UV filter and strange bokeh

<p>I have mentioned this previously on a thread in this forum, but now I have some (not very) pretty pictures to go with some words. </p>

<p>My colleague got into photography recently and bought a 450D, 18-55 and 55-250. He wanted more reach without paying canon prices, and bought a Sigma 150-500 OS. On both the 55-250 and the 150-500 we've seen a strange effect of diagonal lines in the bokeh. To begin with it seemed like camera shake, and was much more prevalent at the long end of the lens (in both cases). The direction seemed consistent (and after a while far too consistent for camera shake). </p>

<p>Here's an example of how the bokeh looked (this is one of the test shots from our final diagnosis of the problem):<br>

<img src="http://www.askisaac.com/images/500_1.JPG" alt="" /><br>

As you can see, the bokeh shows strange diagonal lines, but the focused subject remains free of any camera shake problem. </p>

<p>Turning the filter a bit changed it to vertical:<br>

<img src="http://www.askisaac.com/images/500_2.JPG" alt="" /></p>

<p>And removing the filter entirely:</p>

<p><img src="http://www.askisaac.com/images/500_3.JPG" alt="" /></p>

<p>The effect, and solution, were identical on his 55-250 lens also. </p>

<p>This is not an effect that I've seen before, although I think it is probably tied in with cheaper filters (I use canon and Hoya HMC Pro filters, my colleague use(d) Jessops and Kenko filters). However, I do have a few shots that have displayed "odd" bokeh and I wonder if it might not be a similar effect. </p>

<p>Hopefully these images might help someone else struggling with the same issue. </p>

<blockquote>

<p> </p>

<p > </p>

 

</blockquote>

<p>I rarely use UV's or any filters as the "norm" on my lenses unless I am in extremely harsh conditions for the reasons your test shots are showing. Then I will use mostly B&W filters. However, I have also had pretty good look with Hoya HMC's.</p>

<p>I don't want to start a filter vs. no filter thread here . . . just my opinion!<br>

Almost any filter (used primarily for protection) can and will have an effect on the image. Some good, some bad.<br>

I depend more on the lens hood for protection of my lenses.</p>

<p>I will 2nd James. And your images are just further proof: <em>any </em>filter will degrade an image. It's just a matter of whether the degradation is minimal (or inconsequential) to the effect the filter provides. Meaning, a GND provides a very useful effect when needed. For a UV filter, the effect is protecting the front lens element. If you are shooting on beaches or other venues where sand, mud, dirt, etc might be thrown at the glass, a UV filter may be important. As a general protective device, I am against them! If I must use a filter, then I will try and get the best I can. B + W are what I usually end up with.</p>

<p>I'll try not get into the pro- vs. anti-filter argument, which has already started.</p>

<p>I have taken many thousands of photos with good UV filters, with several cameras and all sorts of lenses, and have never observed this phenomenon. My hunch is that it is one or two factors, maybe together: placement of the light, and a cheap filter. The one time that any filter is likely to degrade an image is if the light source is in front of the lens. The shadows suggest that the light source is above and in front. In that case, a cheap filter will be much worse because of flare and hazing--the point of an expensive, multi-coated filter, just like an expensive coated front lens element, is to reduce flare. However, I have never seen the degradation take the form you found.</p>

<p>Most of the shots at one of my sites, http://dkoretz.smugmug.com, were taken with Hoya HMC or S-HMC UV filters on the lens. At http://dkoretz.smugmug.com/Other/test-shots/9295844_q6Paq#627108121_8jhHi, you will find two shots of a bookshelf. One has a UV, the other doesn't. You can blow them up to full size if you want. I can't recall (or figure out) which is which. However, even with good filters, if the source is in front of the lens, there can be problems, particularly if you are not using a hood.</p>

<p>The point here isn't the filter vs no filter argument, rather that this is what may be seen with a bad filter. Also, this is a separate issue and argument to that of flare. </p>

<p>What little mention I have found on the web of similar phenomina seem to centre around the big sigma lenses - eg 150-500, 50-500. I think that with these lenses several factors combine to make the problem more prevalent:</p>

<p>1) Price. People buy the big sigma zooms because they're well priced for the focal length that they offer. The 150-500 is £700 in the UK, vs over £1200 for a canon 100-400. With that in mind, they are not going to spend the £100+ that a high quality filter costs. <br>

2) Size. The sigmas use 86mm filters, which as per point 1 are expensive, but also they are bigger and mechanically more difficult to keep flat than a smaller filter. <br>

3) Magnification. I think the effect is magnified by the lens. Other manufacturers lenses for 500mm may well not even have a filter thread on the front of the lens (eg canon 500 F4, etc)</p>

