300 DPI seems to be a gold standard for target print resolution. When a new camera comes out everyone gets out their calculator and computes how large a print it could make at its native sensor resolution at 300 DPI.

My wife is a classical musician and always has others over for practice sessions so they provide a steady stream of victims for my printing tests. I've done experiments where I've printed special test patterns at different resolutions (and verified under magnification that they printed cleanly on my Epson 2200). I've also printed regular photos at different resolution.

When I've tested these on people almost no-one could resolve 300 DPI content, and the ones that could were people like me with myopia with their glasses off, putting their nose up to the picture. (We myopics have eyeballs stretched in the depth axis, and this functions like an extension tube on a lens. So we lose our distance vision but we have unusually close "macro" vision)

When I've tested regular photos, at 4x6, if I ask people to hold them at whatever they regarded as a comfortable distance most people see no difference between 250 and 300 DPI. With bigger photos, e.g., 10x15 people look at them farther away and claim to see no difference between 200 DPI and 300 DPI. As a result I've had no problem making gallery-quality 12x18 prints using simple PS CS up-rez'ing from images taken on my old (6MP) D100.

As prints get bigger people look at them from ever greater distance. I recently bought a D300 and I'm floored by the resolution and low-noise of its 12MP sensor, and so given the above I'm wondering if I'll ever need another camera. Put another way, is there some point where we have "enough" megapixels that we can make any size print because as the prints get bigger people just stand farther away? Instead of using DPI should we be using some sort of DPD, Dots-Per-Degree of the viewer's visual field?

When I process RAW files, I use a 360ppi setting and then when I resize for print, I just allow the ppi to fall where may, not allowing resampling to occur unless my ppi range falls outside of a 180-460 range.

300ppi at the printing dimensions is based on most commercial photo printers. Its their native resolution I suppose. I know the lab I use for prints is a 300ppi lab and I get the sharpest prints when I size to this resolution. However some printer have other resolutions such as 400ppi.

Epson uses 360ppi as the native resolution and I've found a significant difference in print sharpness and detail when I use the native resolution in comparison with equivalent prints at higher and lower resolutions. A 400ppi file is noticeable less sharp than a 360ppi file. I shoot and scan 4X5 film so I have no trouble getting 360ppi without any upres. I'm pretty sure that a user would be able to see the difference between a 360ppi file and a 400ppi file in a 12X18 print, at least when they get within a few inches. Away from this one specific resolution, I would say they could not tell the difference between a print from a 240 and 400ppi file. The driver does a really good job of handling the scaling.

That said, when shooting with a DSLR and printing larger than the native resolution would allow, its a toss up between upres with Bicubic in Photoshop (and applying print sharpening) and letting the print driver do it. I don't really have a good conclusion on this yet. The only advantage is that you'll have absolute control over sharpening if you do the upres in photoshop.

And finally:
I've heard that Canon uses a 600ppi native resolution. I don't have a Canon printer to test this. I have no idea what resolution HP uses in their pro printers.

first, regular people have no idea what to look for so even if there are obvious differences to a photographer or a fine art print buyer, they won't see them. second, when you make a really big print (think 3ft by 2ft), people will look at it frlom a distance at first, but then they walk up close to "have a really good look" and see if the photographer really knows what they are doing. 12MP is adequate for what i do and no more. it limits me to16x20 prints with acceptable quality. a 39MP medium format back has been on the back burner for a while and getting the D3X at 24MP is a certainty.

Herb...

plnelson wrote:
When I've tested regular photos, at 4x6, if I ask people to hold them at whatever they regarded as a comfortable distance most people see no difference between 250 and 300 DPI. With bigger photos, e.g., 10x15 people look at them farther away and claim to see no difference between 200 DPI and 300 DPI. As a result I've had no problem making gallery-quality 12x18 prints using simple PS CS up-rez'ing from images taken on my old (6MP) D100.

As prints get bigger people look at them from ever greater distance. I recently bought a D300 and I'm floored by the resolution and low-noise of its 12MP sensor, and so given the above I'm wondering if I'll ever need another camera. Put another way, is there some point where we have "enough" megapixels that we can make any size print because as the prints get bigger people just stand farther away? Instead of using DPI should we be using some sort of DPD, Dots-Per-Degree of the viewer's visual field?

first, regular people have no idea what to look for so even if there are obvious differences to a photographer or a fine art print buyer, they won't see them. second, when you make a really big print (think 3ft by 2ft), people will look at it frlom a distance at first, but then they walk up close to "have a really good look" and see if the photographer really knows what they are doing.

I find this to be true. While people will view a painting or a poster from a distance, they have a tendency to move as close to a photograph as physically possible.

Although that's less true of portraits, it's extremely true of landscapes. People tend to expect a landscape to reveal more and more detail as they move closer to it. People are satisfied with the sharpness of a portrait if the facial hair is resolved, but the audience's resolution expectation of a landscape is effectively unlimited.

