How large a print can be made from an uncropped Nikon D700?

Started 5 months ago | Discussions thread
Mark Scott Abeln
Mark Scott Abeln Forum Pro • Posts: 16,182
Large prints

HenrilensUK wrote:

In this case it would be a picture to hang at home, so it needs to be able to be examined from a few inches. There seems to be a fair debate on line with confusion between DPI and PPI. Guys, I came to DPreview for a definitive answer. Thanks if anyone is able to help resolve this.

As others mentioned, viewing the print within a few inches is far beyond what's found even in quality printing. Maybe your eyesight is poor like mine, and without glasses, I can focus as close as 5-½ inches. With glasses, I focus on my computer monitor at arm's length. But even if you plan on examining the print from that close (I define "few" as three), be aware that if your eyesight has good acutance, you'd see the dots that are put on the paper by the printer, and so the illusion of a photograph would disappear.

Daguerrotypes, which were early large format photographs with ISO values of something like 0.05 (!!!), had such fine detail that magnifying glasses were often used to see closer. Generally, film negatives and their contact prints have enough detail for inspection with a loupe, so perhaps you'd rather explore 8x10 inch film photography. But with any given film size, a sensor of the same size can typically make much larger prints with decent quality.

Civilians typically view images from a bit less than arm's length if holding them in their hands, or if wall-mounted, from far enough back so that they can clearly see the entire image. By the second criterion, if you view a print from a distance equal to its diagonal width, no matter the print's size, then according to popular measure for human vision, you need only about one megapixel to make a tolerable 3:2 aspect ratio print, assuming you don't view it any closer.

With your 12 MP camera, you could stand about 1/3rd or 1/4th of the diagonal width away from the print and get a similar visual quality. You could view an 8x10 inch print from three or four inches away and still get good visual quality.

If you are viewing an image from farther away than one diagonal width, then you need even fewer pixels, which is why billboard photography does not demand lots of megapixels. Also, if your print won't be hung in bright lighting, then the resolution demands are even smaller.

For the arm's length criterion, a commonly quoted value is 300 pixels per inch, but please note that this is requirement is for text and line art. Apple Retina screens, which are designed to have pixels invisibly small for the typical viewer, use resolution values of around 225 pixels per inch for viewing from 20 inches, up to about 400 pixels per inch for viewing from 10 inches away. But this is again is for good text quality: photographs need far less resolution, a minimum of 86 pixels per inch when viewing at arm's length.

By the arms-length method at 86 PPI, your D700's 4,256 pixels x 2,832 pixels could go up to 49.5 x 33 inches, and since arm's length is approximately 20 inches, please note that the viewing distance for this print size is about 1/3rd the diagonal width, and so nearly match what I mentioned earlier. You could restrict the print size to 14 x 9.5 inches by the overly-restrictive 300 PPI value, but I don't think this would be necessary for a photograph.

I've read that a 10% pixel count difference is a "just noticeable difference" in images, or more precisely, if a group of people compare two images side-by-side, identical in every way except for the number of pixels, only half will detect a difference if the pixel counts are off by 10%. But if you view an image all by itself, and another image with 10% more pixels much later on, I would guess that virtually no one would be able to reliably identify the higher megapixel image; it would have to be much higher. Don't take the specific 10% value too seriously, but instead consider how people would evaluate an image hanging on the wall: without something much better or much worse to compare it to, they may not appreciate minor quality changes, but only really big ones.

Suppose you can't help yourself, and you still want to pixel peep a huge print; you can get away with even less resolution if the image is processed and printed well. One thing which harms pixel peeping is image artifacts not found in the original scene; certainly you'd want to eliminate fringing due to chromatic aberration, so be sure to use a good lens or at least good software correction.

Noise is a digital artifact that can harm large prints, and so shooting at base ISO and overexposing as much as you can without harming your most significant highlights can help, as well as avoiding lifting shadows. Suppose you can get away with overexposing by a stop (as is often the case with raw capture); you can then lower exposure by a stop for the print and obtain lower noise then if you exposed correctly in the first place.

You can effectively lower ISO to whatever value you want by taking multiple exposures, and then blending them together: for example, ten photos at ISO 100 becomes ISO 10 after blending, and you'll triple your dynamic range. High Dynamic Range photography or HDR can work well by putting more exposure to the shadow areas, or you can use panorama stitching to effectively get a larger sensor, with more megapixels and greater dynamic range as well. Having a clean image should be a main technical goal for making large prints.

In-camera JPEG sharpening, or capture sharpening in raw processors, can be visually distracting if you can actually see the sharpening haloes in a large print. What some print makers suggest is to *not* do this sharpening but rather perform it much later in the workflow, after you know exactly how large you are going to print and how close you are going to view the print. One suggestion is to upsize or resample your image to the exact pixels per inch required by the printer driver (for example, on my printer, that is 360 pixels per inch), so that one pixel on your image maps to exactly one pixel on the printer. Once the image has been resized, then you can add sharpening.

The method used for this upsampling is quite important, and it ought not introduce any noticeable artifacts. The Mitchell interpolation method produces smooth results, as does the similar Lanczos method, found in some advanced software. The common Bilinear method, found in some software, will produce noticeable artifacts. There are some Fractal and Artificial Intelligence (AI) software for resampling that does a good job: they do create fake detail, but often plausible detail.

JPEG compression, especially if high compression values are used, can produce visible artifacts that will punish pixel peeping.

-- hide signature --

Dots per inch or DPI is rarely directly useful for general photography, as it usually describes specific implementations within a printer and its driver software. "DPI" and "PPI" roll off the tongue similarly, so confusion is to be expected, even among those who do know the difference.

Printers come with a set of inks: with cyan, magenta, and yellow being used on the cheapest color printers; most all other ones use black also, and adding other inks are useful for getting a wider gamut of color. The printers do not mix the inks, and they can only send dots of pure ink to the paper; but these are small dots, placed side-by-side with minimal overlap. Using multiple ink dots in close proximity will give you the visual impression of a wide range of colors, but only if you don't look too closely. There are a multitude of ways of patterning these dots to give you a range of colors, and they vary with printer models and driver settings.

So DPI doesn't tell you much, except that you can know that DPI will always be greater than PPI, with the exception of printers which can only print black dots and no grayscale, such as we find with old fax machines. Desktop printers often have a DPI value that is four times the PPI maximum resolution, so each pixel will be printed with a block that is four dots on a side, or at most 16 dots total.


 Mark Scott Abeln's gear list:Mark Scott Abeln's gear list
Nikon D200 Nikon D7000 Nikon D750 Nikon AF-S DX Nikkor 35mm F1.8G Nikon AF Nikkor 50mm f/1.8D +2 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow