designed to seek edges and local contrast, we sharpen images to feed the addiction (hopefully in moderation). Analog photogs know that grain gives the impression of detail, and it's a convincing illusion—except in the sky! As for the sharpening parameters of contrast level and radius, the former is dialed up or down according to good taste. The latter is dictated entirely by the PrintSize/ViewingDistance factor, a.k.a. subtended angle of the print to the eye. DPI and other such measures are irrelevant to sharpening. In analog space, a gelatin print or projected slide can deliver the same eye-candy satisfaction if you take the (considerable) trouble of goosing those media "in post": the original USM process.

Interesting, but what is false above? Sharpening need not be done as the last step before printing and after setting the final resolution OR sharpening is unnecessary (the latter being a minority position) ? I note that there are various methods of sharpening and increasing local contrast. Most digital cameras don’t even offer the option of turning off internal sharpening of JPEG’s.

 

[kciuk aka Ken60]

When all this digital stuff took off I was quite an early adopter, and I remember going into the lab where they worked on my 5"-4" chromes. I told them that I was using digital now and one of the guys said to me can we play with a file please, we want to try out our new print system.

I took in a 3mp image from a Canon EOS D30 , and while I watched they output an A3 size print. It was beautiful, detailed and quite saleable.....worthy of at least the wedding output from a Hassy. It does not take long to understand that to get that quality from a 3MP file required there to be an expensive rip on the system.

 

[Keith Cooper]

If there is nothing to magic numbers, then you are, in essence, suggesting that the printer algorithms work just as well as resolution adjustment in software, at least when down-rezzing. Few really care about this, by the way, because up-rezzing is where the question of native resolution becomes more important. Should we up-rezz with specialized software or let the printer handle it? If there is no difference, then why take the extra step? Well, for one thing, up-rezzing with software allows sharpening at final resolution. We, of course, realize that your experiment was limited, as you indicate above. However, your conclusion that there are no longer magic numbers is also misleading, except in the limited context of your down-rezzing experiment. It may be that the same conclusion would apply where uprezzing, but, again, allowing the printer algorithms to adjust resolution either up or down from the original resolution of the photo file precludes sharpening at the final resolution of the print.

What I was looking at was whether feeding increasing levels of detail into the driver is reflected in the detail in the print. It is. It also seems that the increase of detail is a steady increase, rather than showing any discontinuities (or favoured values) - However this would need rather more specific (likely synthetic) test images to analyse (see the LL thread link earlier in this thread).

Now, whether that increasing level of detail comes from simple bicubic interpolation or something like Gigapixel AI is neither here nor there. Using 'better' printer settings may also have an effect, but I feel that is a second aspect.

There is nothing in my testing or other discussions to suggest that the concept of 'native resolution' is any longer worthy of significant attention.

As ever -for anyone wondering, please do some experiments...

I await actual results, rather than just what seems likely or reasonable ;-)

I'll definitely be returning to this with a bigger printer, but it has to be said that we are at the level where actually seeing any differences becomes problematic ;-)

 

[kciuk aka Ken60]

But Keith you already have a bigger A2 Epson 5000 , or is this yet another bigger printer being tested

 

[rambet]

If there is nothing to magic numbers, then you are, in essence, suggesting that the printer algorithms work just as well as resolution adjustment in software, at least when down-rezzing. Few really care about this, by the way, because up-rezzing is where the question of native resolution becomes more important. Should we up-rezz with specialized software or let the printer handle it? If there is no difference, then why take the extra step? Well, for one thing, up-rezzing with software allows sharpening at final resolution. We, of course, realize that your experiment was limited, as you indicate above. However, your conclusion that there are no longer magic numbers is also misleading, except in the limited context of your down-rezzing experiment. It may be that the same conclusion would apply where uprezzing, but, again, allowing the printer algorithms to adjust resolution either up or down from the original resolution of the photo file precludes sharpening at the final resolution of the print.

What I was looking at was whether feeding increasing levels of detail into the driver is reflected in the detail in the print. It is. It also seems that the increase of detail is a steady increase, rather than showing any discontinuities (or favoured values) - However this would need rather more specific (likely synthetic) test images to analyse (see the LL thread link earlier in this thread).

Now, whether that increasing level of detail comes from simple bicubic interpolation or something like Gigapixel AI is neither here nor there. Using 'better' printer settings may also have an effect, but I feel that is a second aspect.

