| Polarizer behavior with polarized illumination
Hi, Gang,
I just made some rather primitive tests regarding the response of a digital camera, equipped with a polarizer, to an illuminated test target, where the test target was illuminated both by randomly polarized ("non-polarized") and polarized light. Please keep in mind that the experimental setup and conditions were far from ideal, and certain steps I could have taken to minimize experimental error were not taken because, frankly, I didn't have the time.
The camera used was a Canon EOS 40D dSLR, equipped with a Canon EF 24-105 mm f/4.0 L IS lens. The camera polarizer was a Hoya Super circular polarizer.
Exposures were all taken in M mode, with a fixed aperture and shutter speed for any given series.
The test target was a sheet of plain white copier paper. Is was photographed at its center from a distance of about 24 inches, from an angle of about 45° to the normal.
The light source was incident on the target at an angle of incidence of about 45° to the normal, and "from the side" compared to the camera position.
Each series used a consistent illumination setup, all based on a small incandescent floodlight. In each series, three exposures were taken with a consistent aperture and shutter speed:
1. With no polarizer on the camera.
2. With a polarizer on the camera with its axis of polarization along the "6-12 o'clock" axis.
3. With a polarizer on the camera with its axis of polarization along the "3-9 o'clock" axis.
For the first series , no polarizer was used on the light source.
The apparent photometric exposure on the sensor, compared to that with no polarizer in place, was:
For the polarizer in place, 6-12 o'clock orientation: -1.9 stops.
For the polarizer in place, 3-9 o'clock orientation: -1.9 stops.
(A quick check was made to be certain that there was not, for example, a maximum or minimum at an orientation exactly between the two.)
Comment:
No surprise that there is no change with orientation.
The theoretical attenuation of a "perfect" polarizer (to non-polarized light) is 1.0 stop. Practical attenuators are "lossy", as we see in this case (where there is about 0.87 stop of excess attenuation.) I have no idea if this is typical. Somehow I would have expected less, but I have never looked into this through specifications or otherwise.
For the second series, a polarizer was used on the light source. (I had no really suitable one, so I used my fancy sunglasses! I confirmed that they seemed to exhibit classic polarizer properties before the test. This was of course not done "scientifically")
The apparent photometric exposure on the sensor, compared to that with no polarizer in place, was:
For the polarizer in place, 6-12 o'clock orientation: -2.1 stops.
For the polarizer in place, 3-9 o'clock orientation: -2.1 stops.
(Again, a quick check was made to be certain that there was not, for example, a maximum or minimum at an orientation exactly between the two.)
Comment:
No surprise that there is no change with orientation. Even though the incident light is polarized, if the target were perfectly diffuse, the reflected light would be non-polarized. That seems to be the case.
The small difference in the attenuation of the polarizer between the two series' is probably due to various practical shortcomings of the test situation, which I will not belabor here.
Best regards,
Doug
White Balance so easy, even our 5 year old can do it.- Melissa Strickland
Last edited by Doug_Kerr; 12-03-2008 at 08:50 PM.
| 
Linear Mode