<p>I'm not about to take all my filters off simply because I have not had this problem for the most part. There are two instances of pictures I can think of where I've seen strange bokeh - one was with my 300 F4L IS + 1.4X TC, the other my father's 100-400. With reference to point 3 above, and the effect seemingly worse with long lenses, I might try using the 300 without its protect filter (since it has one built-in also... however, a separate one is easier to clean (eg ultrasonic tank)). However, I've been saved a more expensive repair bill by filters on two occasions, so they will remain.</p>

<p>What diagonal lines? All I see is random blotches?</p>

<p>Carl Sagan once said, "there is no question that the canals on Mars are the result of intelligent behavior, the only question is on which end of the telescope is the intelligence?"</p>

<p>Time for new glasses then JDM, or a better monitor, they are very obvious in the top two images. I would be interested to see the same images taken with a lens hood in place though. I believe it is a flare issue combined with the cheap filter and lens.</p>

<p>If you think it might be the filter do some tests with it removed. However that said I have never seen a filter cause anything close to this. The diagnal lines (most visible in the first photo could be from camera shake. At 500mm with a APS-C camera you should keep your shutter at 1/800 of a second or faster (without IS).I would do some tests with the camera on a tripod to ellliminate camera shake. The other possibility is that there may be an issue with the Sigma OS system. I would try some photos with and without IS to determine is IS is working correctly. In short you need to run some tests where each posssible cause is ellliminated so that you can determine the cuase. Once you have identified the cuase you can determine what needs to be done to fix it.</p>

<p>I have viewed it on three different monitors now on three different computers, and none of them are bad monitors, either. My eyes are just checked too. Still just can't see it, can somebody put arrows pointing to them or something?</p>

<p>Curious minds want to know.</p>

<p>Never mind, I finally see them- I was looking for something much larger than these little parallel lines.</p>

 

<p>Here's the arrow I asked for, although it's really not too clear in the slightly reduced post below.</p>

<p>I wonder if this is not more an artifact of image processing though?</p><div></div>

<p>I was working on that for you JDM :-)</p><div></div>

<p>Steven,</p>

<p>The tests were done, it is the filter, as proven by the third picture taken with no filter, all else identical but showing no lines.</p>

<p>I'm with Scott. All I can think of is some sort of flare issue. Perhaps there is some sort of streak on the filter that causes this? Whatever it is, it's something that can be rotated. The elimination of the problem with the removal of the filter certainly nails the filter as the culprit, and not camera shake.</p>

<p>What sort of cheap filter was this? Are we talking no-name Chinese/Indian, or something more up-scale like a Kenko or Hoya monocoat?</p>

<p>It would be interesting to repeat this test with and without lens shading.</p>

<p>It would also be interesting to repeat the test with the cheap filter and then with a good filter -- even if the good filter is too small to mount and is just held in place in your hand.</p>

<p>Finally, a shot of a light source (e.g. a chandalier) would address the streak/flare issue.</p>

 

<p>I still can't see how the filter alone could produce this - but I could imagine some flaw in the filter which reacted with the pixel massaging in the camera or application?</p>

<p>As Gene M. once said,<br /> "It's a mystery"</p>

<p>Addressing a few points raised:</p>

<p>Image processing artefact. The above images are large fine jpgs out of camera, downsized to 20% in MS office picture manager. No other processing applied. The effect has also been seen in RAW files. </p>

<p>Camera shake. I rejected some images not shown here because of camera shake. However, this effect has also been seen at higher shutter speeds and shorter focal lengths (eg efs55-250 @ 1/500th). Also not shown was a further shot taken after the vertical shot when the filter was rotated another 90 degrees and lo and behold the lines went horizontal. </p>

<p>Flare issue. It is a possibility, but seems unlikely that the effect would rotate with the filter if a flare issue. There was no linear grease wiping across the filters (which were brand new) or anything like that. </p>

<p>As for the mechanism, I think it's the flatness of the glass. Look at the red image in this filter test: <a href="http://translate.google.co.uk/translate?u=http://www.optyczne.pl/9.2-Inne_testy-Test_filtr%C3%B3w_UV_-_uzupe%C5%82nienie_Cokin_UV_72_mm.html&sl=pl&tl=en&hl=&ie=UTF-8">http://translate.google.co.uk/translate?u=http://www.optyczne.pl/9.2-Inne_testy-Test_filtrów_UV_-_uzupełnienie_Cokin_UV_72_mm.html&sl=pl&tl=en&hl=&ie=UTF-8</a></p>

<blockquote>

<p>Image processing artefact. The above images are large fine jpgs out of camera, downsized to 20% in MS office picture manager. No other processing applied.</p>

</blockquote>

<p>"No other" is not the same as "none at all". Regardless of RAW or not, the camera is still converting the signal from the sensor into a picture.</p>

