2

Couple of weeks ago, I was assigned a homework in one of my Applied Physics master's course. My professor asked to explain why passive autofocus works only with circular polarizers.

I am not a very experienced photographer(I still haven't bought a DSLR...) but in my compact cameras I don't think that a special polarisation is needed.

Later I thought, that this is because my cameras use active autofocus. I tried to gather as much information as possible on how passive autofocus works, but I haven't found any indication on whether they need circular polarisation.

Is this statement true? Does autofocus require circularly polarised light to work? Why only passive AF systems are polarisation sensitive?

How is the AF system affected by the polarization of the incident light? This has to do with the censor itself?

Thanos
  • 171
  • 4
  • 1
    Despite the title difference, this is the essential question asked by (and answered at) What is the difference between a linear and a circular polarizer? – mattdm May 11 '13 at 13:03
  • Also this, on compact cameras: http://photo.stackexchange.com/q/9488/1943 – mattdm May 11 '13 at 13:14
  • @mattdm: Thank you very much for your comment and help! I am not sure though, if my question is fully answered by your links! It is true that a beam splitter is a "polarisation sensitive" optical item. But I cannot understand why the sensors are polarisation sensitive... What makes the polarisation sensitive? And why only passive AF are polarisation sensitive? – Thanos May 11 '13 at 13:49
  • 1
    it's in some of the answers there, but the quick version is: all mirrors have a (linear) polarization effect, and the AF sensors on SLRs are after the reflex mirror, so if light hits the mirror polarized in just the wrong orientation, it might be entirely filtered out and not reflected to the sensors. AF systems which do not have a mirror in the path are not subject to this issue. – mattdm May 11 '13 at 14:15
  • @mattdm: The thing that I am not aware of a reflection related effect, which affects an already polarised light. Polarisation via reflection affects only one component of the reflected beam(that is the p-component). So, I cannot understand that, how a light can be affected from reflection if it's already linear polarised? – Thanos May 11 '13 at 14:37
  • This might be getting to be better for http://physics.stackexchange.com/. But as I understand it, it's because the mirror is half-silvered and a different amount will be reflected or passed through based on the linear polarization. – mattdm May 11 '13 at 14:41
  • @mattdm: Thank you very much for your time! I've already asked on physics.se but i don't believe to get an aswer... The answer has to do with the censor itself. I am concinnced and sure that the reflection of an already polarized light isn't bad;the linear polarized light itself interacting with the censore is probably the key. Any idea on that? – Thanos May 11 '13 at 15:24
  • @mattdm: From what I know passive AF, works with contrast comparison hence I don't think that an AF system can have hard times with a linear filter. However, the metering system(after which there is the auto-exposure system) is really intensity sensitive. From your expereience, a linear filter can really cause focus trouble or exposure ones? – Thanos May 11 '13 at 18:15

1 Answers1

0

First we need to clear up a confusion caused by photographers too often not having paid attention in physics class. There are two types of polarization, linear and circular.

Linear polarization defines what plane the light is oriented in. It could be horizontal versus vertical, for example. Sky light bouncing off of a horizontal dielectric, like a puddle for example, will be horizontally polarized. This effect depends on angle, and there is actually a angle at which all the light is 100% polarized (all horizontal, none vertical). A linear polarizing filter oriented to pass only the vertical component (the dot product with the vertical direction) will block the horizontally polarized sky reflection off the puddle. With a good polarizer and with the light bouncing off the puddle at just the right angle, it will block almost all of it.

Circular polarization is a separate attribute of light than linear polariation. It is sortof the "handedness" of the light. Right-handed light is blocked by a left-handed circular polarizing filter, and vice versa. These types of polarizing filters are often used in 3D glasses that aren't the red-blue kind. This allows you to not have to have your head perfectly level for each eye to get its intended image. Left-handed light is still blocked by a right-handed polarizer regardless of how much it is turned.

That all made sense until the aforementioned photographers who didn't pay attention in physics class got envolved. As far as I have even seen, photographic polarizers are all of the linear type. This makes sense because they are intended for effects like enhancing or attenuating reflections off of puddles and the like. A circular polarizer wouldn't be useful for that. What is called a "circular polarizer" in photography, is not a circular polarizer as understood by the rest of the world.

