So I bought a polarising filter yesterday that has a dial you can turn to increase/decrease the polarisation effect and mount directly onto the lens. I also have an old polarising filter that was one that you mounted in a rectangular filter holder. By coincidence, I'm doing polarisation in physics at school at the moment, so understand the concepts.

Understanding that my phone screen outputted polarised light, you could turn my old filter around in front of the screen and watch it go from light to dark, then light then back to dark etc... It worked both ways, so if I flipped my filter around, the effect would be the same. So I try it out with my new filter, and begin to rotate it - the effect is the same as the old one as expected. However if I flipped it over, the effect was not an increase/decrease in brightness, it actually made the screen go from warm to cool, then back to warm then to cool as you rotated it (i.e. orange to blue).

I don't understand why this happens - surely it should work both ways like my older filter?

My phone is a Samsung Galaxy S8, so it has a Super AMOLED display.


You're probably comparing a linear polariser with a circular polariser. The linear polariser is a basic filter that only passes light waves polarised in a particular direction. That works either way round, and you can combine two of them to produce a variable density filter - by rotating the second polariser, it passes most of the light when the polarisation directions are the same, and a minimum amount when the polarisation directions are at 90 degrees to each other.

Many older polarising filters were linear ones - and that sounds like what your rectangular filter is.

But there's also a polarising effect when light reflects off glass or water at an angle (how much varies with angle). - that's why rotating a polariser can often reduce reflections from water or glass.

Unfortunately, that can cause problems when it happens in the metering/AF systems of a camera. So the manufacturers came up with the idea of circular polarisers - these are more expensive than plain linear polarisers, and combine a linear polariser (at the front) with a second layer that converts the linearly polarised light into circular polarised light - which doesn't suffer from the reflection effect. But that means they only work like linear polarisers one way round. (For details of how the second layer - a quarter wave plate - works, see https://en.wikipedia.org/wiki/Waveplate).

The colour shift effect is probably similar to that used in geological microscopes, where it can be used to help identify materials in thinly sliced rock samples.

  • 3
    As an addendum; a linear polarizer selects a (linear) polarization state by blocking all others. Circularly polarized light, on average, contains equal parts of all polarization angles. So no matter what, the linear polarizer will always take out (about) 50% of the light. If the source is unpolarized, like daylight, the waveplate has no effect. If it is polarized, it will convert the source from linear to circular, or circular to linear polarization. Your screen is linearly polarized, so the CPL in reverse makes it circularly polarized and thus there is no angle effect the "wrong" way. – Brandon Dube Dec 31 '17 at 23:42
  • 3
    @BrandonDube that deserves to be in an answer, since it directly answers the question of why it doesn't do anything when backwards :) – hobbs Jan 1 '18 at 0:48

The digital camera sports automation that adjusts exposure and focus. These mechanisms are likely dependent on semi-silvered mirrors. These work like mirrored sunglasses; they pass some light and reflect the remainder. When you mount a polarizing filter, it can be an impairment diminishing the effectiveness of this wonderful automation. Because mounting a polarizing filter is often desirable, we need one that will not do mischief.

The polarizing filter we use is actually two filters sandwiched together. The front facing filter is an ordinary polarizing linear screen. This one does the job. It mitigates reflections, cuts haze, and increases saturation. Because it darkens blue sky, white clouds are caused to stand out in bold relief. So it’s the upfront filter that does the deed.

Behind the polarizing screen is mounted a second filter called a retarder. This filter effectively de-polarizes the light. This sandwich filter is now called a circular polarizer. That’s OK because the upfront polarizing screen’s actions are not impaired and neither is the camera’s automation. If you turn this “circular” polarizer around, you effectively mitigate its ability to polarize light.

By the way, early scientists working with materials that polarized light falsely assumed that light had a + and – component, something like a magnet has a North and South Pole. They assumed the filter somehow split light into positive and negative (polarized) light beams. This was proven false but the name polarization stuck.

Most would agree, the polarizing filter is a must have filter.