There is nothing in my testing or other discussions to suggest that the concept of 'native resolution' is any longer worthy of significant attention.

I agree when the PPI for the paper size is around 200+.

But in the very rare occasions where I have an image that is say only about 100 PPI or less for the paper size then I would normally resample the image using Gigapixel AI to the printer's native print resolution so I can see near enough to exactly what I am going to get in terms of detail and sharpness after printing before potentially wasting paper and ink.

As ever -for anyone wondering, please do some experiments...

I await actual results, rather than just what seems likely or reasonable ;-)

I'll definitely be returning to this with a bigger printer, but it has to be said that we are at the level where actually seeing any differences becomes problematic ;-)

--
bye for now
Keith Cooper

 

[Euell]

If there is nothing to magic numbers, then you are, in essence, suggesting that the printer algorithms work just as well as resolution adjustment in software, at least when down-rezzing. Few really care about this, by the way, because up-rezzing is where the question of native resolution becomes more important. Should we up-rezz with specialized software or let the printer handle it? If there is no difference, then why take the extra step? Well, for one thing, up-rezzing with software allows sharpening at final resolution. We, of course, realize that your experiment was limited, as you indicate above. However, your conclusion that there are no longer magic numbers is also misleading, except in the limited context of your down-rezzing experiment. It may be that the same conclusion would apply where uprezzing, but, again, allowing the printer algorithms to adjust resolution either up or down from the original resolution of the photo file precludes sharpening at the final resolution of the print.

What I was looking at was whether feeding increasing levels of detail into the driver is reflected in the detail in the print. It is. It also seems that the increase of detail is a steady increase, rather than showing any discontinuities (or favoured values) - However this would need rather more specific (likely synthetic) test images to analyse (see the LL thread link earlier in this thread).

Now, whether that increasing level of detail comes from simple bicubic interpolation or something like Gigapixel AI is neither here nor there. Using 'better' printer settings may also have an effect, but I feel that is a second aspect.

There is nothing in my testing or other discussions to suggest that the concept of 'native resolution' is any longer worthy of significant attention.

As ever -for anyone wondering, please do some experiments...

I await actual results, rather than just what seems likely or reasonable ;-)

I'll definitely be returning to this with a bigger printer, but it has to be said that we are at the level where actually seeing any differences becomes problematic ;-)

Well now, you have thrown out a new thought in the discussion on “better printer settings.” Since printer settings affecting resolution are somewhat limited, it should be easy for you to elaborate on that thought. You have also implied equivalency between between simple bicubic interpolation and Gigapixel AI, which actually is capable of supplying detail only hinted by the original file, for better or worse. I presently do not use Gigapixel, but clearly it does things bicubic interpolation does not.

I would also remind that you committed to further testing comparing assumed native resolution software interpolation to interpolation accomplished solely in the printer, using files requiring enlargement beyond the resolution afforded by the original file. I’m sure you must have older files of 24 megapixels or less to play with in this regard. These could be used at the assumed native 360ppi of Epson printers to demonstrate any benefit by printing on 13x19 or 17x22 paper.

 

[Rob de Loe]

If there is nothing to magic numbers, then you are, in essence, suggesting that the printer algorithms work just as well as resolution adjustment in software, at least when down-rezzing. Few really care about this, by the way, because up-rezzing is where the question of native resolution becomes more important. Should we up-rezz with specialized software or let the printer handle it? If there is no difference, then why take the extra step? Well, for one thing, up-rezzing with software allows sharpening at final resolution. We, of course, realize that your experiment was limited, as you indicate above. However, your conclusion that there are no longer magic numbers is also misleading, except in the limited context of your down-rezzing experiment. It may be that the same conclusion would apply where uprezzing, but, again, allowing the printer algorithms to adjust resolution either up or down from the original resolution of the photo file precludes sharpening at the final resolution of the print.

What I was looking at was whether feeding increasing levels of detail into the driver is reflected in the detail in the print. It is. It also seems that the increase of detail is a steady increase, rather than showing any discontinuities (or favoured values) - However this would need rather more specific (likely synthetic) test images to analyse (see the LL thread link earlier in this thread).

Now, whether that increasing level of detail comes from simple bicubic interpolation or something like Gigapixel AI is neither here nor there. Using 'better' printer settings may also have an effect, but I feel that is a second aspect.

There is nothing in my testing or other discussions to suggest that the concept of 'native resolution' is any longer worthy of significant attention.

