This has been a wonderful thread and the totality of the discussion including the below email from Erik Allin, Canon Professional, has caused me to compromise my use of a UV filter. In clean environments I will no longer use a UV filter; in questionable environments - shooting near the surf, waterfalls, places where the air is heavy with water, very dusty Australian locations :-), and most of the time in Antarctica - I will use a UV filter.
Due to the vast differences in build and image quality amongst the different UV and protective filters on the market, you will not find any more of an “official Canon” response than what you have already received or read in a manual. The idea that “Either they do or do not recommend” and the search for a blanket, all encompassing Canon statement on the matter is unrealistically simplistic.
As such, I will NOT be giving you an official Canon position, but my own official PERSONAL position on the subject.
ANY and ALL UV/protection filters have some degree of negative impact on image quality. Some more than others. With some extremely high-quality multi-coated UV filters – typically very expensive – the IQ impact is so negligible as to be unnoticeable in the finished printed image to most people. Most UV filters exhibit some IQ degradation that can be seen to some degree in the image. Some UV filters can be quite bad.
If you should choose to use a UV filter, the need for one of a high quality is far more important with digital than it ever was in the film days.
Canon recommends the use of a filter, to enhance the weather-resistant characteristics on certain weather-resistant lenses– the EF 16-35, 2.8 L II as an example where the front of the lens moves while focusing or zooming in relation to the barrel of the lens. Canon has no such filter recommendation on lenses where the front element does not move and the barrel can be better sealed – the EF 70-200, 2.8L IS as an example. To one of your emails, and a comment on the forum you quote from, Canon has not “downgraded their recommendation”; the lenses are different in their mechanical design, and as such the recommendation is different.
In this case the recommendation of the use of a filter is based solely on the weather-resistance characteristics of that specific lens and not on enhancing the image or protecting the front element from impact or damage.
I personally do not use any UV filters.
In terms of protecting the lens – or “your investment” as the camera store sales person will phrase it – a matched lens hood is far more protection from damage and impact than any filter ever will be. And a lens hood has NO negative impact on IQ, and in most instances has a positive impact on IQ. Buying, AND USING, a lens hood is the best investment one can make on protecting “your investment” and improving the IQ of your images.
Buy the matched Canon lens hood, and use it.
Canon’s branded UV/protection filters are OK, but nothing special. They are provided in the catalog as a convenience so that some smaller camera dealers that carry Canon product, but may be too small or out of the way so that a sales representative from a filter company may not visit, can have filters to sell, should they chose to.
To the question posed in your emails “if they are not necessary, why sell them to the public?”: because some people want to buy them and some stores want to sell them.
Having said all that, Canon brand Circular Polarizers are exceptional and are probably some of the best filters on the market from any company – again, personal opinion.
I asked Erik whether in "dirty" environments and especially with lenses that have a moving front element whether he uses a UV filter. I will provide his answer.
email from Erik Allin:
"ANY and ALL UV/protection filters have some degree of negative impact on image quality. Some more than others. With some extremely high-quality multi-coated UV filters – typically very expensive – the IQ impact is so negligible as to be unnoticeable in the finished printed image to most people. Most UV filters exhibit some IQ degradation that can be seen to some degree in the image. Some UV filters can be quite bad."
I agree with this statement with one caveat: for a high quality filter the IQ degradation is usually not measurable within the noise of the image except in rare circumstances. I'll remove protective filters when I'm doing photography that includes bright subjects, like a moon in deep twilight, or lightning. I do a lot of night city scene images, especially when I'm on travel. (e.g., Hey Jay, check out my Sydney Opera House image at http://www.clarkvision.com/photoinfo...ht.photography it was used by a Sydney advertising firm--funny, it seems like they could have gotten their own image). Since switching to multi-coated filters, I've not had a problem, so sometimes I forget to take off the filter and have never had reflections or other problems.
But this thread has been illuminating. I'll try some tests over the next few weeks with filters and see if I can find problem filters. I have a variety of test methods that I can design that anyone can use.
The problem people seem to be observing with filters on telephoto lenses is probably due to the following. Filters are very thin, and on small lenses, the apertures are very small. A filter must be flat and uniform to at least 1/2 wavelength of light--that's about 1/5 of a micron. A 50 mm f/1.4 lens has an aperture of 35 mm but image quality is limited by other aberrations. Stop down to f/8 and the aperture is only 6.25 mm, so the filter performance needs to be 1/2 wave only over a 6.25 mm spot, which is fairly easy (good quality window plate glass meets that spec). But a 400 mm f/5.6 lens has a 71 mm aperture, so the 1/2 wavelength criterion is over a much larger area and harder to achieve. It is this larger area of required flatness that probably creates the problem, rather than the magnification of the long focal length.
So again, if anyone has a filter that they believe is causing a problem of soft focus, I would like to examine it. I'll even return it if you wish. Contact me by email or PM. Roger
Reply With Quote
