Polariser filter

Discussion in 'Beginner Questions' started by peterlove, Jun 23, 2022.

  1. When all theory has been expounded, do the friggin' experiment!

    Here's a set of comparisons between a linear and circular polarising filter, on all the DSLRs I could lay my hands on:
    1. An old Mk1 Canon 5D fitted with 50mm f/1.8 Canon 'plastic fantastic' lens -
    5D-no-filter.jpg
    5D-Linear.jpg
    5D-CPL.jpg

    2. Almost equally old Nikon D700 fitted with 50mm f/1.8 AF-D Nikkor - D700-no-filter.jpg
    D700-Linear.jpg
    D700-CPL.jpg

    3. Nikon D800 with same 50mm AF-D Nikkor -
    D800-no-filter.jpg
    D800-Linear.jpg
    D800-CPL.jpg

    4. Nikon 'DX' D7200 with 35mm f/1.8 AF-S lens -
    D7200-no-filter.jpg
    D7200-Linear.jpg
    D7200-CPL.jpg

    5. Sony A7Riv with 50mm f/1.8 AF-D Nikkor lens via adapter (manual focus) -
    A7Riv-no-filter.jpg
    A7Riv-Linear.jpg
    A7Riv-CPL.jpg

    All the above are SOOC JPEGs with no post work done apart from resizing and adding title text. All shot at f/4, ISO 100, Aperture priority, and AWB. Except the D7200, which was shot at f/2.8 to equalise the D-o-F and the D700 that was set to its base ISO of 200.

    Surprisingly, there is no discernable difference between the linear and CPL filter shots taken with the same camera, despite there being a slight visual tint and density difference when looking directly through the filters. I guess AWB and auto-exposure really do work to level out any difference!

    I was also surprised to find that the CPL filter was visually denser and gave an average 1/2 stop slower shutter speed, but as you can see the end results are difficult to tell apart.

    WRT AF: None of the cameras struggled to find focus, or missed focus enough to worry about. Although focus speed seemed a bit slower with either polariser fitted. I put this down to having a dimmer, and maybe less contrasty image to focus on.

    I deliberately threw the lens well OOF before each shot, and let the camera AF do its thing - except for the adapted lens on the Sony, which was manually focused.

    Here are 100% crops of the AF focus area from each camera:
    Canon-5D_crops.jpg
    Nikon_D700_crops.jpg
    Nikon_D800_crops.jpg
    Nikon_D7200_crops.jpg

    And to round things out the cropped focused area from the Sony MILC.
    Sony-crops.jpg

    Conclusion: The older DSLRs with AA filters did show quite a dramatic colour shift when fitted with a polariser. However the type of polariser appears to make little to no difference.

    Also, with the cameras tested, there were no severe side effects from using a linear polariser. In fact the extra half stop of light from the linear filter might be advantageous.

    So based on the above results, it's my opinion that reports of the death of the linear polariser are grossly exaggerated. Further, the spread of fear and panic over using linear polariser on DSLRs and AF/AE film cameras was IMHO largely unfounded, and a cry of wolf to stimulate sales of 'upgraded' filters.

    Maybe the above experiment should have been done years ago?

    Oh, P.S. The filters used were an old Nikon Linear polariser and a slightly newer Hoya CPL.
     
    Last edited: Jul 6, 2022
    NHSN likes this.
  2. The D7200 seemed to be focused the best with the most contrast separation between sky and clouds. Maybe lower exposure saturated the colors more.
     
  3. The D7200 is the only one of those DSLRs that doesn't have an AA filter Alan. And besides, there's a bit of white cloud behind the tree branches that's helping the contrast for focusing.

    I was dogged by a stiff, blustery breeze that moved the clouds constantly and blew the tree branches about that I was using to Auto Focus on. So I wouldn't draw too many conclusions about absolute AF performance.

    The point is that the AF on each camera worked equally well (or poorly) between no-filter and either flavour of polarising filter.

    Myth busted - in my opinion.
     
  4. OK, first, the big problem with aliasing is that it can interact with the Bayer filter.
    It can generate false colors with high spatial frequencies. Monochrom digital cameras won't have that problem.

    The explanation for the D800 and D800E is here:

    https://cdn-10.nikon-cdn.com/Images...012/Moire-D800-D800E/Media/OLPF_schematic.pdf

    For the D800, the first birefringent sheet separates into two spots depending on polarization. If the incoming light is horizontally or vertically polarized, it will only get one spot.
    (My thought would have been to do it at 45 degrees, but it doesn't say that.)

    Then there is a λ/4 wave plate to convert those spots into circular polarization. The second birefringent sheet separates those into vertical spots.

    If the first step doesn't separate the spots, you could get aliasing in that direction.

    And yes naturally polarized sources will cause effects, though most aren't so strongly polarized as filters.
    And maybe with a little luck, strong polarization and high spatial frequency don't tend to occur together.
    While we all like the super high resolution, not so many subjects have those high spatial frequencies.

    The D800E is interesting. One might guess that they just leave out the birefringent sheets, but I suspect that the rest of the system is based on their thickness.
    So, they remove the λ/4 plate, and replace it with a plain glass sheet, and then put the second birefringent sheet to undo the first one.

    There are then some recommendations to reduce aliasing, such as stop down the lens a few stops, so diffraction will increase the spot size.
     
  5. That's all very well, but it's still possible to provoke moire effects with the D800. For example, by copying half-tone dots at a magnification where the dot spacing is close to the photosite spacing.

    I suspect that the lack of efficacy of any AA filter that doesn't also totally negate a high pixel density, is the reason that manufacturers simply dropped the inclusion of AA filters altogether.

    Whatever. This is wandering way off topic from the linear v circular polariser debate/mythology. The type of which, as shown above, seems to make very little difference to either camera operation or sensor image.

    And I'm not about to buy a Canon Pellix or Sony a77 to find exceptions to the rule.

    BTW, kudos for finding the Alt-graphics key combo to get the lambda symbol!
     
  6. Well, also that the site spacing is approaching the lens resolution.
    As Nikon notes in the linked article, stop down a few stops, and diffraction will do it.

    My dad bought a Pellix when they were new. Some years later, it was stolen out of the trunk of the car.

    About 6 years ago, I got one from a Goodwill auction for $44. The mirror is a little dusty, though.

    The Pellix manual doesn't mention polarizing filters. I was trying to remember if I knew that years ago.

    Oh, the lambda comes from a web search, and then cut/paste.
     
  7. Nah! Most Planar-clone standard 50mm lenses will resolve over 200 lppmm @ f/5.6 and there are no 140 megapixel full-frame sensors that'll image that natively without using pixel-shift.
    Or, for flat artwork copying, just throw the focus off a tad.

    In fact Nikon's flaky AF might do that for you automatically. ;)
     

Share This Page

1111