As ever -for anyone wondering, please do some experiments...

I await actual results, rather than just what seems likely or reasonable ;-)

I'll definitely be returning to this with a bigger printer, but it has to be said that we are at the level where actually seeing any differences becomes problematic ;-)

Well now, you have thrown out a new thought in the discussion on “better printer settings.” Since printer settings affecting resolution are somewhat limited, it should be easy for you to elaborate on that thought. You have also implied equivalency between between simple bicubic interpolation and Gigapixel AI, which actually is capable of supplying detail only hinted by the original file, for better or worse. I presently do not use Gigapixel, but clearly it does things bicubic interpolation does not.

I would also remind that you committed to further testing comparing assumed native resolution software interpolation to interpolation accomplished solely in the printer, using files requiring enlargement beyond the resolution afforded by the original file. I’m sure you must have older files of 24 megapixels or less to play with in this regard. These could be used at the assumed native 360ppi of Epson printers to demonstrate any benefit by printing on 13x19 or 17x22 paper.

Keith does an enormous amount of testing and writing to support people who want to print. He does it all for free.

Here's a thought: can you do this test and post and write up your results?

 

[NAwlins Contrarian]

I also used Qimage for a while until I realised I could get just as good, if not better prints for some photos, without it.

In what way(s) is any other software--other than something like Gigipixel AI if you have to do a massive upscale--producing better prints of any photos, compared to Qimage? And what software is producing these better prints? I'm pretty skeptical of the "better" claim, but I'm certainly willing to learn.

Regarding "as good", I certainly think that in many ways, any properly-functioning color-managed software should produce pretty similar prints. IMO the ways that Qimage distinguishes itself are somewhat better resampling and sharpening than e.g. Lightroom, plus power and convenience features. In many cases a print from many other programs should / will look very similar. But compared to maybe all but a few, Qimage is more flexible and full-featured, and probably a little better for image quality.

But hey, maybe I'm wrong, I'm interested to hear.

 

[Euell]

If there is nothing to magic numbers, then you are, in essence, suggesting that the printer algorithms work just as well as resolution adjustment in software, at least when down-rezzing. Few really care about this, by the way, because up-rezzing is where the question of native resolution becomes more important. Should we up-rezz with specialized software or let the printer handle it? If there is no difference, then why take the extra step? Well, for one thing, up-rezzing with software allows sharpening at final resolution. We, of course, realize that your experiment was limited, as you indicate above. However, your conclusion that there are no longer magic numbers is also misleading, except in the limited context of your down-rezzing experiment. It may be that the same conclusion would apply where uprezzing, but, again, allowing the printer algorithms to adjust resolution either up or down from the original resolution of the photo file precludes sharpening at the final resolution of the print.

What I was looking at was whether feeding increasing levels of detail into the driver is reflected in the detail in the print. It is. It also seems that the increase of detail is a steady increase, rather than showing any discontinuities (or favoured values) - However this would need rather more specific (likely synthetic) test images to analyse (see the LL thread link earlier in this thread).

Now, whether that increasing level of detail comes from simple bicubic interpolation or something like Gigapixel AI is neither here nor there. Using 'better' printer settings may also have an effect, but I feel that is a second aspect.

There is nothing in my testing or other discussions to suggest that the concept of 'native resolution' is any longer worthy of significant attention.

As ever -for anyone wondering, please do some experiments...

I await actual results, rather than just what seems likely or reasonable ;-)

I'll definitely be returning to this with a bigger printer, but it has to be said that we are at the level where actually seeing any differences becomes problematic ;-)

Well now, you have thrown out a new thought in the discussion on “better printer settings.” Since printer settings affecting resolution are somewhat limited, it should be easy for you to elaborate on that thought. You have also implied equivalency between between simple bicubic interpolation and Gigapixel AI, which actually is capable of supplying detail only hinted by the original file, for better or worse. I presently do not use Gigapixel, but clearly it does things bicubic interpolation does not.

I would also remind that you committed to further testing comparing assumed native resolution software interpolation to interpolation accomplished solely in the printer, using files requiring enlargement beyond the resolution afforded by the original file. I’m sure you must have older files of 24 megapixels or less to play with in this regard. These could be used at the assumed native 360ppi of Epson printers to demonstrate any benefit by printing on 13x19 or 17x22 paper.

Keith does an enormous amount of testing and writing to support people who want to print. He does it all for free.

