I was just thinking that the resolution of a print doesn't seem to be discussed very much.

Looking at my Harman FB AL prints, they definitely seem a bit sharper than other prints I've made in the past on these heavier style papers. I haven't done a comparison with say Epson premium glossy, though this might be instructive (later).

Given that the surface structure of this new paper is different to previous papers (according to the specs) I guess this has an impact on how ink is laid down, and the final apparent resolution to the viewer.

This only bothers me because of an interest in LF photography. Silver process contact prints possess an incredible level of detail. Will this ever be achieved by inkjet printing.

Has anyone quantified this? Is inkjet printing inherently limited in terms of resolution. Does anyone care?

I'd like to see more of the information from my 100Mb files visible in a 10x8 print. Perhaps this is only a vain hope though.

I'll let others with greater expertise comment on the resolution capabilities of inkjets.

Instead I'll bring to your attention the work of Clyde Butcher. He shoots with large format cameras and prints both chemically and with a large format inkjet.

Assumed 8-bit RGB, this corresponds to 645 dpi. The native resolution of many inkjets is 360dpi (I have no experience with them).

The most expensive and best silver halide printers (Oce, Noritsu, Frontier) work with 300dpi. This is so good, that I can view prints with magnifier and see details perfectly (I bring TIFFs with 300dpi to the photo station).

Which printer did you have in mind, when referring to silver and the 100MB files?

Off the top of my head, I'd say a 100MB file is pretty much wasted on an 8 by 10. Assuming you print at 300 dpi a 10 inch print needs 3000 pixels by 2400 pixels. Even at 16 bits, this is much smaller file. I don't think you'll see all your detail until you get into significantly larger prints.
Bill

Hi!

I've been wondering this resolution problem also. Canon's printer drivers are 600dpi (at least the smaller models) although this may not have much to do with how the ink droplets are laid on the paper. I've have not done any testing comparing different printers but with a small Canon consumer printer (Pixma iP5300, which Canon claims to offer theoretical 9600x2400dpi) the best results so far have been when feeding it native 600ppi files from 33Mpix mfdb. Still I think those prints have some kind of "powdery" appearance that probably has to do with how the individual ink droplets are placed and the fact that this printer only uses 4 inks and as such is far from creating a convincing illusion of continuous tone.
Proper sharpening is crucial and it depends very much on the image content. 600ppi native files from a sharp lens and digital back reproduced in a small (8x12") print need almost no sharpening and as such print very naturally. As a comparison 11Mpix 1Ds files in 300ppi have to be sharpened quite substantially to have similar appearance and I've not got that last bit of crispness and datailing that larger files at 600ppi have. Maybe it's my bad technique or an illusion. Unfortunately I'm not able to make comparisons with good contact prints regarding the detailing and sharpness The tone reproduction in BW contact print is from another universum compared to my cheapo printer.
Anyway, I would love to see some side by side comparisons between the state-of-the-art inkjet prints and large contact prints.

Cheers,
J

(Deleted by the author, because irrelevant)

This subject is discussed quite extensively in Harald Johnson's "Mastering Digital Printing" Second Edition. I won't try to summarize it here. As well, the subject is discussed in the LLVJ Camera to Print tutorial.

Based on some personal testing, observation, discussion with specialists and reading broadly on this topic I'll suggest the following. First it's important to get the definitions right. Resolution has a very specific meaning - it is defined as pixels per inch or per cm. Apparent sharpness of an image is another matter altogether - affected by quite a number of variables from the capture, through image processing to the final print. Resolution is one of those factors, but above 360PPI in the image file, not a determinative one. The quality of capture (raw vs jpeg, focus, lens quality e.g. absence of chromatic aberation and distortions) the use of sharpening, sharpening techniques, the use of mid-tone contrast techniques, any resampling done and the maner of resampling, the paper surface, the manner in which the printer driver rastorizes the file data and how it lays down ink dots will all affect apparent sharpness. The number of inks would affect the quality of tonal transitions, colour gamut and colour rendition, but not necessarily apparent image sharpness.

One cannot make a general presumption that a single factor such as the printer having a native resolution of 600 PPI will make for sharper-looking images than a printer with 360 PPI native resolution. These numbers are the resolution values at which the printer drivers rastorize image data. It is most unusual that the human eye would see any significant difference of image sharpness north of 360 PPI; as well so many other factors as mentioned above are likely to make a much more visible difference to outcomes.

With a very good, in focus digital capture and good image processing and printing technique using quality equipment from capture to print one is able to achieve an extremely high level of image detail.

From my experience and many others I know that print for a living or create images daily, we all agree that anything over 360ppi is wasted information. (both on Epson and HP printers) The images run the gamut from med. format digital backs and digital slrs to scanned 6x7/ 8x10" transparencies. The print sizes they have tested also run the gamut from 8x10" to 60"x 90". If anything, most of us tend to use 360 ppi for 20x30 and smaller, while using 270-300ppi for larger sizes. This makes sense since viewing distance changes with print size. I've heard of extreme cases where 600ppi for a scanned 8x10 tranny printed smaller than 16x20 actually helps, but according to them it's very slight. One thing to note, all these folks are printing color images on glossy or satin papers, B&W may require different resolutions. Also, many of the people using Epsons have used rip devices or software.

This info by no means is large enough of a sample to be scientifically significant but I have yet to see an image that benefitted from 400ppi or more of info. There was some anecdotal info for a few years on Epson printers, that one should use a factor of 9 for resolution (270, 360ppi, etc.) I've tried these and other resolutions that didn't meet this criteria and haven't seen a difference. Others say they thought there was a difference, but again, ever so slight. On my z3100 I tried prints at 270, 300 and 360 (600ppi native printer) to see if 300ppi would be better since it is exactly 1/2 the native ppi. Again, it didn't seem to matter.