<p>I don't know <em>that</em> is what is happening, but the regular and repeating character of the effect seems more likely to occur from some kind of electronic conversion than from something in the glass of the filter itself.</p>

<p>I think it is a filter-lens issue. It might not be a straight filter quality is the issue here though. It could be a problem caused by the lens interacting with the filter. Having the best filter (I like Hoya) only means that it's surfaces should be very flat, and well coated. It is still possible that the filter used here is defective, and either not flat, or stressed in it's mount to the point of slight distortion, which can cause plenty of problems by the time the light makes it to the image plane.</p>

<p>Even if the filter is otherwise perfect, it's two glass surfaces still reflect some of the light passing through (both to the outside, and to the inside of the lens, and at various angles), and they add to the number of reflective glass surfaces already built into the lens with it's element stack. It is also possible that if even one element or structure inside the lens is badly coated, malformed, left bright, or who knows what else, the result can be the funky image effect when a filter is in place on that lens. Sigma builds some nice lenses, but they often use unorthodox design methods to get there, and there have been at least two or three Sigma lenses over the years that were delivered to retail outlets in less than perfect condition....</p>

<p>I would try to duplicate the problem with borrowed filters, and then see if it could made to follow a filter on a similar lens, or even happen with any filter in the same situation with a similar lens. A little detective work should pay off...</p>

<p>I never had any noticeable IQ issues with high quality UV filters in the past, although I stopped using them as "protective" elements a number of years ago. I most always use hoods, and I just don't see how any possible good can come from adding two more glass surfaces to a lens unless it is for a desired image effect. Still, there are millions of filters in use for just that purpose, and you shouldn't expect obvious image problems if you choose that route.</p>

Hello all,

 

Isaac Gibson is very very likely right. I was alerted by a user on DPreview that my own "strange" bokeh problem might have to do with the filter. I then searched the Net and found this thread. Thus I created an account to confirm. Although ten years ago, proof is never too late.

Here is my own experience: THIS is what I got, using a Kenko UV filter on a Nikon Coolpix P900:

I believe the diagonal lines are rather obvious.

 

 

 

Then take this one, also with Kenko UV on same camera, notice the little stripes, creating a sort of "hatching" effect in the out of focus grass:

 

 

Then this photo, taken WITHOUT the Kenko culprit - as you can see the bokeh of the grass is as it should be:

 

Hello all,

 

Isaac Gibson is very very likely right. I was alerted by a user on DPreview that my own "strange" bokeh problem might have to do with the filter. I then searched the Net and found this thread. Thus I created an account to confirm. Although ten years ago, proof is never too late.

Here is my own experience: THIS is what I got, using a Kenko UV filter on a Nikon Coolpix P900:

I believe the diagonal lines are rather obvious.

 

Then take this one, also with Kenko UV on same camera, notice the little stripes, creating a sort of "hatching" effect in the out of focus grass:

 

 

Then this photo, taken WITHOUT the Kenko culprit - as you can see the bokeh of the grass is as it should be:

 

[ATTACH=full]1333364[/ATTACH]

 

 

I say that confirms it and if anyone has this problem, this is the easiest thing to confirm or rule out. Take a photo with and without the filter.

When I first got into shooting with DSLRs I bought a UV filter, mostly because I didn't know any better. I came to the realization that the lenses I use have great coatings on their own, and the glass quality far excedes a cheap $100 filter. Why pay $700-$2500 for a high quality lens and stick some cheap piece of glass in front of it.

 

On a rare occasion I may use a circular polarized filter if shooting into water or through glass windows and I have used ND filters shooting the eclipse and a few times experimenting using manual flash in day light with fast lenses wide open.

 

But, 99% of the time I don't use lens filters. It is just a choice.

 

Here is an entertaining video by Tony Northrup about UV filters and I believe he explains it well.

;) It's just a choice. Why would you want that funky bokeh? :cool:

Hello Mark,

 

Thank you for your post and the link - I must say, I winched, seeing Northrup hacking at his canon lens with a vengeance... Anyhow, perhaps the scratches on that lens did not affect later photos, but the lens on my Canon Powershot G9 most certainly did not perform well after being damaged, a big, deep scratch - all photos after the damage showed a slightly over-exposed area directly under the scratch.

 

Cheers,

 

Mabel A.

Oops, spelling correction in my text, frist line: "winched" should of course be "winced".

Sorry, I have to be pedantic - 'frist' ?

 

Welcome aboard, anyway - very interesting 'effect' !

Good evening Tony,

 

o_O Oops.... spelling correction in my spelling correction.. "frist" should of course be "first".

 

Lol ;-)

 

Yeah, the effect is "interesting" and totally ruinous for the photos....

Nice subject matter, anyway - and the lines in the image of the stork could almost be rain in the background ?

Rain on a radiant, sunny day.... ? Rain is the bane of photographers....