For that reason, I always judge the sharpness of my prints at reading distance, regardless of size.

That "300 dpi in an 8x10 print viewed at normal reading distance" target figure is much older than digital imaging. It was a reality with film and it's a reality in lithography and other printing. Obviously it's not exact--it's a "human norm," but it's what manufacturers use to determine the specifications of autofocusing systems and lens formulae.

my ipf5000 native DPI for highest quality photo printing is 16 bit 600 dpi.

Actually it is 300 PPI (pixels-per-inch) not dots per inch with respect to output resolution.

The 300 ppi standard has its roots in the mid-1980s when laserprinter and laser imaging scanners were developed and is based in perception / resolution of B&W images, not color. It was found that an oversampling rate of 1.7 to 2.0 to 1 of pixels to halftone printing dots was needed to have a B&W laser printed image perceived as continuous tone. 150 dot-per-inch offset printing was standard at the time for coated papers, hence the 150 dpi x 2 = 300 ppi convention.

I recently explained this in more detail here: http://www.fredmiranda.com/forum/topic/645943/0#5740715

Color reproduction is more complex in than B&W in that the full range of color is produce from the physical process of the white light passing through the transparent inks and being filtered, then having that single color reflect off the white surface of the paper, back through the ink layer to to eye where the color is perceived. Dots of the cyan, magenta, and yellow inks overlap to produce the intermediate colors and the black is used to compensate for the cross-contamination of colors in the inks which result in 100% C+ 100% M + 100% Y printing as a dirty brown instead of pure black. The upshot of all that in terms of resolution is the fact the overlapping of the four (and now as many as 12) overlapping color dots when printing can make color images perceived as sharper at lower resolutions than a B&W: it's a quirk of human perception which has a hair-trigger for contrast differences. There is more contrasts between individual black dots and white paper than in a mish-mash of overlapping color dots.

There are many variables affecting how an image will be perceived at various output resolutions: print size, viewing distance, lighting conditions, and output device. Your approach of testing different output resolutions is good empirical way to determine the optimum resolution, but the tests results will differ if any of the variables change. Also keep in mind that regardless of the output resolution you select for the ink-jet printing the printhead which deposits the dots has nozzles with fixed physical spacing which is forming the dots. The same is true for photo-based commercial printers and dye-subs.

The critical factor is whether or not the original pixel dimensions of camera capture need to be changed to make the print. Camera pixels = original detail and anytime a file in up- or down-sized there will be some loss of original image content due to averaging of pixel values. A 3000 x 2400 pixel camera file is needed for an 10 x 8 print at 300 ppi. That's 7.2 million pixels. So any camera over 8 MP can easily print an 8 x 10 without resampling. But what if you want to print a 6x4? At 300 ppi that is a 1800 x 1200 file? The original must be down-sampled about 60% to print at 300ppi.

To make a 16 x 20 print the same file needs to be up sampled. Is it better to resize 6000 x 4800 to print at 300 ppi, keep the original file size at 3000 x 2400 and print at 150 ppi, or spit the difference printing at 240ppi? That's something which can only be determined by testing and direct comparison at the anticipated viewing distance and lighting.

The best baseline for making visual comparisons would be to start with a file out of the camera printed at whatever size the native resolution of the printer (e.g. 300ppi pitch of the print head) will produce. Then determine what is the optimum viewing distance where that print is accepted by your eye as continuous tone.

Chuck

i was at a friend's show this weekend and most of his work is 100 megapixels and up. his latest work tends to be at 1-2 gigapixels and printed suitably large.

Herb...

RDKirk wrote:
I find this to be true. While people will view a painting or a poster from a distance, they have a tendency to move as close to a photograph as physically possible.

Although that's less true of portraits, it's extremely true of landscapes. People tend to expect a landscape to reveal more and more detail as they move closer to it. People are satisfied with the sharpness of a portrait if the facial hair is resolved, but the audience's resolution expectation of a landscape is effectively unlimited.

I work with a Canon iPF5000, did an interesting test, had a scanned TriX neg at 3200ppi, the imae is a little grainy. I upsampled it to 16x20 at 180ppi and 16x20 at 300ppi. At 180ppi the grain structure in the image was awful and mushy, grains clumped etc. At 300ppi I got the TriX looking grain structure and the image looked a lot better.
I generally print 4x6's at 600 ppi but when I have printed 8x10 and larger, even at 16 bit (really 10 bit) and 600 ppi the images take a lot of looking to see a difference between them and 8 bit x 300ppi. I do scan and process RAW files in 16 bit and drop to 8 bit as the last function.

