Polarizers and Image Noise

[Geoff F]

Following on from my Project 2013 thread and a few similar posts, we have mentioned the use of polarizers and whether they were worth using.

For general use, I have found that I can mostly achieve the same results, or better, from different shooting and/or processing options.

But I intended to do some more experimentations.

Looking at the final edits, which is what mostly matters, I found virtually no difference with regard to saturation, brightness, contrast, etc. But there does appear to be a considerable difference concerning the amount of image noise.

Some close crops to show the difference. An overcast day and I was shooting at 90 degrees to the light.





The difference will be most noticeable at full screen size.

There was also quite a bit of variation in the available shutter speeds, as you would expect, which ranged between 1/100 with polarizer to 1/500 without for the same scene.

I just used some slight noise reduction while converting the Raw files; not specialised software.

[revi]

Strange, the difference in noise. Although I did notice that the first image had a -0.33 EV exposure bias.
(Seems too little to explain the difference in noise, unless you hit a threshold somewhere).

[Mike Buckley]

Geoff,

When shooting on an overcast day, the polarizer isn't going to have much effect on the sky even when shooting at 90 degrees to the sun. That's because the polarizer affects direct reflections much more than diffuse reflections and the clouds diffuse the sunlight.

When using the example of the boat being reflected in the water, the relationship between the camera and the sun (the light source) is irrelevant. That's because the boat's reflection is a direct reflection rather than glare. Direct reflections are not affected by a polarizer. The water could have been replaced by a mirror (the water's surface is acting much like a mirror), in which case the polarizer will have no effect other than to act as a neutral density filter, which is what is occuring in your photo of the boat.

You mentioned that you don't see much difference in saturation when using a polarizer. Try photographing a grass lawn or plant leaves outside on a sunny day that produces glare on the leaves. The polarizer in that situation will dramatically reduce the glare and eliminate the need for post-processing techniques that try at best to emulate reduced glare. That reduced glare has the effect of increased saturation because the color of the leaves is seen rather than the glare.

Keep in mind that the position of the polarizer (I'm referring to the rotated position) will affect the glare only in one curved area, not all curved areas, of the items being photographed. As an example, rotate a polarizer when shooting a curved car hood on a bright sunny day. You'll see that as you change the rotated position, the part of the hood that is being affected also changes. So, when you are addressing glare in a situation when there are lots of curved surfaces and/or flat surfaces that are at lots of different angles to the camera, the polarizer will only affect some of the glare. In that situation, you need to decide which glare is the most important to eliminate.

[revi]

Sorry, I have to disagree wrt reflections in non-metallic surfaces (including water): such reflections are partially polarised,
and thus they are affected by a polariser (angle wrt to the sun is irrelevant though). One of the uses of a polarising
filter is the removal of reflections on water to see what's under the surface.

The glare you mention on a lawn is also composed of individual reflections on leaves, and as such removed (in part) by a polarising filter.

Note that this only works for reflections on non-metallic surfaces, metallic surfaces do not partially polarise the light. Thus
the reflection from the metal surface in a mirror is not removed by a polariser, where the reflection on the glass surface is.

[Geoff F]

I did try a polarizer on my macro lens last summer to see if it helped to reduce glare problems on shiny insects. But no success; it just darkened the whole scene but the glare areas remained,

Much the same as metal surfaces I suppose.

With this scene, I used a mix of polarizer strengths and the one shown here was the best.

And I particularly chose this location, and light angle, as a direct result of some comments/suggestions from my Project 2013 postings. Where it was suggested that polarizers are most effective when shooting across the angle of light not along it.

I have 'messed about' with polarizers on and off for a few years but always ended up putting them to one side and continuing with a bit of exposure compensation or various forms of HDR and similar alternatives.

But I will continue experimenting with the noise improvement to see if it really does help in the long term. I did notice that in the original Raw images the shadows were a bit brighter with the polarizer.

Which I suppose might be caused by some reduction in the strength of the brightest areas, resulting in having less extreme shadows, and therefore less noise.

I will keep experimenting; nobody is so old that they can't learn something new!

[Graystar]

I don't know what's going on in your photos, but a polarizer doesn't generally increase noise. Only a change in ISO affects noise.

Noise is directly related to the amount of light collected, and nothing else. A polarizer can increase the apparent noise in a blue sky because the polarizer is eliminating some of the polarized light from the sky (and thus, collects less light from the sky...which also makes the sky a darker blue.) However, the effect is selective...only affecting the sources of polarized light. The sky would be affected, and nothing else. In your sample, the entire image is affected, and it's impossible for that to be caused by a polarizer.

