Sharpness and Aperture Selection on Full-Frame DSLRs

(Note: This article was originally posted in 2007 and I should probably update the test using newer gear – though the point of the test and the post remains.)

Last month I wrote about a set of tests (“Full Frame Lens Test“) that I conducted with my lenses and my Canon 5D body. My object was simply to better understand how the camera/lens combinations would behave so that I could make better decisions about appropriate lenses and apertures while making photographs.

One discovery was that, compared to using a crop sensor DSLR, I can get excellent results when I shoot at smaller apertures with good lenses on the full frame body. I tended to avoid apertures smaller than about f/8 on the crop sensor camera, but there seems to be little or no real liability in using f/11 or even f/16 on full frame.

To illustrate I put together the following composite image. (The image appears in reduced form on this page. Click the link to see the full size version.)

Diffraction Blur Test Image
A sequence of text photographs illustrating diffraction blur at several apertures on a full frame camera

The example includes five versions of a small section from near the center of a photograph taken with the Canon EOS 5D using the EF 24-105mm L IS lens at a 50mm focal length. The camera was on a tripod, MLU and a remote release were used, and the AF was turned off. The images are 100% crops – in other words, actual pixel size is displayed in these tiny excepts from the much larger original images. (You would virtually never view a print at this magnification. These are equivalent to tiny sections from a print that might be about 5 feet wide!) The images have been slightly sharpened in post-processing, but are otherwise unaltered.

I shot at apertures of f/4, f/5.6, f/8, f/11, and f/16. In terms of the sharpness of this portion of the image, I am quite certain that all five examples are plenty sharp for making prints. That said, there are some differences. To my eye:

  • The f/4 and, to some extent, the f/5.6 versions are slightly but noticeably softer at this magnification.
  • The f/8 and f/11 versions seem to me to have approximately equal sharpness. Some parts of the f/8 image seem slightly sharper, but other parts of the f/11 image seem sharper. In the end they are pretty darn equivalent, though I’d maybe give the f/11 a very slight edge overall.
  • The f/16 image may be slightly less sharp than the f/8 and f/11 versions, but the difference would not be noticeable in a print, even a rather large one. In any case, f/16 appears sharper than either f/4 or f/5.6.

After doing this test I no longer hesitate to shoot at f/11 or f/16. Not only does this give me the possibility of getting greater depth of field when I need it, but it also means that I can compensate for corner softness on some lenses (e.g. the 17-40mm) by using a smaller aperture without fear of losing center sharpness.

(Addition: 4/23/07 – Other Canon L lenses seem to give similar results, including my 17-40mm f/4 L and my 70-200mm f/4 L.)

Added 2/23/08:

In response to a question in a photo forum I put together a sample image showing corner sharpness from the same original images used in the example above. (The earlier example shows 100% crops from near the center of the frame.)

(image temporarily unavailable) Canon 24-105mm f/4 L IS lens corner sharpness test

Technical info: Shot using a Canon 5D with the Canon EF 24-105mm f/4 L IS lens at a 50mm focal length. In aperture priority mode, the aperture was changed manually between shots. Initial focus was with AF, which was then switched off before shooting the series. Camera was on a tripod and MLU and remote release were used. Shots were converted from RAW with ACR and no additional post-processing applied. Print made at this resolution would be approximately five feet wide. The crop is from the far lower left corner of the frame.

In addition to noting the softer image in the corner at f/4, also note that the image is a bit darker due to the expected increase in corner light fall-off (“vignetting”) at the largest aperture. Sharpest version in this series shot with a FF body seems to be at f/11 as in the center crop example above. But note that f/8, f/11, and f/16 are not very different in overall sharpness – and in the end any of these apertures would produce a very sharp print.

In response to another forum discussion, I have added another example, this time using the Canon EF 50mm f/1.4 lens and showing performance at f/5.6, f/8, f/11, and f/16. In other respects the test is the same as described already in this post. This image is seen to the right and below.

(image temporarily unavailable)

Conclusions: Based on this set of images and other taken with different lenses under similar conditions, I have come to some conclusions that work for me with my Canon lenses and full-frame Canon 5D body.

