Viewing distance-digital vs.silver prints

[Kirk Gittings]

I was recently going over some new prints from my Epson 4000 with a freind of mine who is a master digital printer. He was asking me why I printed at 2880 dpi, because "you whould have to stick your nose to the print to see the difference" between 2880 and 1440.

I always looked at my silver prints that close, because I wanted them to hold up well when looked at very close. Why would I approach inkjet prints with a lower standard?

[Michael E. Gordon]

Kirk: even under a loupe, virtually no difference has been detected between 1440 and 2880 dpi by most everyone who has conducted comparison tests. Just curious what you've seen with your own work and if you're just 'blindly' printing at 2880. The only thing I've seen from 2880 dpi is more lost ink and slower printing times.

www.mgordonphotography.com

[chris jordan]

Kirk, I heartily agree with you about the print quality issue-- the only people who talk about "viewing distance" are people making low-quality prints. One of the wonderful things about really good photographic prints is that you can walk right up and look closely and see more than you could from a distance. It's always disappointing to walk closer to a print and see it fall apart into grainy junk up close.

However, on the Epson question, it is pretty well established in the printing world that there is no difference in print quality between 1440 and 2880. I personally have done many tests, of targets made out of 1-pixel lines running every which way, with tiny text and miniature white circles on black backgrounds and other details much sharper than would ever appear in a print, and I can detect NO difference between 2880 and 1440, even with a loupe. If there were even the slightest improvement with 2880, I would use it, but there isn't. I've heard many others say the same thing.

So, by all means be a stickler about print quality (the standard for digital is just as high as with darkroom prints), and you can stick with 1440. And make sure you're feeding your printer a 360-dpi file-- that's the biggest quality factor in the Epson workflow.

best,

~cj

www.chrisjordan.com

[Oren Grad]

Kirk - You should approach inkjet prints with whatever standard standard feels right to *you*. The judgment should be yours - print a range of test pictures both ways and view them however you traditionally like to view prints to be sure you are happy with them. If you can see a difference that matters to you, then print at 2880 and forget what anybody else says you're supposed to want.

[Oren Grad]

> whatever standard standard feels right <

Oops - I guess some of us need to double up the words-per-inch, not the dpi...

[Bob Chambers]

My question is what looks better from any distance, the digital print or the silver print?

[Leonard Evens]

I think people tend to exaggerate what can be accomplished vis-a-vis large prints viewed very close up. Consider the issue of diffraction. Let's assume that you are viewing a 10 X enlargement at 250 mm, and that you took the picture at f/22. One common estimate says that in the negative diffraction would limit you to about 68 lp/mm. That would translate in the print to about 6.8 lp/mm, which is just at the level many people think they can detect a blur. Of course, you could open up more to reduce diffraction, but then you run into the same problem with depth of field. You significantly reduce the DOF you might otherwise expect. I think in practice we don't often do even that well.

I attended an exhibit of Thomas Struth's impressive photographs of museums interiors and other public spaces. He started with an 8 x 10 camera and used the highest quality techniques for making prints. But when I got really close, the image clearly deteriorated significantly. All my experience with viewing large prints confirms that.

It is nice to have the illusion that a photograph has infinite detail and the closer you get the more you see. But it is, when all is said and done, just that, an illusion.

[paulr]

It's not possible to generalize if digital looks better than analog, or vice versa, but neither realm puts any inherent limit on quality.
You'd have to compare a specific digital standard to a specific analog standard. 35mm fujichrome to a certain 5 megapixel camera, for example. It's like asking the same question with audio: CDs generally sound better than cassette tapes; LPs generally sound better than CDs, 24bit 96khz digital generally sounds better than LPs, 1/2" 2-track 30ips analog master tape sounds better still ... and so on.

[paulr]

"It is nice to have the illusion that a photograph has infinite detail and the closer you get the more you see. But it is, when all is said and done, just that, an illusion."

I think this is true. with small enlargements our eyes can be the limiting factor. With big enlargements the illusion gets preserved for the most part because we instinctively step back to look at them.

[julian_4860]

Leonard I also saw some Struth prints and they are no way near state of the art. His 'Paradise' pics, which I love, don't hold interest as you get closer. He was using a lightjet. Now the apparent res of a lightjet is inferior IMO to Epson printers. The day before the Struth prints, I saw some of Chris Jordan's and, believe me, no comparison. Folk talk about 'digital prints' as if its a known, uniform beast, and it isn't

[Brian Ellis]

It isn't a question of lowering your standards by using 1440 vs. 2880 or which looks better when viewed close up. There just isn't a noticeable difference between the two at any distance under any circumstances I know of.

