[Eruditass]

Does anyone know of any good comparisons for linear polarizers, similar to the excellent circular polarizer test done by LensTip?

Unfortunately, a number of those manufacturers don't offer the same coatings and whatnot on their Linear Polarizers.

[Digitalis]

nope nothing comes to mind, though in my experience Linear polarisers are more efficient than circular types and transmit more light which can come in handy. Just use common sense when you're using it, and the issue of whether is is coated or not isn't going to be such a deal breaker. The fact is that all filters cause filter flare to varying degrees, the better ones still produce it however it's impact is reduced, but not eliminated, as their manufacturers so often claim.

The statement that linear polarisers foul up the AF in modern digital cameras is weapons grade BS, it's just scare marketing to make you buy more expensive filters.

[amoringello]

Not much direct help, but I fell for the coatings BS and bought an expensive circular polarizer. It works as well as the $30 polarizer. Yes, in about 0.5% of the time I might have noticed less glare where I might have expected some, but the cost does not outweigh a majority of its usefulness.

100% agree with Digitalis. Linear polarizer v.s. AF or metering is 100% BS. I don't know if it ever truly affected any particular sort of mechanism, but it certainly does not affect current cameras.

[bdery]

in my experience Linear polarisers are more efficient than circular types and transmit more light

There is no physical reason for that. A circular polarizer IS a linear polarizer, with another wave plate added. This second wave plate has no reason to degrade the image quality in any way.

Polarization is one of the most complex concepts in optics, and because of that it leads to many misconceptions. The quality of the polarizer will have much more to do with the result than the type.

Regarding various brands, anything from a reputable manufacturer will do fine. I own and use a polarizer from Optex, nowhere near the reputation of say, Hoya, and it works like a charm.

[Digitalis]

A circular polarizer IS a linear polarizer, with another wave plate added. This second wave plate has no reason to degrade the image quality in any way.

I never said it did, all I'm saying is that linear polarisers transmit approximately half a stop more light through them than circular ones do, because they don't have the second wave plate.

[Eruditass]

The statement that linear polarisers foul up the AF in modern digital cameras is weapons grade BS, it's just scare marketing to make you buy more expensive filters.

I've heard reports of it mucking up other metering systems, up to +/-3 EV. Less so on the autofocus system. Depends on how it was designed. People have said they contacted Pentax, who said it did not affect it.

Nevertheless, I took my cheap 55mm LP on my kit lens and did a non-reflective white wall test. I got +/- 0.15 EV while rotating the filter around: pretty much negligible.

I want at least one LP so I can simulate a variable ND filter.

Options that I've found:
Marumi
B+W (Super expensive, might as well get top of the line Marumi Circular which is well rated. Only one that mentions multicoating, though)
Hoya
Tiffen

Nevertheless, I have heard too many reports of crappy UV filters to pick one randomly, especially since I will be stacking two. And if you haven't read the lenstip test, it's pretty thorough: Polarizing filters test - Introduction - Lenstip.com

[bdery]

all I'm saying is that linear polarisers transmit approximately half a stop more light through them than circular ones do, because they don't have the second wave plate.

I understand what you meant. But with a well manufactured polarizer, the second wave plate will reduce, at most, light passing through by 4%. And any coating will lower that number. There is no reason why that wave plate should absorb half a stop of light.

[Digitalis]

There is no reason why that wave plate should absorb half a stop of light.

well that is what it appears to be doing, and it is certainly not impossible for it to absorb so much light.

I've heard reports of it mucking up other metering systems, up to +/-3 EV.

that only happens with metering systems that use a beam splitter.

[bdery]

well that is what it appears to be doing, and it is certainly not impossible for it to absorb so much light.

I never said it was impossible for physical objects to absorb or reflect light. I'm saying there's no reason why a good filter would do that. A sheet of cellophane might, of course.

Would you mind describing your comparison procedure? Here's how I would do it:

If you have different filters from different brands, then comparisons are worthless because coatings, glass elements, assembly procedures will be different. If you have filters of the same model from the same brand, then it's possible to compare them (assuming coatings are the same for CP and LP). To do so, it's necessary to identify the polarization axes of both filters (which can or cannot be aligned with the side mark that can or cannot be present). It's reasonable to say that finding the man axes within ±5° is possible, so that should be the uncertainty we base ourselves on (though I'd have to calculate how much ±5° would affect polarization, that's far from linear).

Once the filters are well characterized, it should be possible to measure their exposures readout when spot focusing on a static subject if the light souce does not vary.

Then, all that's left to do is to try to factor in the thickness of the filters (which will affect the direction of transmitted light) and the distance between the back surface of the filters and the front of the lens. Then it could be possible to evaluate, within a significant uncertainty value, the performance of each filter.

Or we could try to find the spec sheets of the filters, look at their light transmission specs, and trust the manufacturer. But I'm betting both filters will have ">97% light transmission" as their specs, if they are of the same brand and model.

Long story short, there are many, many variables that can affect polarization but which do not relate to the actual transmission of the filters. And if they are different models, it gets worse

...

I'm re-reading my text before clicking "Post" and I want to stress out something : I hope this does not sound arrogant in any way. That's not at all my intent. My objective is to make sure we stand on the same ground when when discussing optical performances. I'm a scientist completing a PhD in optics and that makes me methodical. I'm not trying to be "right", or "wrong", only accurate.

Good day.