I've also read over the years and don't know if it still holds true but native digital camera files (vesus scans) can use lower resolutions because the files aren't second generation like film scans. In other words, you don't need 300ppi to get the same quality as a 300ppi film scanned image. I don't have any personal experience that could validate this.

- Doug

Ctein explained a couple of years ago why digital capture tends to look better than scanned film of the same nominal resolution. (Ctein is a quirky but incredibly knowlegdeable printer with expertise ranging from dye transfer to digital. His book Post Exposure is simply brilliant in its discussion of perceptual and contrast-control principles underlying printing.) It's not so much the "second generation" issue. Digital capture has creamy smooth tonality due to its underlying image structure, while scanned film is limited in its tonal smoothness by the granularity of film grain. Even when nominal resolution of a film image is higher than a corresponding digital capture, the film grain interferes with sharpness and (especially) tonal smoothness, often just below the level of conscious perception of what is going on.

The biggest advantage of large format film over smaller formats is not so much resolution per se, but the smooth tonality provided by the more modest degree of enlargement required for a given print size- the film grain isn't enlarged enough to become objectionable. Digital capture gains you a large fraction of that advantage because of its inherent tonal smoothness. This is why a high resolution D-SLR like an Eos-1Ds II gives you an image file that looks at least as good as a scanned 645 or even 6x7 cm film image, despite nominally lower resolution.

While I agree on this you may also find that stretching it to the ultimate smoothness like what happens in noiseless, rasterised vector computer designs the printer will produce banding in gradations. Then very little noise added produces much better gradations. The noise itself not really visible but the image quality in general much improved. This will be less a problem with new higher bit workflows up to the printer's head but right now it shouldn't be forgotten as a remedy for that kind of banding.

I general noise seems less a problem in the actual print than on the monitor. Both of digital and analogue origin.

There is another pro for having some grain, I rather have analogue grain in a blow up where not enough data is justifying that size than doing the same from similar small digital data with little noise. It could be the taste for grain that we developed in almost two centuries that is the factor in this case. Especially with B&W this grain is nicer than the artefacts of extrapolation and sharpening a small digital file. Combined, like with scanned MF frames blown up, it will disguise the digital artefacts and give some textural hold for the eye where absolute smoothness would show the lack of data or more sharpening shows the digital artefacts. Maybe we will adapt our taste to new artefacts, mobile phone takes will do part of that job I guess. Reminds me of that endless zooming in on a digital picture by Harrison Ford in Blade Runner, I find it hard to believe that we ever will get normal image technology that goes down to the grain of atomic particles.

So at the two ends of noiseless image data, plenty data and not enough, you may find that some noise may help. I still agree with what you have written above.

Back to the resolution of inkjet prints: there's enough resolution in paper coatings and the modern 300-600-1200 PPI or 360-720 PPI native resolution printers that even a young eye will not see a loss of detail if there was enough data in the image file in the first place. Not only that but the digital control of that data up to the print has been improved much if compared to analogue printing. Any choice of other media with less resolution like matte paper up to coarse canvas is a choice that is based on other aspects but many will still hold much detail. I don't think it is a good idea to focus on resolution figures for gloss papers if many are already beyond the resolution of the eye or on resolution figures for media that wasn't selected to get optimum image quality in the first place. Please, no pixel peeping on inkjet papers, there are many things more important like gloss differential etc.


Ernst Dinkla

try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

My guess is several things are going on that have nothing to do with file size=quality.
First, a native sized file (non-interpolated) should always print better assuming you not enlarging it much, since uprezzing or downsizing files will not look quite as good (especially uprez). The "From Camera to Print" videos available on this site cover this issue. They basically said, keep the file size native if possible for the best output unless sizing it up quite a bit (like 150% or more). For example, if your file at 8x10 was only 580ppi, they would say to leave it not round it up to 600ppi.

Second, a 33mg mfdb will outdo a 11mg 1ds file most any day if done with right lighting/asa, etc. Larger sensor, probably better lenses, etc.

I wouldn't be surprised if you downsampled your 33mg files to 300-360ppi and they looked as good as the 600ppi even with the resampling of info involved in that process. Bicubic smoother works well for sizing up but not sure best method to downsize or if it matters.
- Doug

Ernst hits the nail straight on as usual. Screening resolutions are outside the scope of the actual dot placement precision , nor accuracy. The coatings, and media surface are the determining factors of the actual potential of print resolution beyond the driver/writing system specs.
You can however print resolution charts to find ideal horizontal/vertical sweet spots and the effects of composite ink splats will have on res lines.

I say this as you will be surprised of the differences by different printers , all of which exploit some form or other of Micro weave, or HP's PhotoRhet. Both employ the power of edge diffusion which are nice in the way of gamut but hard to keep out of the way of resolution charts.
I have some res charts that Mark Enders sent me if you like.

In addition to Neil's offer of resolution charts, here is a link to a page with Mike Chaney's test files at his Qimage site.

Qimage Print Quality Challenge

With these files, you can test the differences on your own printer.

Mike's opinion is that in many cases you CAN see a difference in print quality when printing at higher resolutions, say 600 vs 300 ppi.

Of course, subject matter and viewing distance play a large part, with viewing distance perhaps being just as important as resolution. After you print out the test charts, try viewing them at different distances (multiples of 5 feet) and see if the resolution still has an impact.

All that said, Ernst has a good point about the pixel peeping!

Mark

When I need to downsample I use the Adobe-recommended Bicubic Sharper and the resilts seem fine to me - in fact, combined with output sharpening, I sometimes find I should moderately reduce the opacity of the sharpening layer set for optimal results.