Camera Resolution and Maximum Print Size
How many pixels can we actually use?
December 19, 2009

At some point most photographers wonder how many pixels they need in order to print an image file at a given size. Of course there is no hard and fast answer. A print that's perfectly acceptable to one may fall short of someone else's standards. In addition, the combined factors of shooting technique, lens quality, and noise can make huge differences in the quality of the original file, and lower quality means smaller maximum print sizes. With all this said, and assuming image files are of the highest quality that one can get from a DSLR, I'll explain the guidelines I use. Your mileage may vary.
.
Six clean mega-pixels from a 2003 vintage APS-C format DSLR can make 12x18 inch print that's clearly superior to one made from a 4000 dpi scan from 35mm film. Much beyond a 12x18 inch image area film starts having an advantage over a 6 mega-pixel camera because it can resolve more detail. But resolution isn't everything, and the lack of noise in files from a DSLR is a huge advantage. Numbers aside, 12x18 inch prints from the 6 mega-pixel DSLR simply make a better impression on those who view them, including me. With this as a standard it's an easy mathematical exercise to determine what resolution is required to get the same print quality in larger sizes. Full frame sensors and other camera improvements since 2003 push the quality into territory formerly occupied by medium format film. Given enough pixels we ought be able to make stunning 40x60 inch prints from DSLRs in the future, right? Well, not quite.
.
It's all a matter of area. Making something like a 24x36 inch print from a 24x36 millimeter sensor stretches the limits of every part of the image making system to a ridiculous degree, no matter how many pixels the sensor has. The tiny sensor area must be magnified over 645 times to make a print that large. That magnification makes the smallest lens aberrations and every other image defect plainly visible. These issues affect medium format cameras to a much smaller degree because the image and its flaws are magnified only about a third as many times to make a comparably sized print. By virtue of the tiny physical size of a "35mm" format image sensor, today's highest resolution DSLRs already require perfect technique and the best prime lenses (not zooms) used at their sharpest apertures (usually F8 - F11) if full advantage of their resolution is to be realized. But in real world situations it is often impossible to shoot at the sharpest aperture due to depth of field requirements for the image at hand. Using the best zoom lenses at apertures like F16 for depth of field takes a huge toll on image resolution, and hand holding is completely out of the question regardless of the lens and aperture used. Don't forget that just a little wind can often cause a tripod to vibrate almost imperceptibly. The shot might look great at around 11x14 inches but fall apart completely at 20x30 inches. It's easy to see that for any given sensor size there is a point beyond which more pixels become useless. The 35 mm formats (APS-C, APS-H, FX "full frame") are not quite there yet in theoretical terms, but in practical terms I think they are very close indeed.
.
I haven't gotten as far off track as it might seem. The point is that if we can't make practical use of full frame 35mm resolutions beyond roughly 30 or so mega-pixels in the field, it means it's equally hard to make use of APS-C sensor resolutions of beyond about 19 mega-pixels. Personally, I think anyone working outdoors with zoom lenses would be doing very well to make full use of 24 mega-pixels on a regular basis, and that corresponds to 15 mega-pixels in an APS-C image sensor. If you take a look at the latest camera offerings you'll see we're rapidly approaching and even surpassing these numbers. Hopefully camera makers will soon stop their race for more pixels and concentrate on reducing sensor noise (RAW file noise) instead. I have little confidence this will happen because so many confuse pixel count with image quality, and they buy based on that notion.
.
Another thing to keep in mind is that the useful depth of field in an image becomes shallower as the print size becomes larger. That means you need to use smaller apertures, and get less overall image resolution due to diffraction as a side effect. Only tilt-shift lenses can help here, and only in certain situations. For what it's worth, depth of field markings on some prime lenses and most older lenses are calibrated for making an 8x10 inch print! If you use these and try to make a 20x30 inch print you'll be quite disappointed. Lastly, image processing before printing is far more critical when printing larger sizes. This is especially true of sharpening. When printing small format captures at these maximum sizes everything has to be perfect. Digital image capture has removed the constraints imposed by film itself, like grain, but the constraints imposed by the physical size of a given capture format are still present. They always will be, and that's why medium format cameras exist. Digital has given us much higher image quality, but no matter how many pixels one has there are still significant constraints when printing a 35mm digital capture to sizes larger than 16x24. Making larger prints from a 35mm format DSLR capture requires certain image characteristics captured using perfect technique under all of the right circumstances. That does not happen very often, and higher pixel counts will not improve the situation. Only a larger image sensor format can do that.
.
With all that said, when used with the best files you can pull from a full frame (36x24 mm) image sensor, using the guidelines below can help create prints that compare very well with those originating from medium format film. Don't forget that since it takes four times the pixels to double the print size, a couple mega-pixels at the higher resolutions does not make a huge difference.
.
.

Camera Resolution / Print Size Equivalents

Mega-pixels

Max. Print Size FX

APS-C

6

12x18 inches

X X

10.7

16x24 inches

X X

16.7

20x30 inches

X X

24

24x36 inches

X  

.
Happy pixel counting!
.

.
Notes:

These are guidelines, not rules, and the numbers are approximate, not exact. The true maximum size for any image depends on a number of factors, including the image itself. Those lacking fine details that need to appear sharp can probably go larger. Those that require gobs of depth of field may not go as big unless they were shot with a tilt-shift lens. I've found that printing part of an image at its actual size on a smaller piece of the same paper, and then viewing it from the distance I'd view the full sized print helps in determining when quality starts falling off noticeably. It also saves a lot of paper!
.
Best results with most printers are had by up-sizing the image file to the required dimensions at the printer's native resolution, or half of that resolution, in Photoshop before sharpening and printing. Photoshop does a better job of interpolating than most printer drivers. The "bicubic smoother" algorithm in Photoshop generally works best for up-sizing, but other algorithms can be more appropriate in special circumstances. I actually prefer "bicubic" for up-sizing when the amount of up-sizing is not dramatic. This retains more sharpness at the risk of making high contrast diagonals appear "stepped".  
.
Camera makers have been reducing sensor noise and simultaneously nullifying the noise reductions by packing more pixels into their image sensors. This has increased pixel counts while keeping RAW file noise about the same over time. That's fine until pixel counts get beyond what can be used practically. The extremely low noise at high ISO settings from recent cameras is only available in the camera's JPEG files, and is a result of firmware processing of the RAW file inside the camera when the JPEG is created. Since JPEG files have such severe limitations, reducing noise at its source in the image sensor, and consequently in the RAW files, would be very beneficial.

Home | Index

Copyright 2009 Dean M. Chriss
Content may not be reproduced or distributed without written permission.
All Rights Reserved.
 e-mail