So what are photographic "circular polarizers" and why do we have them? Originally, straight linear polarizers were used in photography. They did exactly what they were intended to do, which was messing with puddle reflections, scattered light from the sky, etc. Then along came cameras with built-in light meters. In a SLR, there is nothing between the lens and the film or sensor while the picture is being taken, so the linear polarizers are fine for that. However, to do metering before the exposure usually requires bouncing light off or a mirror. The light also needs to go to the viewfinder, so there are often partially-reflecting mirrors envolved to split the light. Unlike regular silvered mirrors where light bounces off a metal layer, partial mirrors usually envolve light bouncing off of a dielectric somewhere. That light is then linearly polarized. Or put another way, such a partial mirror splits the light in a different ratio depending on polarization angle. That's bad if the exposure meter thinks its seeing a known fraction of the light coming in.

The solution was to add another layer to the polarizing filter on the front of the lens. The front of the filter passes only the portion of the light aligned with whatever orientation you rotate the filter to. A second layer was added that you can think of scrambling the linear polarization. In more technical terms, this is referred to as a "1/4 wave plate", as pointed out in a comment by Mattdm. Now the partial mirror, which is in effect a somewhat polarizing filter, will pass about the same fraction of light in each direction regardless of how the linear polarizer at the front of the lens is rotated.

This situation is so common that "circular polarizers" (actually a linear polarizer with a de-polarizer after as discussed above) are now more common in photography than true pure linear polarizers. In a lot of cases, the "circular polarizers" will be more available and cost less due to volume.

Olin Lathrop
  • 17,402
  • 1
  • 32
  • 68
  • You are mistaken here. The second layer doesn't "sort of scramble" the light; it is a quarter-wave plate. This results in circularly-polarized light — in other words, this is a real circular polarizer. – mattdm May 11 '13 at 13:13
  • @mattdm: I was trying to describe the relevant effect, which is that light no longer has a single linear polarization. I have edited the answer to refer to your comment. – Olin Lathrop May 11 '13 at 13:29
  • @OlinLathrop: Thank you very much for your comment! The thing is though, that I am interesting to understand why passive(and not active) AF systems are "polarisatoin sensitive"... Any help/idea on that would be more than welcomed! – Thanos May 11 '13 at 13:51
  • A better way to say it would be that although it is a circular polarizer, it still has the effect of filtering out light that isn't polarized in a certain linear orientation, which is the effect that is useful in photography. – mattdm May 11 '13 at 14:00
  • @Thanos: "Passive autofocus" (more correctly, phase detect autofocus) involves several mirrors. In "active autofocus" (contrast detect autofocus), the light shines directly on the sensor or film without encountering a mirror. You should now be able to connect the dots... :) – Chinmay Kanchi May 14 '13 at 02:21
  • @ChinmayKanchi: Contrast detect autofocus and Phase Detect autofocus are two methods of passive autofocus. The active autofuc involes the emission of either an ultrasound pulse(old techique) or visible or IR light(modern technique). Measuring the time delay of the pulse we get an estimation of the distance between the lens and the object to be captured. – Thanos May 14 '13 at 06:00
  • @Thanos: Fair enough. However, CDAF should not be affected by a linear polariser, since there are no mirrors in the path of light. If an IR/ultrasound pulse is being transmitted, then I would think that it's pretty obvious why a polarising filter doesn't affect it! – Chinmay Kanchi May 14 '13 at 06:21
  • @ChinmayKanchi: You are right. I believe that both passive AF methods shouldn;t be affected by the filter, because in both methods what is measured is the difference in intensity or position. I believe that auto epxosure is more fragile there! – Thanos May 14 '13 at 06:36
  • @Thanos: PDAF will be affected, because it's working with a fraction of the light available to the sensor in the first place. Anything that cuts down further on that fraction is going to affect PDAF. – Chinmay Kanchi May 14 '13 at 10:09