  • In general the sharpest whole aperture seems to be around f/11.
  • It is very difficult to distinguish any resolution differences at f/8 or f/16 – there are subtle differences when viewing the test images at 100% magnification on my monitor but these are essentially invisible in prints.
  • f/5.6 or f/22 will tend to be a bit less sharp, though perhaps not for the same reasons. At f/5.6 I begin to notice a bit more of the diminished sharpness as a lens is opened up – more on some lenses than on others. At f/22 the effects of diffraction become just a bit more noticeable. However, if the shot demands it I do not hesitate (much) to use either of these apertures as the very slight decrease in sharpness is quite tiny if visible at all in a print and both provide some other advantages in certain situations. (I’ll even use the largest f/4 aperture on the test lens when isolating the subject is important or when low light demands it – and the results will typically be just fine.)
  • At larger apertures the performance becomes more tied to the particular lens so it is more difficult to make any generalizations beyond the fact that vignetting increases and sharpness will be less optimal.
  • The smaller apertures decrease any corner light fall-off (“vignetting”) or softness, generally to a point where both are insignificant.
  • With all of this in mind, unless I have a reason to select some other aperture I typically use f/11 as my general starting point when shooting with my full-frame DSLR body.

G Dan Mitchell is a California photographer whose subjects include the Pacific coast, redwood forests, central California oak/grasslands, the Sierra Nevada, California deserts, urban landscapes, night photography, and more.
Blog | About | Flickr | Twitter | FacebookGoogle+ | 500px.com | LinkedIn | Email

Text, photographs, and other media are © Copyright G Dan Mitchell (or others when indicated) and are not in the public domain and may not be used on websites, blogs, or in other media without advance permission from G Dan Mitchell.

8 thoughts on “Sharpness and Aperture Selection on Full-Frame DSLRs”

  1. Someone asked about how to interpret this for a cropped sensor camera since these photographs were made with a full frame camera.

    Roughly speaking the results from a particular aperture in this full-frame test would be similar at an aperture about two stops larger on a 1.6x cropped sensor body. In other words, what happens here at f/16 is close to what would happen on your cropped sensor camera at close to f/8. (Technically it would happen at an aperture just a bit smaller than f/8.)

    Dan

  2. You are welcome, Steve. I do these little tests as a matter of course with new equipment. The goal is to understand the performance characteristics of the gear a bit better right at the start. In this case the results informed my decisions about aperture selection in some unanticipated ways.

    Dan

  3. Thanks for your time in putting this together Dan.

    I’m looking at going full frame in the next few months,and looking at canon.

    All the best.

    Steve.

  4. SNoyes:

    I’ll agree that noise reduction and other forms of post-processing can improve an image that is not quite optimum, and I’ll also agree that image “flaws” that may be visible on the screen an 100% magnification may well be immaterial in a print, especially if the print is not overly large. For example, I can certainly imagine a situation in which the diffraction that will be present at f/18 on a crop sensor camera might not be visible in the final image – in fact, it is quite likely that it wouldn’t be visible in an image reduced down to an online jpg.

    I’m puzzled by the “increased noise” report though, unless your sharpening method is one that acts on noise as well as other image elements.

    By the way, I did do some post processing on the images shown in my test. My position is that the most meaningful results are those the represent how we actually make our photographs. With that in mind I felt that applying a general set of sharpening processes to the images – the same as I would do with a “real” image – would give the best result in terms of actual practice.

    I’d be interested in hearing more about your testing methodology.

    Dan

  5. Interesting test Dan. I have found, however, on a crop (40D) sensor I can typically PP shots of up to f/18 or so and get back almost all of my “lost” detail and sharpness due to diffraction. This also comes at a cost of increased noise. So while I agree with people that images will progressively get softer as the aperture is closed down, proper PP must also be taken into account in these comparisons.

  6. Hi Rob, and thanks for posting.

    I think you are right to generally limit yourself to f/8 on crop since the diffraction sets in a bit “sooner” (if by “sooner” we mean at larger apertures) on the smaller format. In my example, for all intents and purposes, f/8, f/11, and f/16 would look essentially identical in a print, and even the “softer” f/4 and f/5.6 examples would look quite good.

    I’m sure that you could push to f/11 on crop in some situations, in the same way that I’ll sometimes push to f/22 on my 5D. Although there is a tiny bit more diffraction softeness at these smaller apertures, it turns out to be pretty minor in an actual print and sometimes the greater DOF (or the slower shutter speed!) can be more important than squeezing out that last bit of “sharpness.”

    Dan

  7. Interesting information. The f11 shot looks great here and I wouldn’t hesitate to shoot it on the 5D. I tend to limit to f8 on my 1.6 crop. Thanks for the test.

Join the discussion — leave a comment or question. (Comments are moderated and may not appear immediately.)

This site uses Akismet to reduce spam. Learn how your comment data is processed.