Many photographers leave UV filters on their lenses more-or-less permanently. Many others do not. Who is right? Should you buy UV filters (or clear protective filters) for all your lenses?
This question is highly controversial amongst photographers and the forums of DPREVIEW have seen many long discussions/arguments on the subject. The only thing we can say for sure is that there is no definitive right answer. In this article, I will attempt to explain my own point of view and discuss the main arguments for and against using UV filters. Most of these arguments apply equally to clear protective filters.
I assume that the UV filter being used is of similar optical quality to the lens elements themselves. This is generally true for the best multi-coated filters from good manufacturers, but may not be true for the cheaper filters.
The filter blocks UV light and removes the blue cast from images taken in very bright sunny conditions
This argument is almost completely spurious for modern digital cameras. With old film cameras it was often necessary to use a UV filter because film is extremely sensitive to UV light. However, digital sensors are generally rather insensitive to UV, so the problem doesn't arise to anything like the same extent.
Having said that, I have seen some evidence that for certain lenses a UV filter can reduce the purple fringing caused by longitudinal chromatic aberration. The purple fringing of longitudinal chromatic aberration only occurs in particular circumstances and is not to be confused with the much more common coloured fringing caused by lateral chromatic aberration (most noticeable in the corners of the frame).
My personal view is that these effects are almost always insignificant and do not provide a good enough reason for using UV filters on a regular basis with digital cameras.
The filter provides protection for the lens
There are two types of protection to consider.
Firstly, protection against damage caused by rough handling or dropping the lens/camera -
I doubt if anyone has done a proper scientific study of this, but personal experience suggests that a mishap that damages the filter will probably also damage the lens. I have seen no good evidence that the presence of a filter significantly reduces the chances of seriously damaging the lens.
Secondly, protection against dust, dirt, smears and scratches on the front element of the lens - The presence of a filter on the lens certainly protects the front element, as the dust, dirt, smears and scratches get on the filter instead. Which is preferable?
The filter is flat and easily removed, which makes it much easier to clean. Also, if it does get scratched, or gets so dirty that it is too difficult to clean thoroughly, then it typically costs much less to replace than the lens.
On the other hand, many photographers argue that lenses do not need cleaning very often and the chances of scratching the lens are very low, so it is better to save your money and go without the filter.
The filter causes a loss of image quality
This is true in theory (except possibly in those rare cases of lenses that have been specially designed for use with a filter). However, the loss of image quality is likely to be very small in practice and so the real question becomes: Is the loss of image quality significant to me?
In trying to answer this, there are several different aspects of image quality that need to be considered:
Flare and ghost images
I use the term flare to mean an overall veiling of the image (or parts of the image) due to stray light, while ghost images are secondary images of very bright light sources, usually badly out of focus and sometimes showing extreme coma, astigmatism and chromatic aberration as well.
Both flare and ghost images are caused by unwanted reflections or scattering from the various exposed surfaces within the lens and camera body. The glass surfaces of all the lens elements will contribute, as will the glass surfaces of the filter. Typical modern lenses contain up to 15 or more elements, and the addition of one more element (the filter) is not likely to make much difference in most practical circumstances.
I have never seen any convincing evidence that the presence of a good quality filter increases flare to any noticeable extent with ordinary camera lenses.
However, there is a particular circumstance in which the presence of a filter may cause noticeable ghost images. With some lenses, when used at full aperture (or nearly so), light reflected from the sensor back through the lens may be reflected from the rear surface of the filter back into the camera producing a ghost image on the opposite side of the optical axis. Ghost images of very bright lights are often visible in night shots taken with a very fast lens at full aperture if a filter (any filter, the type is irrelevant) is being used.
Although much fainter than the primary images, they can be very noticeable as they will be in focus if the lens is focussed at infinity and the lights causing them are in focus. These ghost images will disappear if the filter is removed, or if the aperture is reduced sufficiently (i.e. the F-number is increased).
Both the above images are overexposed and this makes the ghost images more noticeable. Even so, it is only the brightest lights that produce ghost images that are bright enough to be seen (and then only when they occur against a relatively dark background). In daytime images, it is extremely rare for ghost images to be noticeable unless the sun is visible in the frame (and the air is clear so the sun shines brightly).
Notice that there is considerable flare near the floodlights and this is the same in the two images. The leftmost floodlight produces particularly strong flare which can be seen both as a fuzziness and spreading of the floodlight itself and also as a purplish haze which seems to surround the crane on the extreme left.
Loss of light
I have never seen any convincing evidence that a good quality UV filter causes noticeable loss of light through the lens. Indeed, that would not be expected as the filter is just one additional glass element and most modern lenses already have at least 7 elements and often twice that number or even more.
Loss of resolution
Again, I have never seen any evidence that this is significant for a good quality filter on normal camera lenses. Good quality filters should have optically flat surfaces that do not disturb the direction of the light rays passing through the filter. If there is any slight variation from optical flatness (as may occur with a very cheap filter), the effect will be most noticeable with extreme telephoto lenses because of their magnifying effect.
My evaluation of the evidence is that there is no really compelling evidence either to use a filter or not, except in a very few situations when it is better not to use a filter to avoid in-focus ghost images.
Personally, I do have UV filters on all my lenses and only remove them in those very rare situations for which I know they may cause ghost images. My main reason for using filters is that I like to keep my lenses very clean and I feel more confident in cleaning the filter than in cleaning the surface of the lens.
However, I think those photographers who choose not to use filters have a sound case as well!
The effect of a dirty lens (or filter)
In comparing images taken with and without a filter, one thing I have noticed in doing the tests is that even a slightly dusty lens occasionally has a noticeably deleterious effect on the image. This only occurs in extreme lighting conditions such as when the sun is shining brightly and is within the image frame, or very close to it. Under such circumstances, light scattered by dust particles on the front element of the lens or on the filter can significantly increase the stray light falling on the image.
It's generally not worth worrying about a little dust on the lens (or filter). In normal circumstances dust on the front element has no visible effect at all. But, if you are shooting into a bright sun or other very bright lights, then it is a good idea to clean your lens (and filter) first.