: Just adding one point on filters, and I do not argue whether or not this flaw is visible in video applications (but it can easely be proven by raytracing) concerns extra spherical distortions.
: Putting a perfect (prefectly flat, parallel, no wavefront distortion...)piece of glass in front of a lens allways add sphericcal aberations (astigmatism, coma..). Only if the glas has "no thickness" or all incoming rays are parallel (object at infinite distance) there is no such distortion. So "thick" filters used in wide angle application have to be suspected.
: Personally I don't care, because I don't use them mainly because of the glare risks.
: Andre

This is a good example of "over-theorizing", methinks...;-)
In practice, with any tough conditions I can think of (and
have tried), with good-quality filters (coated, acceptably
flat, properly-mounted), there is NO image degradation
that can be seen on film as a result of adding a filter,
except under unusual conditions (shooting straight at
sun with long lens, using a poorly-designed zoom lens
with strong point-source lights in a dark field, etc.).
Even things that really should cause problems (using
filters in combination [like a polarizer with a color
filter], resulting in parallel flat surfaces) are STILL
rarelya problem... To not use filters due to
theoretical problems that do not often show in the
"real" world is to miss the opportunities for changing
or improving image character or for offering cheap
protection for lenses...
I'm a nut about lens sharpness, especially edge/corner
sharpness (see my lens evaluation list, on my web site,
under "I babble", at: www.David-Ruether-Photography.com) - I have
checked a LOT of lenses (and lens-filter combinations...).