Are circular polarizers required with mirrorless cameras?

[JohnFrim]

I know what polarizers do, and I know the technical differences between circular and linear polarizers, so no need to explain any of that.

Circular polarizers became a necessity with DSLRs because of issues related to both autofocus and metering systems because these take their information from mirrors inside the camera box. The mirrors actually polarize the light going to the sensors to some extent, so depending on the angle of the linearly polarized light entering the camera (when using a linear filter) the combination of linear filter and focus/metering mirror could cut light intensity much like crossing two polarizers.

Since mirrorless cameras have PDAF sensors embedded right on the imaging sensor (not sure about CDAF) I was wondering if there is no longer a need to use circular polarizers. Linear polarizers are cheaper; alternatively one could spend the same $$ and get a very high quality linear polarizer.

J

 

[le_alain]

Quite clearly I think that just like phase detection is hurt by a linear polarizer on DSLRs, it would also be affected on these cameras - it's just optics.

Can you developp?

the light is not reflected to reach on sensor PDAF in a mirorless (but with a miror in a DSLR),

is there another way for the polarizer to affect phase detect focus ?

 

[jackalopemaui]

Hi John,

contrast-detect AF is obviously fine with linear polarizers. And linear polarizers are more interesting to work with anyway because you can select which direction of the light waves you block - that's much more interesting than a circular polarizer which blocks all directions.

However, I'm sure you're aware that some more recent mirrorless cameras (btw I hate this term - it suggests that somehow the norm would be to have a mirror, but it's not) manage quite decent AF tracking by using both CDAF and some PDAF pixels on the sensor. Quite clearly I think that just like phase detection is hurt by a linear polarizer on DSLRs, it would also be affected on these cameras - it's just optics.

I know what polarizers do, and I know the technical differences between circular and linear polarizers, so no need to explain any of that.

Circular polarizers became a necessity with DSLRs because of issues related to both autofocus and metering systems because these take their information from mirrors inside the camera box. The mirrors actually polarize the light going to the sensors to some extent, so depending on the angle of the linearly polarized light entering the camera (when using a linear filter) the combination of linear filter and focus/metering mirror could cut light intensity much like crossing two polarizers.

Since mirrorless cameras have PDAF sensors embedded right on the imaging sensor (not sure about CDAF) I was wondering if there is no longer a need to use circular polarizers. Linear polarizers are cheaper; alternatively one could spend the same $$ and get a very high quality linear polarizer.

J

A circular polarizer doesn't block light in all orientations. That would make it a ND filter. It's a linear polarizer which blocks light to an extent based on its orientation. Perpendicular light is 100% blocked, 45 degree light is partially blocked, parallel light is 0% blocked, etc. The light coming out of the linear layer is all aligned, and then it goes through a circular polarizer layer which makes it circular polarized

 

[DFPanno]

I have been following several photography forums over the past few years, and while I am often truly amazed by some of the expertise out there, I am more often very entertained by the "expertise" out there. After awhile you get to know who you can trust for opinions and advice.

People are simply trying to help you as best they can.

I generally simply thank everybody for their contribution and quietly go with the answer that seems most correct.

IMO it is a little silly to go to forums in the first place when their are so many on-line resources available from pro photographers.

Incidently - I found expert answers to your poorly crafted question with a couple of keystrokes and I didn't have to ask anybody anything.

 

[Zapirian]

My typo -- should have read Brewster's angle. Wikipedia has quite a lot of technical info about polarization, but they end the article with a very broad comment about cameras that does not address the situation with mirrorless cameras. I think they need to update the page.

Just an FYI, the way Wikipedia works is that everyone can update the information. If you find that you have information that exceeds that of the entry, possibly after researching and discussion in this thread, you can add it yourself (will probably have to be verified by a Wikipedia employed editor though)

 

[antoineb]

light is a wave as you know. Either its normal random light in which case it is a mix of waves with random orientations over 360 degrees (over a circle perpendicular to the direction of observation). Or it is polarized light, where one given direction is privileged while the other directions are reduced to an extent (but not 100% suppressed unlike what you wrote).

As for an ND filter: indeed if you use two linear polarizers and one is free to rotate vs the other then you have made a variable ND filter.

Hi John,

contrast-detect AF is obviously fine with linear polarizers. And linear polarizers are more interesting to work with anyway because you can select which direction of the light waves you block - that's much more interesting than a circular polarizer which blocks all directions.

However, I'm sure you're aware that some more recent mirrorless cameras (btw I hate this term - it suggests that somehow the norm would be to have a mirror, but it's not) manage quite decent AF tracking by using both CDAF and some PDAF pixels on the sensor. Quite clearly I think that just like phase detection is hurt by a linear polarizer on DSLRs, it would also be affected on these cameras - it's just optics.

I know what polarizers do, and I know the technical differences between circular and linear polarizers, so no need to explain any of that.

Circular polarizers became a necessity with DSLRs because of issues related to both autofocus and metering systems because these take their information from mirrors inside the camera box. The mirrors actually polarize the light going to the sensors to some extent, so depending on the angle of the linearly polarized light entering the camera (when using a linear filter) the combination of linear filter and focus/metering mirror could cut light intensity much like crossing two polarizers.

Since mirrorless cameras have PDAF sensors embedded right on the imaging sensor (not sure about CDAF) I was wondering if there is no longer a need to use circular polarizers. Linear polarizers are cheaper; alternatively one could spend the same $$ and get a very high quality linear polarizer.

J

A circular polarizer doesn't block light in all orientations. That would make it a ND filter. It's a linear polarizer which blocks light to an extent based on its orientation. Perpendicular light is 100% blocked, 45 degree light is partially blocked, parallel light is 0% blocked, etc. The light coming out of the linear layer is all aligned, and then it goes through a circular polarizer layer which makes it circular polarized

 

[JohnFrim]

Hi John,

And linear polarizers are more interesting to work with anyway because you can select which direction of the light waves you block - that's much more interesting than a circular polarizer which blocks all directions.

No, that's not correct. Circular polarizers are, in fact, linear polarizers with a quarter wave plate AFTER the linear polarizer in the light path. The quarter wave plate phase shifts the electric field vector from the magnetic vector of the linearly polarized light by 90 degrees, making the resultant composite vector travel like a corkscrew. This corkscrew rotates clockwise or counterclockwise depending on the angle of rotation between the linear element and the quarter wave plate when they are assembled at the factory. Note that this is NOT the angle that you play with when rotating the filter assembly. If you mount the circular polarizer backwards you actually get linearly polarized light coming out, and some CPLs have been sold with the elements mounted backwards!!

Quite clearly I think that just like phase detection is hurt by a linear polarizer on DSLRs, it would also be affected on these cameras - it's just optics.

No, it is not the sensors that are affected by linearly polarized light. It is the interaction between linearly polarized light (when using only a linear polarizer on the camera) and the partially silvered mirrors (which tend to polarize the light going to the sensors) that causes the problem. Because in this situation you effectively have 2 linear polarizers in the light path from the scene to the sensor (PDAF, CDAF, meter, it does not matter) the intensity of light reaching the sensor is affected by the angle of rotation between the 2 polarizing elements. Thus, as you rotate the linear filter on the lens to cut glare you are changing the intensity of light reaching the sensors -- on cameras with mirrors or beam splitters in the light path -- and that can lead to focussing or metering errors. By using a CPL (mounted correctly in the holder) the light entering the lens is circularly polarized, and it will be of the same intensity after it bounces off the mirrors no matter what the angular rotation of the filter. Now, with a mirrorless camera you don't have that second polarizing element in the light path, so a linear filter should work just fine.

I know this sounds like I am answering my own question, but that is in fact the hypothesis on which I was working. I wanted to confirm this hypothesis from folks with knowledge in the subject area, or with hands-on definitive experience.

I think there have been lots of comments and outside references to support my hypothesis.

Thanks to all.

 

[JohnFrim]

light is a wave as you know. Either its normal random light in which case it is a mix of waves with random orientations over 360 degrees (over a circle perpendicular to the direction of observation). Or it is polarized light, where one given direction is privileged while the other directions are reduced to an extent (but not 100% suppressed unlike what you wrote).

As for an ND filter: indeed if you use two linear polarizers and one is free to rotate vs the other then you have made a variable ND filter.

Regarding your subject line that light cannot be circularly polarized, please check Wikipedia or many other sources. Surely you are not in the camp that thinks circular polarizers are called that because they are "round" filters.

Movie glasses use right vs left circular polarizers to present the appropriate images to the eyes for the 3D effect. Linear polarizers could be used as well, but they would be sensitive to the tilt of your head. Circular polarizers work no matter what angle you tilt your head.

 

[120 to 35]

I know this sounds like I am answering my own question, but that is in fact the hypothesis on which I was working. I wanted to confirm this hypothesis from folks with knowledge in the subject area, or with hands-on definitive experience.
I think there have been lots of comments and outside references to support my hypothesis.

Thanks to all.

You can also ask in the Photographic Science and Technology forum on DPReview.

 

[le_alain]

Hi John,

And linear polarizers are more interesting to work with anyway because you can select which direction of the light waves you block - that's much more interesting than a circular polarizer which blocks all directions.

No, that's not correct. Circular polarizers are, in fact, linear polarizers with a quarter wave plate AFTER the linear polarizer in the light path. The quarter wave plate phase shifts the electric field vector from the magnetic vector of the linearly polarized light by 90 degrees, making the resultant composite vector travel like a corkscrew. This corkscrew rotates clockwise or counterclockwise depending on the angle of rotation between the linear element and the quarter wave plate when they are assembled at the factory. Note that this is NOT the angle that you play with when rotating the filter assembly. If you mount the circular polarizer backwards you actually get linearly polarized light coming out, and some CPLs have been sold with the elements mounted backwards!!

Quite clearly I think that just like phase detection is hurt by a linear polarizer on DSLRs, it would also be affected on these cameras - it's just optics.

No, it is not the sensors that are affected by linearly polarized light. It is the interaction between linearly polarized light (when using only a linear polarizer on the camera) and the partially silvered mirrors (which tend to polarize the light going to the sensors) that causes the problem. Because in this situation you effectively have 2 linear polarizers in the light path from the scene to the sensor (PDAF, CDAF, meter, it does not matter) the intensity of light reaching the sensor is affected by the angle of rotation between the 2 polarizing elements. Thus, as you rotate the linear filter on the lens to cut glare you are changing the intensity of light reaching the sensors -- on cameras with mirrors or beam splitters in the light path -- and that can lead to focussing or metering errors. By using a CPL (mounted correctly in the holder) the light entering the lens is circularly polarized, and it will be of the same intensity after it bounces off the mirrors no matter what the angular rotation of the filter. Now, with a mirrorless camera you don't have that second polarizing element in the light path, so a linear filter should work just fine.

I know this sounds like I am answering my own question, but that is in fact the hypothesis on which I was working. I wanted to confirm this hypothesis from folks with knowledge in the subject area, or with hands-on definitive experience.

I think there have been lots of comments and outside references to support my hypothesis.

Thanks to all.

Thanks

 

[teddoman]

I know this sounds like I am answering my own question, but that is in fact the hypothesis on which I was working. I wanted to confirm this hypothesis from folks with knowledge in the subject area, or with hands-on definitive experience.

I think there have been lots of comments and outside references to support my hypothesis.

Thanks to all.

You can also ask in the Photographic Science and Technology forum on DPReview.

I have a feeling the OP needs some of the heavyweights there to answer this.

 

[Hanno van Tartu]

I know what polarizers do, and I know the technical differences between circular and linear polarizers, so no need to explain any of that.

Circular polarizers became a necessity with DSLRs because of issues related to both autofocus and metering systems because these take their information from mirrors inside the camera box. The mirrors actually polarize the light going to the sensors to some extent, so depending on the angle of the linearly polarized light entering the camera (when using a linear filter) the combination of linear filter and focus/metering mirror could cut light intensity much like crossing two polarizers.

Since mirrorless cameras have PDAF sensors embedded right on the imaging sensor (not sure about CDAF) I was wondering if there is no longer a need to use circular polarizers. Linear polarizers are cheaper; alternatively one could spend the same $$ and get a very high quality linear polarizer.

J

I bought a Nikon linear 'Polar 52mm' filter for about USD 17 on eBay a week before I received my Fujifilm X100T. It is the old type with very wide front rim, in practically new condition. I have used the combination without noticing any disadvantage. Automatic focus and automatic exposure seem to have no problems.

In order to answer your question, I put my 'Nikon circular polar II 52mm' filter (worth about USD 80) on the X100T and cannot see any relevant difference to the linear polarizer.

You might ask, how I screw a 52mm filter onto the 49mm male thread of the X100T. I use a Kiwifoto 52mm adapter -- just for the purpose of being able to use my collection of 52mm filters and other attachments. This adapter has a 49mm female thread that goes on the X100T and a standard 52mm female thread for the filters.

I just jumped in at the deep end and was lucky. I recommend you do the same -- what have you got to loose?

 

[Billy Ray Valentine]

I know what polarizers do, and I know the technical differences between circular and linear polarizers, so no need to explain any of that.

Circular polarizers became a necessity with DSLRs because of issues related to both autofocus and metering systems because these take their information from mirrors inside the camera box. The mirrors actually polarize the light going to the sensors to some extent, so depending on the angle of the linearly polarized light entering the camera (when using a linear filter) the combination of linear filter and focus/metering mirror could cut light intensity much like crossing two polarizers.

Since mirrorless cameras have PDAF sensors embedded right on the imaging sensor (not sure about CDAF) I was wondering if there is no longer a need to use circular polarizers. Linear polarizers are cheaper; alternatively one could spend the same $$ and get a very high quality linear polarizer.

J

A: No. But good luck getting high-quality linear polarizers. I gave up and kept my old cir-pols from the DSLR era.

"DSLR era" nice joke...

 

[teddoman]

I know what polarizers do, and I know the technical differences between circular and linear polarizers, so no need to explain any of that.

Circular polarizers became a necessity with DSLRs because of issues related to both autofocus and metering systems because these take their information from mirrors inside the camera box. The mirrors actually polarize the light going to the sensors to some extent, so depending on the angle of the linearly polarized light entering the camera (when using a linear filter) the combination of linear filter and focus/metering mirror could cut light intensity much like crossing two polarizers.

Since mirrorless cameras have PDAF sensors embedded right on the imaging sensor (not sure about CDAF) I was wondering if there is no longer a need to use circular polarizers. Linear polarizers are cheaper; alternatively one could spend the same $$ and get a very high quality linear polarizer.

J

A: No. But good luck getting high-quality linear polarizers. I gave up and kept my old cir-pols from the DSLR era.

"DSLR era" nice joke...

And let's not forget the "digital compact camera era"

 

[GaryW]

i have used 3 or 4 different liner polarizes on my 5Dc and 20D and never had any AF or metering problems the makes are cokin,vivitar ,hoya and a Minolta ..i do now have a hoya cpl will have to try it to see if any different

I agree. While I usually used a CPL on my DSLR, because you're supposed to, I tried a linear one, and didn't notice problems. When I posted about this (years ago), the response I got was that it should have failed occasionally, and I must have just not happened to get into that alignment where it fails to AF. I don't know!

What I haven't seen anyone mention is that there's a big cost in CPLs, not just in price, but in the loss of light. A simpler LP can often pass more light than a cheap CPL, as it has less stuff in the light path-way. (Expensive CPLs may be fine in this regard, but I suspect most people are like me and buying more modestly priced equipment.)

So, I bought a 49mm LP for my Nex. I don't know if it's multicoated. Seems fine. Only loses about 1 to 1 1/3 stop of light.

For the 40.5mm 16-50, I got a cheap CPL because of the price.

 

[antoineb]

reflections DO polarize light, linearly i.e. in a preferred direction, generally not totally of course.

When a light ray enters the DSLR, goes trough the tiny non-reflecting area in the main mirror, it then hits the secondary mirror which sends it to the AF sensors.

As this happens, the secondary mirror polarizes that light ray.

But if that light ray was already polarized by a linear polarizer, most probably with a different angle of polarization, then the 2nd polarization, from the reflection on the mirror, might cancel out most of the incoming light and make the AF sensors inoperative - or even if it cancels only part of the light it might make them slower to react, or react wrongly.

But of course a linera polarizer is much nicer, because you can select which alignement (perpendicular to the direction the light is coming from) you want to favour. We all know the typical example, car windshield, reflections that prevent from seeing inside, a linear polarizer is great there.

Quite clearly I think that just like phase detection is hurt by a linear polarizer on DSLRs, it would also be affected on these cameras - it's just optics.

Can you developp?

the light is not reflected to reach on sensor PDAF in a mirorless (but with a miror in a DSLR),

is there another way for the polarizer to affect phase detect focus ?

 

[JohnFrim]

But of course a linera polarizer is much nicer, because you can select which alignement (perpendicular to the direction the light is coming from) you want to favour. We all know the typical example, car windshield, reflections that prevent from seeing inside, a linear polarizer is great there.

If you are implying that "… a linear polarizer is much nicer than a circular polarizer because you can select which alignment…" then you don't understand circular polarizers. CPLs are LPs with a quarter wave retardation plate added behind the linear component. You can still rotate the CPL to reduce reflections. They perform the same wrt to the effect on the image.