Here's a thought: can you do this test and post and write up your results?

No. That would duplicate what keith is about to do. And, we thank him.

 

[jrkliny]

Qimage is more flexible and full-featured, and probably a little better for image quality.

But hey, maybe I'm wrong, I'm interested to hear.

Qimage Ultimate includes a DFS sharpening algorithm which is better than any sharpen, usm or high pass sharpening available in Photoshop. It also includes more sophisticated interpolation algorithms. At this point neither makes much difference. Topaz software is now the way to go if you need the best for either of these post processing adjustments.

I doubt Qimage has any function which will make a noticeable improvement in print quality. Qimage is all about making the printing process foolproof and easy. Templates make printing easy. The page layouts are easier to use especially when including multiple images on a page. Qimage provides logs making it easy to retrieve exact paper and settings for prints made years ago. Qimage has lots and lots of specialized printing functions not available in Adobe products.

I also like the non-subscription billing practices. I bought Qimage software many years ago for a modest price. It is mine to use forever. After a few years I spent a few dollars extra and got all the past upgrades and an additional year of future upgrades. As I remember that cost me $14. It has now been many years since then. I have not seen any new functions of interest to me so I only did the one upgrade 5 or 6 years ago.

 

[rambet]

I have already posted that I use Gigapixel. Maybe you missed it?

If Qimage works best for you based on your tests and experience that is fine. I am not trying to change the app's anyone uses for printing.

But I'm not convinced based on what I have read in this thread that there is nothing as good or even better out there than Qimage based on my tests and experience.

As the OP, this thread is not about which app is best for printing.

If you are genuinely interested, feel free to start your own thread on the pros and cons of various printing apps and I will be happy to discuss there

 

[Keith Cooper]

But Keith you already have a bigger A2 Epson 5000 , or is this yet another bigger printer being tested

that's not big - it's just 17" ;-)

 

[Keith Cooper]

It does not take long to understand that to get that quality from a 3MP file required there to be an expensive rip on the system.

Maybe on that system, but as you say "When all this digital stuff took off" ;-)

That has not been the case for years ;-)

 

[Keith Cooper]

I await actual results, rather than just what seems likely or reasonable ;-)

I'll definitely be returning to this with a bigger printer, but it has to be said that we are at the level where actually seeing any differences becomes problematic ;-)

Well now, you have thrown out a new thought in the discussion on “better printer settings.”

No, it's one of the two specific questions asked in the article

Since printer settings affecting resolution are somewhat limited, it should be easy for you to elaborate on that thought.

Yes, I have, for the P900/700 PRO-200 and PRO-300

You have also implied equivalency between between simple bicubic interpolation and Gigapixel AI,

No, just that the precise mechanism used was not important to the questions being asked.

which actually is capable of supplying detail only hinted by the original file, for better or worse. I presently do not use Gigapixel, but clearly it does things bicubic interpolation does not.

I would also remind that you

Thanks

committed to further testing comparing assumed native resolution software interpolation to interpolation accomplished solely in the printer,

just to be clear what is this 'interpolation accomplished solely in the printer,' you refer to?

 

[Keith Cooper]

I would also remind that you committed to further testing comparing assumed native resolution software interpolation to interpolation accomplished solely in the printer, using files requiring enlargement beyond the resolution afforded by the original file. I’m sure you must have older files of 24 megapixels or less to play with in this regard. These could be used at the assumed native 360ppi of Epson printers to demonstrate any benefit by printing on 13x19 or 17x22 paper.

Keith does an enormous amount of testing and writing to support people who want to print. He does it all for free.

Here's a thought: can you do this test and post and write up your results?

No. That would duplicate what keith is about to do. And, we thank him.

No, please do have a go yourself - practical results frequently trump armchair wisdom.

The effort of deciding what questions are asked, what they mean, and actually trying for real can be a worthwhile exercise in itself...

My further experiments are also contingent on having the printer here available to do them... A2 is not big enough for some of what I want to look at.

I have actually done quite a bit about aspects of printing lower res files, such as:

Upsizing and sharpening an old photo for an A2 sized print

Editing, upsizing and sharpening part of an old digital photo from a 10MP camera to make an A2 sized print. Image resizing for prints

www.northlight-images.co.uk

This is more of a practical look - rather than addressing specific technical aspects.