Your comment, in which you say you can achieve the same or better results from reframing or post-processing, suggests to me that you don't fully understand what the polarizer is used for. You cannot reproduce the effects of a polarizer in post-processing.

A polarizer only cuts out polarized light...and even then, only light polarized in the right direction. Many surfaces polarize the reflected light, and so a polarizer has some effect on them. Metallic surfaces do not polarize light. However, polarized light that is then reflected from a metallic surface will remain polarized, and so can be blocked by a polarizer.

Polarizers are used to block polarized light when that light is creating adverse effects in the scene. A very good example is of reflections on water. Let's say you're trying to photography koi in a koi pond, but all you're getting in your images are the reflections of people around you who are also trying to photograph the koi. All you're photographing is their reflections from the surface of the water. With a polarizer, you can eliminate that reflection and photograph the koi clearly. That is an effect that is simply impossible to reproduce in post processing. Many times, even repositioning yourself doesn't help.

Polarizers also cut the harsh reflection that you get from foliage. That has a two-fold effect. First, the foliage appears as a more even green color. Second, the reduction of light usually leads to an increase in exposure, brightening the darker areas of the image and producing a more balanced luminance to the scene. Again, this effect is impossible to achieve in post processing.

Polarizers are very useful, but they are very task-specific. Most of the time you don't need a polarizer. But when you do, and the conditions are right for using one, nothing else can do what a polarizer does, and the effects can sometimes be remarkable.

[Manfred M]

I do virtually all of my outdoor shooting with a polarizer. Just about the only time I don't use it is when it is getting too dark out or when I shoot a scene with sky in it with a wide-angle lens.

1. Great for getting deep blue sky and enhancing clouds when one is shooting at or near to right angle to sunlight. One can overdo it so dialing back a bit works better than going for maximum effect.

2. Reducing reflections from plants in all kinds of conditions, including overcast. It kills the reflection off leaves and give crisper shots.

3. Killing reflections off non-metallic subjects. Great for killing glare off of water or glass, regardless of lighting.


I can't explain the noise; the only thing I can think of is a "hot" sensor that is somehow producing (and showing) more noise. I've never seen this in any of my shooting other than after shooting video for an extended period.

[DanK]

I'm confused.

The noisier photo is the one on top, with the file name ending v5. This was shot at 1/500. The less noisy one was shot at 1/100. If the noisy one is the one with the polarizer, it should have been shot with a LONGER shutter speed, not a shorter one, because less light is let in. This suggests to me either that the ISOs were quite different, or that the exposure levels were different and you pulled the noisy one up in post. Either of these could account for noise. Only exposure level and ISO should affect noise. Polarizers shouldn't, if you maintain the same ISO and exposure level.

[Colin Southern]

I don't know what's going on in your photos, but a polarizer doesn't generally increase noise. Only a change in ISO affects noise.

Sorta - it's really a change of signal-to-noise that has the effect. If for example one shot without a polariser with manual exposure settings - then fitted a polariser but didn't adjust the exposure for the attenuation of the polariser (especially if already at a reasonably high ISO) then more noise would be apparent when the image was adjusted in post-production.

[Graystar]

Here's a good article on polarizer use...

http://archive.popphoto.com/pdfs/200.../Polarizer.pdf

[Graystar]

Sorta - it's really a change of signal-to-noise that has the effect. If for example one shot without a polariser with manual exposure settings - then fitted a polariser but didn't adjust the exposure for the attenuation of the polariser (especially if already at a reasonably high ISO) then more noise would be apparent when the image was adjusted in post-production.

Yes, but the problem there was that the person didn't correct the exposure the way it is supposed to be corrected. You'd get the same problem if the light levels dropped and exposure wasn't adjusted. Exposure errors are not valid variations.

[Mike Buckley]

Sorry, I have to disagree wrt reflections in non-metallic surfaces (including water): such reflections are partially polarised, and thus they are affected by a polariser (angle wrt to the sun is irrelevant though). One of the uses of a polarising filter is the removal of reflections on water to see what's under the surface.

I don't pretend to be an expert on this, but I don't think that's accurate. Glare can easily be removed from water using a polarizer. However, I don't think that's true of reflection, which is not the same as glare at least in the practical application of how the two terms are commonly used. The common generality, which of course is subject to some exceptions to some degree, is as stated in Light: Science and Magic: "If the surface is made of a material that conducts electricity (metal is the most common example), its reflection is likely to be unpolarized" and, thus, will not be affected by a polarizer. Clearly, water is highly conductive of electricity.