[Leonard Evens]

Julian,

What I noticed in the Struth prints was loss of depth of field and fuzziness possibly due to diffraction. That was at a level way above anything that would show up because of the resolution of the printer. These prints were wall size, and everything looked fine even from two or three feet away. It was when I got within inches that I could see the fuzziness. I've had similar experiences with other prints, even those made conventionally with an enlarger.

[Bruce Watson]

Kirk,

I think your master printer is off a bit. He's making a generalization, when what matters is the specific printer, ink, paper combination you actually use.

I'm using a 7600 with the StudioPrint RIP, Piezotone inks, on Hahnemuhle Photo Rag 308gsm paper. I first linearized the printer at 1440 and 2880, then ran actual prints where the only difference was 1440 or 2880. I could see the difference from across the room, as could my wife. The 1440 print seemed to be a little bit darker (and yes, I rechecked the linearization, both were fine), and a little bit rougher.

Being an engineer by training, I was very skeptical of this result. So I arranged a double blind test to find out if what I was seeing was "real." I had a third party take the two prints and put them on my viewing wall under the same lights. Then I had some neighbors that don't know jack about printing look at them and pick which one they liked. The third party took their "votes" (I was out of the room), and of the ten votes, seven were for the 2880 print.

It gets worse. I've talked to other people with the same setup who have done the same tests (with their images). One said he got a tie, the other said he got preference for 1440.

I think it depends on your printer - some are better than others. Some can make the advantages of 2880 visable, others can not.

And, it gets worse still. I did the same test with bi-directional and uni-directional printing. I've always heard that uni-directional was the only way to go. But, no one could tell the difference.

So now I'm printing 2880 bi-directional. Go figure.

Clearly, very clearly, YMMV.

[chris jordan]

Hogarth, when you print at 1440 versus 2880, the printer lays down the ink differently. So, if you use the same profile for each setting, you will get different-looking prints. The most accurate way to make the test is to generate separate profiles for each setting, and then print the same file using the two settings each with its own profile. If your profiles are accurate, there will be no difference between the color or overall density or anything else that can be seen from across the room. The only difference will be the 1440 versus 2880 dot-per-inch resolution, which would be visible only with a loupe, and so far no one has been able to see this difference even with a loupe.

And, for those who say digital prints can't be better than analog, there is one factor which I think pushes digital prints over the top: Analog prints can't be any sharper than your original film; in fact, they can never be made as sharp as the original film because of the inherent character of the optical enlargement process. Digital prints, on the other hand, can be made sharper than the original film; in fact they can be sharpened to the point that they look totally awful! In the hands of someone who really understands digital sharpening, however, and who is willing to take the time to get it right, digital prints actually can be made "nose sharp" even at large size, with no visible artifacts from the sharpening process.

~cj

[tim atherton]

Hogarth,

I believe Colorbyte recomends agaisnt using 2880 with Imageprint. According to them, the way the RIP is set up it can generally obtain smoother gradations printing at 1440 than it can at 2880 (or something like that... it might be a better dither pattern...)

[Bruce Watson]

Chris,

StudioPrint, using grayscale inks, doesn't use a profile. It uses a linearization that gives you a straight line between black and white. ICC profiles are for color work. However, you can't use the same linearization for 1440 and 2880 anymore than you could use the same profile in color work. Indeed, you need a new linearization for any change in the RIPs parameters, because any change in parameters changes the way the RIP/printer lay down ink.

I'll agree that 1440 and 2880 lay down ink differently. That is the point. The question is how accurately can the RIP/printer make a pixel (assuming 360 ppi output to the printer) from the four or eight dots of ink. This is a function of the printer, and the RIP's dither pattern. Surely no one expects any two printers to lay down ink dots in exactly the same way. Some are better than others; printers do vary from sample to sample. Even Epson can't escape the laws of physics.

It's not necessary that you believe that I (and 11 other people) can see differences in real prints. It's easy enough to run your own tests and make your own determination, for your own print system. And, I should note, that I think you are much more likely to be able to see difference in grayscale prints than in color prints. In grayscale, you don't have color to obscure the texture, so it's easier to see.

It's a complex issue, YMMV.

[tim atherton]

PS - I'm not sure if that's also the case with the Piezo inks in the system instead of UC inks

either it may possibly be that you are are getting prints which are perceptually sharper, but not as "smooth" in the tonal changes :-)

[tim atherton]

Here's the info I was looking for:

"Output using ImagePrint, even on glossy-type papers, is better if you
don’t use the 2880 dpi setting. The 2880 profiles are only there because so many people asked for them. On the Epson 2200/76/9600 printers 1440 dpi allows a variable ink dot size whereas 2880 dpi uses just the smallest dot. Therefore, ImagePrint can actually create smoother tone transitions with 1440 dpi than with 2880 dpi."

[Bruce Watson]

Tim,

I don't know either. I asked my "judge" to ask the "subjects" if the prints were different, and if so, which print they prefered. I didn't ask the sharper / smoother question. You can only get so much out of free help before they get bored and wander off ;-)

[Bruce Watson]

Tim,

I believe my version of StudioPrint (10.5?) doesn't do variable dots. I think it just uses smallest dots for both 1440 and 2880. I say that because I think the newest version (11.0) claims to have enabled variable dot size. And once you go to variable dot sizing, all bets are off - you have to run the tests again (argh).

Again, I think it's a question of a specific RIP / printer / ink / paper combination. I suspect for some 1440 will be better, and others 2880 will be better. My objection is to the blanket statement brought up by the original poster that indicated that 1440 is *always* better, no matter what. I've seen at least one instance where 2880 is better.

One more time, it's a complex issue, and YMMV.

[tim atherton]

darn - souldn't read the list so late at night - read Imagepritn for Studioprint... when I know full well it's Studioprint that runs with the Peizo stuff!

[chris jordan]

I still use the Epson drivers and Bill Atkinson's profiles for color, but I've seen some amaaaaaaazing B&W prints with some of the new RIP's you guys are talking about; far better than what Bill's profiles can create. My experience generally with the newer Epsons is that they can print everything that's in my file-- film grain and all.

I'm still trying to figure out "YMMV". Yo Mama Moves Vaccuums?

[Kirk Gittings]

Great replies, I learned much from the discussion. I had a feeling there would be many points of view.

I'm one of those guys who takes a 8x loupe to an exhibition. Such as the "Ansel at 100" when it was at the Art Institute of Chicago. I made the guards very nervous. I like prints to be beautiful on a micro level also and many of Ansel's are.

I also went to an exhibition of another friend of mine when I was considering getting into digital darkroom. He had a show of platinum prints from digitally enlarged Imagesetter negatives in Santa Fe. I took my loupe to that show too (much to his chagrin) and I hated the "offset" look to the dot pattern that I saw. I ruled out Imagesetter negs. at that point.

I believe (and maybe I am nuts) that I can see a slight difference in the 2880 dpi prints. I use a Epson 4000 with Imageprint and am currently experimenting with the Ultrsmooth fine Art paper. They seem slightly richer to me at 2880 like a silver print made on a paper with a hair more silver in it or the difference in developing a silver print for 5 minutes instead of four. They are slightly deeper and richer. If only I can see it so be it. I've always believed that an artists standards should exceed their clients. That way you only have to worry about your own expectations.

Then again...........

[JanE]

May I use the opportunity to ask a possibly silly question; Where exact do you see the differences between, say Bill Atkinson profiles and the Epson own profiles?
I've tried both, and get decent results with both, depending on your monitor settings.

The big question is, which way gives the most out in tonal separation for true photographic work?

Thank's, Jan, Finland.

[julian_4860]

One of the problems with BW digi printing is that there is no ICC standard, so you can't profile. All you can do is linearise the rip so that the inks lays down in a uniform manner. If you use a rip to get a fair comparison, you need to linearise the inks for each resolution AND make a new softproof for each res too. Differences in tonality could well be IMO the need for a new softproof. And you need to do this for each paper too. One of the benefits of the latest Piezo (the one that doesn't need a rip) is that Jon Cone has come some way to devising a 'standard'

[Glenn Kroeger]

Chris:

Your mileage may vary.

[paulr]

"Analog prints can't be any sharper than your original film; in fact, they can never be made as sharp as the original film because of the inherent character of the optical enlargement process. "

this might be academic, since I don't know anyone with the patience to put it use, but unsharp masking is (originally) an analog process.
so it's not that you can't sharpen analag; it's just that you probably won't.

[Kirk Gittings]

I just did some test prints on premium luster paper with the ImagePrint to look at this issue, one printed 1440 4 pass HS, one 1440 8 pass HS and one 2880 HS through image print all with the balck point set at 50%. While it is true that the dot differnce is all but invisible with all of them, there is a distinct difference in the look of all of them. As someone said before they each lay down ink very differently. The 1440 4 pass seems to have the deepest blacks, then the 8pass is abit thinner in the blacks but smoother tone overall, then the 2880 which has the most open and lightest shadows and incredibly smooth tone overall, but a "thin" feel to the inks.

The one I now like the best with this print at least is 1440 8 pass with the black point setting at 0 to deepen the blacks a bit.