Recent mirrorless cameras are very susceptible to dust, because the sensor is fully exposed when changing the lens. However, many recent mirrorless cameras (Sony, Canon) are full frame.

Is dust less of an issue with larger sensors?

I think this could be the case, because large dust particles won't attach as strongly to the sensor as small dust particles do (contact area is proportional to size squared, whereas mass is proportional to size cubed), and I think large dust particles will be shaken off in an easier manner by the ultrasonic sensor cleaning cycle.

So, the real issue is dust particles that attach strongly to the sensor and won't be shaken off. On a full frame, these smaller dust particles cover less pixels, and with a large aperture, light anyway has an easier way to find its path around the small dust particles.

Googling for "dust particle typical size" finds:

Big particles are considered to be between 2.5 and 10 microns in size, where small particles are 2.5 microns and smaller. 2.0 microns are 50 times smaller than a typical human hair.

A typical full frame pixel is 5.8 microns, whereas a typical crop pixel is 3.8 microns.

  • Why do people automatically assume that if a camera does not have a mirror, it also does not have a shutter? Or at least that the shutter must be open when the camera is turned off? – Michael C Jul 21 '19 at 13:18
  • The "typical" size of a FF pixel is irrelevant if one has a sensor with pixels that are smaller or larger than "typical." – Michael C Jul 21 '19 at 13:31
  • 2
    A typical full frame pixel is 5.8 microns, whereas a typical crop pixel is 3.8 microns. Is that today's numbers or what? I mean today we have full frames that is 61 MPix and 15-ish years ago a full frame was less than 10 MPix. Obviously the size of the pixels must have shrunk to 1/6 of the size in the past 15 years – Andreas Jul 21 '19 at 16:01

You're right: it's less of an issue, and your reasoning is also right.

To see this consider the absurd case: a really huge sensor. Let's imagine a sensor so large that its pixels are a centimetre on a side. To obscure a pixel you would need a bit of 'dust' a centimetre on a side, and as you say the physics of scale is such that an object that large simply won't stick to a vertical surface, not least because that's the kind of scale where gravity matters.

In fact the pixel size doesn't really matter. Imagine an even more absurd sensor: one the same size (say 10m on a side, which would be one megapixel sensor with square-centimetre-sized pixels), but with normal-sized pixels, a sensor which would have of the order of a trillion pixels (4 trillion if they are 5 micron pixels). Now obviously dust would obscure individual pixels on such a sensor, so if you pixel-peeped on it you would see dust. But, assuming you looked at such an image in a reasonable way then a bit of dust would have to obscure a huge numbers of pixels: at least tens of thousands of them. So dust is not a problem for a sensor like this eiter.

Obviously such sensors are absurd, but the scaling question is what matters: as sensors get bigger, the kind of dust which will stick to a sensor doesn't get bigger, so it becomes less important. More dust does get on the sensor (the amount of dust presumably goes as sensor area), but it becomes less visible.

And indeed there is very good evidence for this for another kind of sensor: photographic film. Dust is a problem for film not, usually, when it is exposed but when it is printed: it is a significant problem to keep negatives really clean when you are printing them, and you quite often have the frustrating experience of making a really nice print (on paper which has a significant per-sheet cost!) only to discover obtrusive dust. For 35mm film, this is a real problem, and people have blowers and soft cloths &c &c to try to keep things clean. But one of the delights of printing from large-format film (say 4x5in, which I print from) is that dust is almost a non-problem, because, although there is dust, it is so much less visible with such a large neg.

| improve this answer | |
  • 1) you don't need to fully obscure a sensel to have a visible result, and 2) you can have several dust specks on a single sensel, and the bigger the sensel, the more likely it is... The only thing that would matter is dust "density". – xenoid Jul 21 '19 at 12:49
  • @xenoid: this is only partly right. What actually matters is the variation in dust density over scales which are visible. If every pixel is 4% darker than it should be with some being 5% darker you will not see this. If most pixels are at theiir expected brightness but some are 98% darker, then you will see this if the pixels are large enough in the image. So what matters is the statistical properties of the dust, and it's obvious (well, obvious to a physicist) that these are better for larger sensors for the usual statistical mechanical reasons. – user82065 Jul 21 '19 at 13:38

Personally, I've had to clean FF sensors on my 5D Mark II and 5D Mark III a number of times over the years I've used them. I've never had to clean the sensors on my 7D or 7D Mark II.

The two aforementioned FF cameras have an aggregate of 147,000 shutter actuations.
The two aforementioned APS-C cameras have an aggregate of 153,446 shutter actuations.

The real issue with dust and mirrorless cameras is probably the shorter registration distance between the throat of the lens mount and the sensor, combined with the poor practice of many mirrorless camera manufacturers that leaves the shutter open when the camera is turned off or a lens is removed. There is at least one mirrorless camera that closes the shutter in the "off" state and also closes it when it detects a lens being removed.

| improve this answer | |

Sensor size, pixel density, and dust particle size don't matter much if you expose the sensor to dusty environments.

Is dust less of an issue with larger sensors?

Take two differently sized sensors with the same pixel density. The larger sensor, by virtue of its larger surface area, is more likely to capture dust. In an extreme case, suppose a single dust particle were released in an otherwise dust-free room containing a laptop and table. The particle is more likely to land on the floor, than the table, than the laptop.

Is dust less of an issue with... [sensors with a lower pixel density] ?

For your analysis to work:

  • Dust particles that are the size of a single-pixel would have to be a significant problem on the more dense sensor. However, the Bayer color array and demosaicking may hide single-pixel sized dust particles. So for a dust particle to show up on the more dense sensor, it probably has to involve a 4-9 pixel cluster. Assuming 4- vs 6-micron pixels, the same particle is also capable of affecting 4-9 pixels on the less dense sensor, though perhaps to a lesser extent.

  • Sub-pixel dust particles have to not show up at all in the final image created by the less dense sensor. If a particle shows up, but just isn't noticeable, that doesn't mean you don't have a dust problem. It means you have a dust problem that you haven't noticed yet.

  • The built-in dust removal system has to be equally effective in all sensors. Higher-density sensors tend to be newer than lower-density sensors. They may have more effective dust prevention and removal systems, such as electrostatic discharge.

  • The effect of anti-aliasing filters has to be ignored. Anti-aliasing filters may hide the appearance of some particles, while worsening the appearance of others, by spreading them out into a fuzzy blob.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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