[Digitalis]

I have never tested this, I have always taken the Linear polariser Vs Circular conundrum with a degree of scepticism. So testing this will be an interesting challenge, I accept.

Instead of using modern multicoated filters I opted for single coated filters since no manufacturer these days makes a fully multi-coated linear polariser,considering the addition of coatings could potentially alter the results I went for filters that were all equal. Secondly I used oversized filters (62mm on a lens with a 52mm thread) and placed them directly in front of the lens thereby eliminating any variances that could be introduced by the air space between the filter frame and the front objective. The lens used was a pentax SMCP-K 50mm f/1.2 stopped down to f/2.8 to eliminate vignetting. Flash was used at 1/8th power and the camera was tripod mounted and aimed at a white wall. I ascertained the alignment of the polarisers with the use of a rather handy mark on the frame of each filter, which was I visually confirmed to be the correct alignment - incidentally the filters used are all top quality Hoya filters made in Japan.

from left to right : circular polariser has a luminance of 53, the linear has a luminance of 73, and the control image with no polariser used is a perfect 128 - middle grey, these are crops from the centre of the image and were converted into greyscale.

[bdery]

Digitalis, many thanks for running the test. I only have CP at home, and not of the same brand, so I couldn't run the test here.

Your test settings sound good to me, as long as you verified the alignment of the side mark (I've seen polarizers where the mark is not aligned at all with any axis, or with no marks).

The luminance ratio between the filter-free shot and the LP shot is 57%. A perfectly randomly polarized source illuminating a polarizer will show a 50% ratio, decreased by the reflections at the two interfaces of the filter (normally 4% each, that number decreasing when good coatings are applied). So either your wall polarizes light to some extend, or the filter was not perfectly aligned (quite probable, it's not as if filters had sub-degrees precision )

The ratio between the two filters is 72%. dismissing interfaces refecletions (since we can't know the effectiveness of the coatings, even though the CP has one more interface), and assuming the same uncertainty (7%) for the positionning of the second filter, it means the ratio is between 65 and 79%.

It's more than I expected (or the ratio is lower, if you will), though still less than a half-stop. It's probable that the quarter-wave plate absorbs some of the incident light.

thanks again!

[Digitalis]

I verified the polarisers alignment using live view with the histogram under tungsten light, the wall I was using had no observable specularity, and I bounced the flash to further eliminate surface specularity from affecting the results. Additionally The use of bounced flash off a white reflector provided a large quantity of diffused non-polarised light. The test results surprised me too, while the difference between the two polarises is less than a stop there is still a noticeable difference between the two filters, and yes I would say the quater wave plate is the culprit with the loss of light.

I would say a error tolerance of 7% is pretty fair because no polariser I own has sub degree markings - Tiffen filters used to have such markings as I recall.

[jolepp]

Hmm ... could it be that the flash reflectors polarize some of the light ? I suppose this could be resolved (to some degree at least :-) by taking comparison shots with the polarizer rotated 90 degrees (?).

[Ben_Edict]

I've heard reports of it mucking up other metering systems, up to +/-3 EV. Less so on the autofocus system. Depends on how it was designed. People have said they contacted Pentax, who said it did not affect it.

The circular polarizers are only necessary for cameras, which have a semi-transparent mirror or the like between the lens and the exposure meter. That is for example what the Pentax LX had. No Pentax DSLR has such a construction, as all of them take the meter reading inside the prism housing and get the light reflected directly from the main mirror. So, no polarizer will affect the metering, whether it be a linear or circualr type. And that is true for almost all current DSLRs.

AF is different, as in front of the AF sensors, there is a polarizing element (aka a semi-transparent mirror). BUT this will not lead to AF errors or malfunction. In the worst case it will readuce the amount of useable light for the sensor, which may lead to the AF stop working in very dim conditions. But in those conditions, you are highly unlikely to use the polarizer, anyway.

So, in theory using a linear polarizer instead of the circular type, may affect AF function. In practice this effect is irrelevant.

Nevertheless, I took my cheap 55mm LP on my kit lens and did a non-reflective white wall test. I got +/- 0.15 EV while rotating the filter around: pretty much negligible.

I want at least one LP so I can simulate a variable ND filter.

Options that I've found:
Marumi
B+W (Super expensive, might as well get top of the line Marumi Circular which is well rated. Only one that mentions multicoating, though)
Hoya
Tiffen

Nevertheless, I have heard too many reports of crappy UV filters to pick one randomly, especially since I will be stacking two. And if you haven't read the lenstip test, it's pretty thorough: Polarizing filters test - Introduction - Lenstip.com

I personally prefer high quality multicoated polarizers, because we are dealing here with 2 glass plates and 4 glass-to-air surfaces. That increases the probablity for flareing ang ghosting very much. I don't care much about the reduced loss of light, as the polarizer will reduce the light much more.

Ben

P.S.: we have had a fair amount of in-dpth discussions of polarizers a couple of months ago and you'll find a lot of info in those threads.

[wjjstu]

AF is different, as in front of the AF sensors, there is a polarizing element (aka a semi-transparent mirror). BUT this will not lead to AF errors or malfunction. In the worst case it will readuce the amount of useable light for the sensor, which may lead to the AF stop working in very dim conditions. But in those conditions, you are highly unlikely to use the polarizer, anyway.

So, in theory using a linear polarizer instead of the circular type, may affect AF function. In practice this effect is irrelevant.

This post seems to indicate otherwise.