  • Light is an electro-magnetic field and has polarity. The name polarization doesn't have anything to do with a + or -. – Brandon Dube Jan 1 '18 at 2:30
  • 2
    @ Brandon Dube --- Check out Étienne-Louis Malus (French) engineer, physicist, and mathematician who coined the word polarization to describe a phenomenon of light. – Alan Marcus Jan 1 '18 at 4:27

Polarisers are fun, and display quantum mechanics effects that can confuse people. Light, and all electromagnetic waves, can be considered to have two parts, with one that is a quarter wave behind the other. If the two parts have their electric fields aligned then we have a linear polarisation (and there are many directions of alignment). If the second part has turned to the side then the overall wave effect turns in a helix (spiral). The old filter selects just for the aligned waves, while the new filters select for the ones arranged as a helix (with a particular turn direction).

Many folk think that the linear aligned state is a 'quantum state' (i.e. countable) while in (quantum) reality it is only the two spiral states of left and right circular polarisation that are distinct countable states.

For a fun test, take three linear polarisers (e.g. polarised sunglasses at a store). The first set in a normal 'up' direction, the second set at right angles (across), and observe no light comes through. Now add the third at 45° at the back — still nothing, but now put that third polariser in the middle — Observe what happens, and wonder why! (Linear polarisation is not a quantum state..) See this MinutePhysics video for a demonstration.

The use of linear polarisers in quantum cryptography can fool people into thinking they have broken the laws of probability (see Bell's Theorem), when in fact it is just that the probability calculation are done on a circle not a square (those 45° linear polarisers are not a classic 50:50 selection!)


DSLRs and digital cameras with "phase detect" autofocus have problems when fed with polarised light. So a "circular polariser" or CPL filter will, after polarisation, employ a "quarterwave plate" for turning linearly polarised light (after the polarisation proper) back into something that has polarisation in two directions. This effect is not independent of the light's wavelength, so it doesn't really create fully circular polarisation for all wavelengths, but somewhat elliptical polarisation. For the purposes of phase-based focusing and metering (including analog camera TTL flash metering), this is good enough, and the photographic process itself is not affected.

For a digital camera with only electronic viewfinder and merely contrast-based focusing, a CPL is not actually necessary and a linear polariser will be fine.

Now when you use the CPL in reversed orientation before a source of linearly polarised white light like a typical laptop screen, the non-perfect circular polarisation achieved after passing the quarterwave plane (which is only truly quarterwave at one particular wavelength of light) will mean that the orientation of the linear polariser behind it is not entirely unrelated to how the different wavelength portions of the white light manage to get through.

If you are using polarised sunglasses, you may get into trouble with the LCD screens on cameras or smartphones. When you are lucky, sunglass and screen producer have adopted polarisation conventions that differ by 45°, making the combination work reasonably either in landscape or portrait mode. Even better would be a quarterwave plate on the LCD or the polariser glasses. But I don't think that's customary.


There are 3 kinds of polarising filters about:

  • 1st Linea (for Manual focus film cameras).
  • 2nd Circular (For manual focus film cameras), and
  • 3rd Circular (for DSLRs and general autofocus cameras which contain both so as not to upset focussing and or exposure components).

Unfortunately some manufacturers don’t understand anything about them which is why your circular polarising was mounted in a rotating bezel when there is no point in rotating it.

You need to have a linear component in the circular polarising filter if you’re using a DSLR or auto focus camera.

So best to do the phone screen trick before you buy the filter. That way you can at least tell that it has a linear component even though it says CPL on the side.

  • 2
    This is incorrect. There aren't circ. polarizers for film distinct from circ. polarizers for DSLRs. A circular polarizer is just a linear polarizer sandwiched with a quarter-wavelength retarder. The order matters: light going through a 1/4-𝜆 retarder first, then through the linear polarizer, just comes out linearly polarized. But going through the linear pol. first then through the 1/4-𝜆 comes out circularly polarized. It has nothing to do with "Oriental manufacturers don't understand anything about them...". – scottbb Feb 21 '19 at 20:37
  • 1
    Also, it's it's mounted in a rotating bezel so you can align it. You want that capability with a circular polarizer too — there certainly is a point. – mattdm Feb 22 '19 at 13:22

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.