Polarization and UV/IR Photography: Rebooting an Antique Technology

[OlDoinyo]

For the modern photographer, the polarizing filter is one of the most useful tools: it can darken washed-out skies, modulate reflected glare, and even modify the apparent texture of surfaces. Even in an era where most filters of the past have been replaced by equivalents generated by digital post-processing, it remains as much an indispensable tool as it has been since Edwin Land invented polarized polyvinyl alcohol sheet material in 1928. Before this invention, the only transmissive polarizing filters were certain hand-cut mineral crystals, which were small, rare, and generally known only to scientists.

 

Invisible-light photographers, however, have been left behind. It turns out that ordinary polarizing filters do not transmit ultraviolet, and they polarize infrared light only barely beyond the visible limit. However, a different approach may be possible in the infrared and ultraviolet realms. Landscape painters of the 19th century were known to make use of polarizing devices as visualization aids, but these depended on very different principles from today's filters. They utilized the principle of "reflected glare" (specular reflection off a non-mirrored interface between two media) to help generate an image with polarized light. An example of such devices may be seen at the Desert View overlook in Grand Canyon National Park--they consist of polished slabs of dark stone mounted on hinges which can be rotated to various angles to view the reflection of the landscape.

 

The optical principles behind this device were elucidated in 1808 and involve what has come to be known as Brewster's Law, where the angle (relative to the perpendicular) at which an incident, horizontally polarized beam of light will be maximally transmitted (not reflected) is given by

 

B = arc tan (n₂/n₁),

 

where n₁ and n₂ are the refractive indices of the two media (most often air and some other substance, respectively;) the angle thus defined is called Brewster's angle and is the angle at which the polarization is most pronounced. For green light and an air/glass interface, Brewster's angle is commonly quoted as 56 degrees from normal (i.e. 34 degrees from the plane of the surface.) For infrared light, the refractive index is lower, so that we can expect a slightly lower Brewster's angle; the reverse is true for ultraviolet, where the refractive index is higher. It is in the vicinity of Brewster's angle that we may expect maximum polarization performance.

 

Testing this idea involved finding subject matter which will exhibit obvious polarization properties (i.e. a sky nearly perpendicular to the sun direction), improvising an aimable mount to be placed before the camera, and finding suitable reflectors that can be mounted. It was decided that uncoated glass filters provided an easy choice for a reflector, due to their high degree of surface flatness. The filter should ideally absorb the transmitted component of the incident light. A wide variety of filter types might be used for the UV version of this experiment, from yellow to orange to red to black-IR. For the IR version, one of the blue-green glasses seems the best choice. Due to what I had available, I chose 72mm filters; I used a Rocolax 1K for the UV trials and a Kolari deconverting filter for the IR trials. The camera itself was fitted with the Steinheil 50mm lens and either a Baader U2 filter or a B+W 093 filter, as appropriate. The filters were mounted on a stepper ring taped to a hand-cut foam wedge mounted on a tripod independent of the camera itself. The entire setup appeared thus:

 

 

A scene close to my house on a fair day was chosen with the sun angle about 70 degrees from the main sky area visible. For each trial, three images were acquired: one reference image, directly from camera to subject; one "parallel" image, with the reflective surface edge-on to the sun; and one "perpendicular" image, with the filter's lateral axis perpendicular to the incoming sun. The plastic wedge served to keep the reflector/subject angle in the vicinity of Brewster's angle. The reflected images were substantially dimmer than the reference images: trial-and-error revealed about a 2.5-stop drop for the ultraviolet images and about a 4-stop deficit for the infrared images. The setup as pictured above turned out to be very temperamental and not suitable for real-life field use, as the two tripods had a tendency to get fouled in each other during adjustments, and alignment proved a tricky matter; but with some perseverance, images were obtained as seen below:

 

 

 

 

 

 

 

The sky in the UV reference image appears somewhat different than the others due to a slightly different workup. The other two images appear fairly similar, but I believe that the sky in the parallel image is slightly darker than that in the perpendicular image, indicating that some polarization effect has been observed. The reflection passband appears very flat, and color rendition is very similar to that in the reference image.

 

The IR set is of slightly better quality and consistency, showing a more pronounced sky-darkening in the parallel image vis-a-vis the perpendicular image (although sky-darkening is seldom needed in infrared photographs.) The images are monochromatic, because the reseau dyes do not absorb infrared light in this wavelength range. Again, there is little sign that the reflection passband is different from that of the reference photograph. A window reflection also appears in the perpendicular image that does not appear in the others.

 

It may be concluded that this old technique shows some promise in enabling photography in polarized ultraviolet or infrared light beyond the range of ordinary polarizing filters, though much refinement would be needed to make this method practical; larger reflectors would almost certainly be needed, as well as a more integrated design that does not require two tripods. Perhaps a more refined version of this method would make use of something like the old Spiratone Circo-Mirrotach, which could be mounted on the front of the taking lens without any additional support and which could rotate on the lens axis directly without the need for more complex alignment procedures.

[Andrea B.]

Clark, thank you for this excellent write-up of your polarization experiment.

 

Of course I was not familiar at all with the idea of using specular reflection in this way to produce a polarized image. So that was an interesting bit of scientific history to learn about.

 

And while your setup indeed does appear "tempermental", you still did prove the concept. Kudos!!

 

I would like to apologize for being so delayed in my comment. We were in Scotland during the first two weeks in May and I am still playing catch-up.

[nfoto]

Same excuse as Andrea. I really look forward to any further developments.

 

Is the mentioned Spiratone gizmo like those "spy" or "shoot around the corner" attachments you could purchase as indispensable gadgets for your camera many moons back in time?

[OlDoinyo]

Is the mentioned Spiratone gizmo like those "spy" or "shoot around the corner" attachments you could purchase as indispensable gadgets for your camera many moons back in time?

 

The very same. They were small and fixed at a 45 degree angle, so would not work for this as-is.

 

I also have heard of nanowire grid polarizers since I posted this. Perhaps that would be another useful approach.