Stop compensation with square polarizing filter

[rpagliari]

I have a square polarizer filter by Lee (100mm system). I was told by them that compensation is up to 1.5 stops depending on the orientation.

I typically use it either straight on, or rotated by 90 degrees. How do I know how many stops do I need to account for?

[Bob Salomon]

The exposure is the same in all positions, the polarizer does not change it’s density with rotation.

[Kiwi7475]

The exposure is the same in all positions, the polarizer does not change it’s density with rotation.

That’s something I always wondered... you’re right of course but the point I think is that the mount of light hitting the film will change based on orientation, simply because the polarizer acts like a filter for certain EM wave polarizations (basically all but linear polarized). So it would seem that the exposure and therefore the stop compensation has to be different depending on its orientation. So for example it will darken the sky or the water reflections if oriented appropriately, but not the shadow of a tree for example. So if you’re exposing for the shadows I guess you are right, but if you exposed for the highlights or some global average (which one shouldn’t do) then you would made to think you need to compensate for the loss of the polarizer.
Is this correct?

[reddesert]

I don't think there's any way to precisely answer this question other than "It depends."

The amount of light removed by the polarizer depends on how much polarized light is in the scene, and how much you are orienting the polarizer to remove that light. The polarizer will remove 50% of all unpolarized light (1 stop), and more than that of linearly polarized light, depending on how you orient it. Typically, a polarizer is used to selectively darken linearly polarized light from reflections or scattered light from the sky, and frequently one wants that darkening effect, so one doesn't necessarily want to add extra exposure to compensate for it. But you have to keep that in mind. If you pointed an averaging meter at a scene with a lot of sky, and then used a polarizer to darken the sky by a lot, you're taking away more than 1 stop of light, but whether you want to compensate for it depends on how you want the final image to turn out.

This is true for all polarizers regardless of manufacturer or shape.

[Jeff T]

In my experience, if the scene has lots of reflected light and the filter is set to cut all the reflected light then highlights will be reduced disproportionally compares to the shadows, thereby compressing the dynamic range of the scene. Sometimes I put the polarizer on the spot meter and check the highlights but it's not as accurate as using digital camera and with a histogram. At times when I used the maximum recommended compensation and I get great white clouds but the light blue sky has gone to dark blue on Velvia. In hind sight, I should have gone with a longer than recommend compensation time to get that baby blue color in the sky and probably sacrifice a little detail in the highlights.

[Mark Sawyer]

In my experience, if the scene has lots of reflected light and the filter is set to cut all the reflected light then highlights will be reduced disproportionally compares to the shadows...

Except for the light seen looking directly into the light source, all light is reflected light. Unless they're generating photons independently, highlights and shadows are just differing amounts of reflected light.

[Pere Casals]

The exposure is the same in all positions, the polarizer does not change it’s density with rotation.

Bob, perhaps it can be added that by rotating the filter we vary the filter density selectively for the share of light that comes polarized, but not for share of light having random polarization directions...

if we rotate the filter we can block highly polarized light from direct sunlight and its reflections, as image is saturated by polarizer less light enters, it's difficult to say if exposure has to be modified or not (this is YMMV) but in fact the ammount of light reaching film may vary if we have polarized light in the scene, that of course not always happens.

[rpagliari]

In my experience, if the scene has lots of reflected light and the filter is set to cut all the reflected light then highlights will be reduced disproportionally compares to the shadows, thereby compressing the dynamic range of the scene. Sometimes I put the polarizer on the spot meter and check the highlights but it's not as accurate as using digital camera and with a histogram. At times when I used the maximum recommended compensation and I get great white clouds but the light blue sky has gone to dark blue on Velvia. In hind sight, I should have gone with a longer than recommend compensation time to get that baby blue color in the sky and probably sacrifice a little detail in the highlights.

I hadn't thought of placing the spotmeter behind the filter. Great idea, I will try do that next time.

[Bob Salomon]

Bob, perhaps it can be added that by rotating the filter we vary the filter density selectively for the share of light that comes polarized, but not for share of light having random polarization directions...

if we rotate the filter we can block highly polarized light from direct sunlight and its reflections, as image is saturated by polarizer less light enters, it's difficult to say if exposure has to be modified or not (this is YMMV) but in fact the ammount of light reaching film may vary if we have polarized light in the scene, that of course not always happens.

Exposure doesn’t change. If you want exposure to change with rotation get a variable density ND filter. They are made by using two polarizes, one turns and the other doesn’t.
But rotating a single polarizer only polarizes light. It does not change exposure.

[Jeff T]

Mark, I agree, instead of reflected light, a better term would be polarized light.

Except for the light seen looking directly into the light source, all light is reflected light. Unless they're generating photons independently, highlights and shadows are just differing amounts of reflected light.