4
\$\begingroup\$

Given a sensor size, the diffraction limit (and cut-off) are determined largely (entirely?) by the density of pixels on the sensor. With the trend of ever-increasing megapixel counts, this means that the diffraction limit reduces with each generation of new cameras.

My understanding is, that past the diffraction limit, the resolving power of the optical system (lens+sensor) can no longer increase, and indeed, begins to decrease somewhere around the diffraction cut-off.

So, in general, when shooting at an f-stop past the diffraction limit of a sensor, does increased resolution (pixel density) offer any benefits? I'm mostly interested in answers regarding the optical trade-off between resolution and diffraction, though answers about more practical benefits are also fine.

To put my question another way, why are high-end camera manufacturers continuing to increase the resolution at the cost of high-aperture image quality? Would they not be better served by improving sensor technology (noise-characteristics, for example), while maintaining resolution?

\$\endgroup\$
1

1 Answer 1

6
\$\begingroup\$

The "diffraction limit", i.e. the f-number at which diffraction begins to reduce image sharpness does indeed decrease as pixel density increases (for a fixed size sensor).

However is not true that resolving power cannot increase past the diffraction limit, just that it will increase by ever smaller amounts, in a situation knowns as diminishing returns.

When shooting at an f-stop that is so far past the limit that the increases are negligible then there is no benefit to increasing resolution (for that f-stop), all else being equal. However in the real world things are never equal so there may be advantages to higher pixel counts when it comes to reducing noise.


Many have claimed diffraction will put an end to the "megapixel race". It will, of course, but not as quickly as some realize. We are still seeing real benefits with APS-C sensors containing 24 megapixels. That equates to a full frame sensor with 54MP, far more than the 'diffraction limited' D800 with 36MP. I predict pixel counts to rise past this level, topping out when diffraction is having a noticeable effect at f/5.6, which is probably 100-200MP for a full frame sensor.

Diffraction is still the ultimate limit to detail however, there's really no escape. Moving to medium or large format wont let you bypass the limit - it's tempting to think so given the larger pixels of the equivalent resolution sensor. But with a larger sensor narrower apertures are required to get the same depth of field, and faster lenses are rarer for large formats.

Using a wider aperture is the only solution, but that too runs into problems as it reduces depth of field, causing large amounts of the image to be out of focus which very quickly limits detail.

The only option then is to reduce your focal length. This results in a smaller entrance pupil for the same f-stop, meaning your depth of field is larger for the same diffraction limit. However chasing this parameter eventually leads you into difficulties again, as focal lengths get shorter, maximum apertures start getting smaller again and detail is limited by standard optical aberrations, unless some advance is made in lens design and materials.

Panoramic multi exposure images will therefore be the future of high resolution. By bracketing focus they can escape the depth of field limits, and by using many images taken from longer focal lengths they can escape a lot of the aberrations present in wide angle lenses.

\$\endgroup\$
7
  • \$\begingroup\$ So, assuming otherwise identical sensors, will an image shot on a 20 megapixel sensor at f/22 (as an example) and then downsized to 10 megapixels be identical to an image shot at f/22 on a 10 megapixel sensor? Will the 20 mp shot show more detail at its native size? \$\endgroup\$ Feb 27, 2013 at 17:00
  • 1
    \$\begingroup\$ @ChinmayKanchi it depends on the sensor size (and hence pixel size), 35mm full frame the 20MP sensor at f/22 will have more detail than the 10MP sensor at f/22. But with micro 4/3rds you may not see any difference in detail between 10 and 20 megapixels at f/22 \$\endgroup\$
    – Matt Grum
    Feb 27, 2013 at 17:05
  • \$\begingroup\$ Does this mean that camera manufacturers will finally give up increasing resolution around the 100 mp full frame, when diffraction begins to appear as early as f/5.6? \$\endgroup\$ Feb 27, 2013 at 17:25
  • \$\begingroup\$ @ChinmayKanchi High performance lenses are already peaking at f/5.6 if you look at the tests. However there will become a point where diffraction, camera shake, and the lack of scope for displaying high resolution images all converge. It ought to be possible to extrapolate this limit from 10x8 photography which is the largest format to have any sort of popular appeal. Film is no longer the limit but the physics is the same. \$\endgroup\$
    – Matt Grum
    Feb 27, 2013 at 17:42
  • \$\begingroup\$ @Matt Grum - what about changes in the way sensors operate. I don't fully understand the physics, but couldn't something like the way the Lytro works be adapted to allow a longer depth of field for a wider open aperture by capturing more of the field? This would also make much higher resolution sensors viable as the current state of the tech is using microlenses to break up the light headed in different directions to refocus it on to different dots of the sensor. \$\endgroup\$
    – AJ Henderson
    Feb 27, 2013 at 19:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.