UV filters

[Roger_Provins]

I don't use them but for those who do this may be well worth reading - preferably before you buy

[Lazza]

Just bought one for a new lens today, knew (liar!) that the B+W would be a great choice!

Tbh I usually use a Marumi filter as a protector on all my lenses especially considering their lower price and find them great and as the test clearly shows/proves they are a very good buy especially when you consider the high price of top end Hoya or B+W filters. As this was a Leica lens though thought it deserved a treat.

Thanks for the link!

**Edit

Just noticed the result for Tiffen - Now that *is* a shock.

[huwevans]

An interesting find, which I've just spent a good few minutes perusing.

I don't use UV filters either, but FWIW I would sound a note of caution about the tests presented here, and that's to say that some of their methodology for assigning scores looks more than a little arbitrary. For instance, their score for visible light transmission falls from a maximum of 10 right down to zero if the filter transmits only 90% in the 390-750nm band. That sounds fine enough, until you stop to think that a drop from 100% to 90% transmission represents a mere 1/6th of a stop - an amount that barely matters in practice.

Similar comments might be made about the scoring of UV transmission. In point of fact, they seem to have designed a test which is really hyper-critical of the precise cut-off wavelength and roll-off, and how well they fit an assumed 390nm ideal. And that's the bit I'd quibble - 390nm is itself a little bit arbitrary, and a filter which cuts off at, say, 420 with a steep roll-off, or 380 with a shallower roll-off, is going to be radically compromised, according to this scoring system. And TBH, I'd say that it's not really obvious that either is definitely bad.

However, looking through the individual tests, they do seem to have pinned down any real problems with flare and ghosting, and I'd definitely pay attention to that part of the test if I were contemplating any purchases.

Finally, they are pretty hard on the expensive offerings from B+W and Heliopan, and especially in their 'Cost to Benefit' ranking. But at least part of the high cost of these brands lies in their use of higher quality brass rings (which don't bind, the way that cheaper alloy rings tend to) - something that doesn't appear to play any part in the benefit analysis, but which obviously counts against them by pushing the price up.

[GeoffR]

I don't normally like using UV filters but my 17-55 f2.8 had one when I bought it so I plan to run a few experiments with and without the filter to see what practical difference it makes.

Mine is a Hoya so I do now know that it is a good one. At least as far as that test goes.

[jeallen01]

Historical conservatism has almost always meant that I buy Hoya filters (and I have a UV on almost everything) - so I quite pleased with the results!

[beejaybee]

390nm is itself a little bit arbitrary, and a filter which cuts off at, say, 420 with a steep roll-off, or 380 with a shallower roll-off, is going to be radically compromised, according to this scoring system. And TBH, I'd say that it's not really obvious that either is definitely bad.

This does of course depend ... not least on the colour correction of the lens in question. Many lenses cease to be achromatic in the violet / UV region, passing a significant amount of "invisible" extreme violet / near UV at say the calcium K line (393 nm) may result in unacceptable chromatic aberration (or "blue bloat" as astronomers call it).

OTOH if you want to image in CaK emission, having an optical system that passes 393nm is critical ... Just in case you think this is unimportant, well it may be to you, but I made this image of the Sun at 1338 UT today 12 May 2009 in calcium K emission.

[huwevans]

Certainly - it's going to be very application-specific. And that's really why I can't go with the idea of a scoring system which arbitrarily designates a shallow roll-off and/or a cut-off that is above or below 390nm as being less good than one which delivers a theoretically perfect cut-off at exactly 390nm. A filter which gives a shallow roll-off that begins at, say, 420nm, could, in theory, be better for general photography than one which passes 100% all the way down to 390nm and then cuts perfectly - although granted it would be no use at all for your CaK imaging application.

That said, personally I haven't used a specifically UV filter for the better part of 30 years, and have never experienced any problem that I could have attributed to excess UV transmission, so really I do wonder how much it matters in practice. These days, I think most people who do use them do so for their (alleged) lens protection capabilities, which, again, aren't addressed in the test.

Incidentally, most Schneider lenses (the only manufacturer I'm aware of which publishes this information) have transmissions that are nowhere near as flat as even the worst filters in that test, and which begin to roll off steeply at about 420nm. Some of the very best of them - like the Apo-Componon enlarging lenses, or the Super Symmar XLs, or Super Angulon XLs - some of the undisputed stars of the large format world - have the worst of all, and are often down to well under 80% transmission by 400nm, from peaks in the high 90s.

So, anyway, FWIW, rather than quibbling over a few percentage points more or less transmission in the visible, I would have quite appreciated a test which addressed the optical flatness of the filters, carried out at small apertures - even if it only reassured us all that there was little practical difference.

[jeallen01]

I don't use them but for those who do this may be well worth reading - preferably before you buy

The link above may not now work, but this one should:
http://www.lenstip.com/113.1-article...ters_test.html