3
\$\begingroup\$

I learned that polarizers did not work with autofocus and we had to get new "CPL" filters, 27 years ago. I even understand how they work and the nature of the light produced, which was a lot more effort in those days.

But I never found out why phase-detection sensors don't like linear polarized light. The old manual-focus ground glass screens featured "split prism" and a ring of smaller units, and we have no trouble seeing the effect with our eyes. The way light is blocked from one side or the other of the lens is not affected by polarization. The line sensors behind the split aperture must have some difficulty—but aren't image pixel sensors the same concept (cmos or ccd light measuring spots) (actually, I've never tried a linear polarizer's affects on a digital camera image so I'm just assuming it doesn't care).

What is it that doesn't like (linear) polarized light?

Improve this question
\$\endgroup\$

1 Answer 1

Reset to default
4
\$\begingroup\$

The light path for the autofocus in an SLR (image) sends it through a semi-transparent window in the middle of the mirror (in effect a beam splitter).

Light, bouncing off of or being transmitted the mirror will become partially linearly polarized itself (much like the light off of a leaf or water). If the light incoming is linearly polarized, some of that light will be lost as part of the process - just as with light going through cross polarizers.

The resulting light loss will show up in two ways:

  • reduction of light hitting the auto exposure sensors leading to the overexposure of the image
  • reduction of light hitting the auto focus sensors leading to a reduction of contrast and more difficulty focusing.

This effect would be variable based on the angle of polarization.

Furthermore, note that anti aliasing filers make use of linear polarization which could also be disrupted by having consistently linearly polarized light being thrown at them.

The solution to this is to have the light polarized in a way that is not affected by a linear polarizer (the semi-transparent window in the mirror). This can be done by placing a quarter wave plate on the back of a linear polarizer making the polarization 'spin' which is known as circular polarization.

Improve this answer
\$\endgroup\$
8
  • \$\begingroup\$ You have mentioned contrast reduction, but phase detection doesnt rely on contrast, does it? \$\endgroup\$
    – Digital Lightcraft
    May 22, 2015 at 15:48
  • \$\begingroup\$ @DigitalLightcraft loss of contrast its a side effect of less light getting through. Any reduction of discernible signal (amount of light) makes it harder. This is in part why autofocus on SLRs doesn't work with lenses that are f/6.3 or slower - there isn't enough light to be able to detect the difference. This means instead of seeing a sharper peak for phase detection (and lining those shapes up), the shapes are all at a much lower 'base' level. This is a loss of contrast. Phase detection autofocus has trouble focusing on contrastless flat surfaces. \$\endgroup\$
    – user13451
    May 22, 2015 at 15:55
  • \$\begingroup\$ If the small aperture is the correct exposure (with a sub-second time), there is enough light. The problem is the lack of physical separation between the two sides of the window. \$\endgroup\$
    – JDługosz
    May 22, 2015 at 18:45
  • \$\begingroup\$ @MichaelT I indicated in my question that I knew how CPL filters work. I'm actually rather interested in quantum mechanics, and in 1988 was digging through a large dead-tree library to find out. It was more challenging before the WWW. \$\endgroup\$
    – JDługosz
    May 22, 2015 at 18:49
  • \$\begingroup\$ @JDługosz If you are interested in the physics / quantum mechanics, you may find physics.SE to be a more appropriate venue to answer the question. Going too deep into the physics and optics realm draws more from one's expertise as a physicist rather than a photographer. \$\endgroup\$
    – user13451
    May 22, 2015 at 18:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.