 

[val1]

No beam splitter so you should be fine with linear.

 

[Juhaz]

But if that light ray was already polarized by a linear polarizer, most probably with a different angle of polarization, then the 2nd polarization, from the reflection on the mirror, might cancel out most of the incoming light and make the AF sensors inoperative - or even if it cancels only part of the light it might make them slower to react, or react wrongly.

Nobody was claiming that reflections don't affect polarization. The whole point was that unlike in SLR, there are no reflections going on in a mirrorless camera, the PDAF sensors are on-sensor, directly on the main light path. So what do you base your assertion of them being affected on?

 

[GaryW]

antoineb wrote:....

But of course a linera polarizer is much nicer, because you can select which alignement (perpendicular to the direction the light is coming from) you want to favour. We all know the typical example, car windshield, reflections that prevent from seeing inside, a linear polarizer is great there.

A CPL is great there too. As someone posted earlier, a CPL still has an LP inside it as the first layer...

l

 

[mladenF]

As a future A6000 owner (or A7000 maybe) I often visit this forum seeking for some information that may be usefull to me in building my next camera system. Intrigued with the fact that this thread is so long and with no real answer I took out my old DSLR and did some tests.

The camera is Olympus E-520, one of early DSLR-s with live view. For TTL operation it uses PDAF, for live view it uses CDAF. So I used it in both ways to see the possible difference in focusing with polarizing filter attached. Since I had used this camera with circular polarizator (CPF) many times, this time the linear polarizer (LPF) was of my concern. I found an old LPF from analog camera 'era' an this is what I found:

I found no difference. The camera operated equaly with or without filter. The only real difference was that I needed 1,5x longer shutter speed with filter attached. With, or without filter, camera focused equally fast (or slow) and the result photos were equally sharp.

And this is how I did it:

I set the camera on aperture priority and single auto focusing. I used kit lens and took photos on 3 different focal lengths: 14, 25 and 42mm (28, 50 and 82 in 4/3 terms). I also took photos from two various distances: from an object approx. 30m away, and from a very close distance. I photographed objects with and without filter in both TTL and live view mode, so both CDAF and PDAF metering took place. While taking photos with filter i rotated filter in many directions.

I also took photos in less systematic way from many reflective surfaces.

I never noticed any difference in focusing speed or quality of photos regarding the filter. Of course, CDAF and PDAF has their ways.

That leads me to conslusion that statement "DSLR cameras focuses badly with LPF" is wrong. Correct statement would be "SOME DSLR cameras focuses badly with LPF".

If there is a problem with PDAF and LPF, it is for sure camera dependent. I dare you guys to do similar tests on your DSLRs an to solve this mistery once for all.

And there is more:

LPFs are much better than CPFs. Very much better. The polarization effect is stronger and control of polarization effect is better. You don't believe. Try it for yourself. If you happen to have both kind of polarizers try them side by side. I did that and here is what I saw:

I choosed an reflective surface and set my position so that I viewed that suface about 45 degrees from source of light that shined on it. I rotated both filters for full circle.

- CPF eliminated reflection partially with little difference between angles of rotation. But it did change color of the scene at various angles.

- LPF did much better. At some positions it completelly eliminated the reflection while at some others did'n affect it allmost at all. Besides, it did't change color of the scene at any of it's positions. LPF has distinctive mark of polarisation direction on its rotating ring.

Having all of this in mind I suppose with great confidence that LPF would not affect A6000 focusing in any way. I might be wrong, but if anybody is in position to do some tests on A6000, please do so.

And more:

Reading of polarisaton on DRLRs mirrors in this thread I asked myself did anybody tried to test that. Just take your DSLR, look through the viewfinder at an reflective surface and rotate your camera. If there is polarisation it will do effect on reflection. I tried this with my OLYMPUS an saw no effect. But, OLYMPUS has penta mirror, not penta prism. I expect that prism might have polarising effect but an mirror for sure does not.

Furthermore, you can check it out with a ordinary mirror. Look in the mirror having some unpolarized source of light reflected - the sun or ordinary bulb will do. But don't look at some reflective surface, that reflection is allready polarized. Rotate your polarizing filter in front of mirror and see what's happening. Does the reflection changes?

Ah, this was long. Hope I wasn't boring very much. I need a beer now. Bye.