Tom

The 300 dpi standard carried over into the photographic world in the 90's with the development of continuous tone digital photo printers using dye-sublimation technology. This continued with the development of LED and laser based printers designed for chemically processed silver halide photo papers. All based on the idea that 300 dpi is above the threshold for the human eye to discern the individual dots, giving the appearance of continuous toned images. Output to these printers is 1 to 1 pixel relationship. Each 300 pixels yields 1 inch on the print. If you need to vary the size you must resize the image - the 300dpi must remain constant. Here the computer is controlling the exact color of each discreet dot on the paper, since each dot can be any individual color within the gamut of the printer. While some drivers can resize the image before printing it, you are better off doing it yourself, where you have more control.

What gets a little confusing is comparing the DPI output of a continuous tone printer such as these (where this 300dpi standard originated in photography) to the DPI settings of inkjet printers. Inkjet printers depend on sophisticated dithering and very high dot densities to achieve their affect. A "240dpi" print from an Epson printer still has far more dots (such as 2880x1440) used to render its image. Even at "180 dpi" many inkjet prints are high quality. Here the printer driver is usually more than capable of rendering the image to a final output size without much regard for the "dpi". There really isn't much need to force your print into a specific DPI for these printers, and some argue it is detrimental, since the print driver has to RIP the image into hundreds of millions of dots anyway. As long as your DPI doesn't fall too low (180 - 200 seems to be the bottom limit) or too high (400 - 480 seems to be the high limit suggested by most), the printer will do a better job than a resize in photoshop to some "artificial" DPI number. The suggested workflow is set your print size without resampling the image, letting the DPI fall where it may, perform appropriate output sharpening, and send the file to the printer for it to size.

If your print falls below or above those limits, what DPI should you use? Here again, DPI is such a misnomer with inkjet printers. Many now advocate a different approach. Rather than trying to force the file into a specific "DPI", use a technique that will guarantee the best quality when resizing .. in this case bicubic at an _even_ increment. So upsize 200% with bicubic smoother or downsize 50% with bicubic sharper. After resizing set your print size without resampling, perform appropriate output sharpening, and send the new file at its new native resolution to the printer. The listed DPI will most likely be some very bizarre number, but again the only real use for that number is to make sure you image falls within a certain parameter.

This sounds eminently reasonable. In the end, though, it would seem to be the empirical result of some very capable users trying to get the most out of the technology. So there is a chance for error, or for missing a way to make it even better due to lack of knowledge (read, disclosure) on what the manufacturers are putting in their hardware and their drivers.

Is there any guidance from the printer manufacturers on which workflow will work best?

Wayne Fox wrote:
The 300 dpi standard carried over into the photographic world in the 90's with the development of continuous tone digital photo printers using dye-sublimation technology.


Actually, it goes back even further than that. When mags were run at 150 lpm printing screens, the original drum scanners (those go back to the 1960's) were set to "double" the most popular printing lpm to account for any loss. "300" was an arbitrary guess that stuck forever. Sort of like the dimensions of a 35mm frame. That was created in the late 1920's to make use of all the cimema film available, and the standard cimema frame (18mm high x 24mm wide) was doubled for the first 35mm camera.

In a practical sense nowadays, different outputs really need different file ppi's. For general offset work, the 300 is simply an anachranism. Digital files are cleaner, better, and 1.2x~1.4x the final printing lpi will work just fine. For Epson printing, the larger the print, the further away it will be viewed, hence the lower file ppi you can get away with. For my work, which is printed anywhere from 8x10 to 17x22, I can print with files as low as 180ppi with fine results. Some folks say they really can tell the difference in prints between files sent in at 300 and 360, but that is only with a high magnification loupe.

jjlphoto wrote:
Wayne Fox wrote:
The 300 dpi standard carried over into the photographic world in the 90's with the development of continuous tone digital photo printers using dye-sublimation technology.


Actually, it goes back even further than that. When mags were run at 150 lpm printing screens, the original drum scanners (those go back to the 1960's) were set to "double" the most popular printing lpm to account for any loss. "300" was an arbitrary guess that stuck forever. Sort of like the dimensions of a 35mm frame. That was created in the late 1920's to make use of all the cimema film available, and the standard cimema frame (18mm high x 24mm wide) was doubled for the first 35mm camera.

In a practical sense nowadays, different outputs really need different file ppi's. For general offset work, the 300 is simply an anachranism. Digital files are cleaner, better, and 1.2x~1.4x the final printing lpi will work just fine. For Epson printing, the larger the print, the further away it will be viewed, hence the lower file ppi you can get away with. For my work, which is printed anywhere from 8x10 to 17x22, I can print with files as low as 180ppi with fine results. Some folks say they really can tell the difference in prints between files sent in at 300 and 360, but that is only with a high magnification loupe.




This is the correct answer. Thanks John!!

tomm101, 'I generally print 4x6's at 600 ppi' . how do you print 4x6 on the ipf5000? my minimum print size is 8 x 10.

Gang up lines of 4 on 17 inch paper, with Photoshop, works well. Cutting them up is a little pain but not bad, I only have CVS anywhere close to my house so I had to figure something out.

Tom