This perhaps explains why the reflection of the boat in the water seems not to have been affected by the polarizer, though I do understand that one could also argue that the polarizer was not placed in the ideal position to make that happen. In the end, though, the real proof of whether a reflection is polarized is whether or not a polarizer removes it. If it does, the reflection was polarized.

[Mike Buckley]

For the record, I'm ignoring the issue of the noise and am not trying to draw any conclusions about it from these two photos. That's partly because I don't know how much noise to expect in this situation because I don't examine noise at 100% unless I first see it at far lower viewing magnifications. It's also partly because this is not at all a scientific comparison of the same situation with only one variable. We know that one image was shot at -1/3 EV compensation and one wasn't. There is also the difference of exposure because the composition is not the same. The histograms are similar but not the same. Indeed, photographing a boat floating in the water, which makes it impossible to produce the exact same composition even when using a tripod, does complicate the issue.

[pnodrog]

Sorry, I have to disagree wrt reflections in non-metallic surfaces (including water): such reflections are partially polarised,
and thus they are affected by a polariser (angle wrt to the sun is irrelevant though). One of the uses of a polarising
filter is the removal of reflections on water to see what's under the surface.

The glare you mention on a lawn is also composed of individual reflections on leaves, and as such removed (in part) by a polarising filter.

Note that this only works for reflections on non-metallic surfaces, metallic surfaces do not partially polarise the light. Thus
the reflection from the metal surface in a mirror is not removed by a polariser, where the reflection on the glass surface is.

As I understand it a mirror or highly polished metallic surface rotates any incident light through a 180 deg angle of polarisation. So if the incident light is polarised a polarising filter may assist in reducing the reflections from a metallic or mirrored surface. As noted above the metallic surface does not partially polarise light.

[Geoff F]

The shot with the polarizer has less noise than the straight shot. That is what surprised me. Hence my suggestion that if you reduce the brightest areas slightly the shadows will then become brighter and have less noise.

[Graystar]

Glare can easily be removed from water using a polarizer. However, I don't think that's true of reflection...

I went and got some water, light, a polarizer, and tried it. The polarizer did remove reflections, though the amount varies with the angles.

It seems best to test this with a deep, dark, pot of water to maximize visibility of the reflection. I did my test with a cup of water because, you know...being in NYC, water costs a fortune (and the fee to drain it away is more than the water!)

[revi]

One thing to keep in mind is that the degree of polarisation on reflection is depending on the incident angle of the light.
IIRC there's no polarisation at 90°, maximum polarisation occurs at ~53° (Brewster's angle) on water.

Here are a couple of examples of the effect of a polariser in different situations.

[Colin Southern]

Yes, but the problem there was that the person didn't correct the exposure the way it is supposed to be corrected. You'd get the same problem if the light levels dropped and exposure wasn't adjusted. Exposure errors are not valid variations.

It really depends on why they're testing. I agree that it SHOULDN'T be a valid variation ... I just wanted to point out that noise is a function of signal to noise ratio though so althogh ISO is one factor that affects it, under-exposure is another. One can't assume a "correct exposure" because in real-world photography "correct" exposure often means different things at extreme ISOs than it does at base ISO eg normal evaluative metering will normally give good "no-noise" results at base iso, but ETTR techniques will be needed at extreme ISOs.

[Mike Buckley]

IIRC there's no polarisation at 90°, maximum polarisation occurs at ~53° (Brewster's angle) on water.

That's really interesting and something that I had never heard of. Thanks for sharing it!

Science: Light & Magic has a more general take: "If the surface looks like a mirror--for example, bright metal--the reflection is likely to be simple direct reflection, not glare," (not polarized direct reflection). "If the surface does not have a mirror-like appearance--for example, polished wood or leather-- the reflection is more likely to be polarized if the camera is seeing it at an angle of 40 to 50 degrees. (The exact angle depends on the subject material.) At other angles, the reflection is more likely to be unpolarized direct reflection."

[davidedric]

"If the surface is made of a material that conducts electricity (metal is the most common example), its reflection is likely to be unpolarized" and, thus, will not be affected by a polarizer. Clearly, water is highly conductive of electricity.

Sorry, resident geek here, but the statement really is a bit too general.

The key is the electronic structure of metals (which is what makes them metals) and which governs how they respond to electrical potential and electromagnetic radiation such as visible light. Conductivity and polarisation are aspects of this, but they are not the same thing.

Pure water is not a conductor. Impure water conducts via a different mechanism, movement of charged ions. Ions are big clunky things, unlike electrons, and I doubt they would be much impacted by ordinary light levels.

Dave