Every time I do this sort of testing, I always come back to reminding myself that the biggest problems in printing are almost never the printer, rarely the choice of paper, and all too often down to basic failures in photography and creative vision ;-)

 

[kciuk aka Ken60]

Sure, what I was suggesting was that even over 20 years ago, the traditional numbers for making a print could be surpassed, and good results made with the right software.

Gigapixel , as good as it is , still requires quite a bit of trial and user intervention to find a path through its myriad of options, and to overcome its overly enthusiastic sharpening and noise controls.

 

[kciuk aka Ken60]

I always come back to reminding myself that the biggest problems in printing are almost never the printer, rarely the choice of paper, and all too often down to basic failures in photography and creative vision ;-)

Amen

One only has to peruse the hallowed corridors where the users of the most expensive equipment, finest lenses, most cutting edge technology reside .... and quickly understand that though they have overcome the restraints of the circle of confusion, they still fall at creating interesting and artistically exciting imagery.

 

[jrkliny]

Every time I do this sort of testing, I always come back to reminding myself that the biggest problems in printing are almost never the printer, rarely the choice of paper, and all too often down to basic failures in photography and creative vision ;-)

Absolutely. My photography and my prints are almost never limited by my gear, by my printer. Even so I need to do some research of my own.

I have no concerns about the original concern about printing at "native" resolution. Qimage Ultimate performs any interpolation as needed to achieve this, needed or not.

I do wonder if I can gain some improvement in my Epson 3880 prints by switching from 1440 dpi to 2880 dpi. I will need to do some experiments. In the past I have used the 1440 setting recommended by Red River and also used for their ICC paper/printer profiles.

I am only rarely interested in maximizing details on my prints. Many photographers seem to be trying to achieve the maximum possible sharpness, front to back. More typically I want the background to be less sharp with lower contrast and lower saturation.









--
Jim, aka camperjim

Special Places Photography

This site contains fine art photographs. Most are available for sale as prints either plain, matted, or framed. Subjects include nature, flowers, and landscapes. Many of the landscapes are from the major American National Parks, including Death Valley, Yellowstone, Arches, Canyonlands...

www.specialplacesphoto.com

 

[Euell]

I would also remind that you committed to further testing comparing assumed native resolution software interpolation to interpolation accomplished solely in the printer, using files requiring enlargement beyond the resolution afforded by the original file. I’m sure you must have older files of 24 megapixels or less to play with in this regard. These could be used at the assumed native 360ppi of Epson printers to demonstrate any benefit by printing on 13x19 or 17x22 paper.

Keith does an enormous amount of testing and writing to support people who want to print. He does it all for free.

Here's a thought: can you do this test and post and write up your results?

No. That would duplicate what keith is about to do. And, we thank him.

No, please do have a go yourself - practical results frequently trump armchair wisdom.

The effort of deciding what questions are asked, what they mean, and actually trying for real can be a worthwhile exercise in itself...

My further experiments are also contingent on having the printer here available to do them... A2 is not big enough for some of what I want to look at.

I have actually done quite a bit about aspects of printing lower res files, such as:

https://www.northlight-images.co.uk/upsizing-and-sharpening-for-making-a-print/

This is more of a practical look - rather than addressing specific technical aspects.

Every time I do this sort of testing, I always come back to reminding myself that the biggest problems in printing are almost never the printer, rarely the choice of paper, and all too often down to basic failures in photography and creative vision ;-)

With ya on that one !

 

[Euell]

I await actual results, rather than just what seems likely or reasonable ;-)

I'll definitely be returning to this with a bigger printer, but it has to be said that we are at the level where actually seeing any differences becomes problematic ;-)

Well now, you have thrown out a new thought in the discussion on “better printer settings.”

No, it's one of the two specific questions asked in the article

Since printer settings affecting resolution are somewhat limited, it should be easy for you to elaborate on that thought.

Yes, I have, for the P900/700 PRO-200 and PRO-300

You have also implied equivalency between between simple bicubic interpolation and Gigapixel AI,

No, just that the precise mechanism used was not important to the questions being asked.

which actually is capable of supplying detail only hinted by the original file, for better or worse. I presently do not use Gigapixel, but clearly it does things bicubic interpolation does not.

I would also remind that you

Thanks

committed to further testing comparing assumed native resolution software interpolation to interpolation accomplished solely in the printer,

just to be clear what is this 'interpolation accomplished solely in the printer,' you refer to?

Sending the file at original resolution to the printer for uprezzing or downrezzing.