What is this swirly bokeh?

[holly_swanson]

<p>After performing completely random test shots in my backyard with a 5d Mark III, I am a bit confused at what I'm finding. The lenses I used were a 50 1.2 and the 135 2.0, both of which have never given me this result when using a 6D.<br>

I do not know if there is a term for it but I think it is best described as swirly bokeh... that's the closest description I can come up with. Take a look. It is in the foreground of all these photos and looks godawful.</p>

<p>Shot with 135 2: http://postimg.org/image/e1ep6bamn/<br>

Shot with 50 1.2: http://postimg.org/image/9drr6z757/<br>

Shot with 50 1.2: http://postimg.org/image/h9b2h0r03/</p>

<p>That is 3 examples. Very horrible if I do say so myself! What is going on here?<br>

These lenses were both being shot wide open at ISO 100 all at speeds over 1/1200. </p>

 

[lex_jenkins]

<p>It's not uncommon, especially with fast lenses wide open. Shoot exactly the same scenes with any Canon full frame dSLR (or 35mm film SLR) and you'll see the same effect. If you haven't seen it before on your 6D: you may not have been close enough to the ground; or didn't include enough out-of-focus foreground in the photo; or weren't shooting wide open.</p>

<p>And if you didn't see it with APS sensor cameras it's because most of the effect is cropped out. Same problem I have with my special effect and soft focus lenses on cropped sensor digital cameras. The lenses were designed for a larger format where the effect would show at the periphery.</p>

<p>Google: <a href="https://www.google.com/search?q=html+link+for+image&safe=off&rlz=1C2LENN_enUS490US562&biw=1040&bih=596&ei=gx1PU7LJNc2dyATV_4HQDg&sqi=2&start=10&sa=N#q=petzval+swirly+bokeh&safe=off"><strong>Petzval lens swirly bokeh</strong></a><br>

The effect is similar, but not identical, to the older lenses like the Petzval, which tended to emphasize the swirls at or near infinity, rather than at close range as you're seeing with your lenses. It was considered a desirable effect in that era. Some photographers still like the effect and have adapted Petzvals to their cameras. But the effect is often minimal unless used on a large format camera. There may be a Lensbaby doodad designed with a smaller image circle to emulate that effect on smaller sensor digital cameras.</p>

<p>Also, <a href="http://www.dpreview.com/forums/thread/3613819#forum-post-52993410">other owners of the Canon 50/1.2 have observed the same effect</a>. And <a href="https://www.flickr.com/groups/20778516@N00/discuss/72157623294611124/?search=swirly+bokeh">another discussion here</a>.</p>

 

[pjmeade]

<p>It seems to be something that's only noticed occasionally in our run of the mill photography, but that's exactly the same background blur/bokeh I've noticed with 400/2.8+1.4x trying to shoot something that's ~4-5m away from a low angle. <br>

As an aside, I called in at the London Lomography gallery/shop recently and they were selling Petzval lenses.<br>

http://shop.lomography.com/gb/lenses/brass-petzval-canon-mount </p>

[Wouter Willemse]

<p>Awful or not is a matter of personal preference, but in the foreground, it is quite distracting; in the background a lot less so to me. Given that both your bodies are full-frame, it is hard to explain why on one body it would show, and on the other not. It should be identical, as Lex wrote.</p>

<p>I'm not optics expert (there are some around, I hope they chime in), but I sometimes do see a similar effect with my Nikon 50mm f/1.2, and to a lesser extend 35 f/1.4 (old manual focus lens - not the current model). Both lenses are not corrected all too well wide open, which under conditions can lead to this different, slightly nervous OoF rendering. I happen to like it in the background blur sometimes, and like these older lenses I have for "defects" such as these - they add a bit of flavour rather than the clinical edge-to-edge sharpness (though it makes the lenses a bit less predictable too - not always an advantage). <br>

Anyway, as far as I understand, it's called <a href="http://en.wikipedia.org/wiki/Spherical_aberration">spherical aberration</a>, but I might be off.</p>

[Robin Smith]

<p>It is all connected with the subject distance, the composition of the foreground/background, and the unsatisfactory light. There's nothing wrong with the lenses - you just have to chalk it up to experience and come to the conclusion that this light, location, time of day, and choice of aperture was not optimum. Even lenses with "good bokeh" will not always produce it. Bokeh probably would have been better in this instance if the lens was stopped down further.</p>

[bobatkins]

<p>Take a look at my review of the Helios 85/1.5 which shows the "swirly bokeh" effect. The cause isn't 100% clear but it seems to be a combination of lenticular highlights near the edges of the image due to vignetting plus significant astigmatism in the out-of-focus areas.</p>

<p>Some people buy certain lenses especially for this effect!</p>

[sjmurray]

<p>I'll second the comments by Robin. Bokeh is highly affected by lighting, distance, contrast, aperture, etc. Also, as I understand it, all fast lenses when shot wide open have some spherical aberration, as Walter has pointed out. Learning to use a lens in the extreme wide open setting takes some practice. As Lex and Walter have noted as well, its hard to explain why a different full frame camera would not show the same results! For me, all lenses have their own distinct characteristics, much like having a set of different "paintbrushes." Each is used for their unique qualities. </p>

[holly_swanson]

<p>Well I tried it again today in better light, with the 50 1.2. The curved foreground bokeh is still present, as you can see at the bottom of this image, and still looks very odd to me.<br /> http://postimg.org/image/y3fuwq2h7/<br /> I know it probably has to do with many other issues but I just want to make sure it couldn't possibly be this different camera. If it could be, that's something I need to know. Otherwise I will just keep working on it. I really had never noticed this before with the 6D, I'm not sure why...that's pretty coincidental to me that it coincides with a body change.<br /> Thanks for the replies!</p>

[laur1]

<blockquote>

<p>Anyway, as far as I understand, it's called <a href="http://en.wikipedia.org/wiki/Spherical_aberration" rel="nofollow" target="_blank">spherical aberration</a>, but I might be off.</p>

</blockquote>

<p>I thought it's due to coma, but I've also seen one other person assign the effect to spherical aberration. I would like to see a definitive answer.<br>

<br>

I sometimes get this with my MFT Voigtlander lenses when used wide open. Here's a shot that emphasizes this effect at close to minimum focusing distance:<br>

<a href="https://www.flickr.com/photos/laurentiucristofor/7213381708/"><img src="https://farm6.staticflickr.com/5348/7213381708_60a172ab8e.jpg" alt="" width="500" height="375" /></a></p>

[gerry b.]

<p>For more extreme examples of this phenomenon, do an image search on "Summitar" and "bokeh". Summitars were Leica fast lenses from the 1940s and 50s that had a ferocious "swirly bokeh", and as alluded to above, some people use them in part to exploit that effect.</p>

[sjmurray]

<p>Holly, your last photo is very nice. One difference from the first three you posted is that it has less foreground, which is the "swirly" part. I think using this lens wide open you get the best results minimizing the foreground in the picture, unless, of course, you are trying to incorporate a lot of distortion in the image as part of overall effect. I do a lot of portraits with the 50mm lens, but I tend to shoot at f 2.5, example here:<br /> http://www.photo.net/photo/11254730 or here http://www.photo.net/photo/13749713 or here http://www.photo.net/photo/14489673<br /> which nicely blurs the background and leaves the person's features in focus, and I don't usually include any foreground, but that is my personal style.</p>

[aplumpton]

<p>Why you get it with the 5D and not the 6D is curious. There must be other factors that were different between the two experiences. Using some lenses wide open can give odd results as they are not usually at their performing best at full aperture, such that out of focus blur (bo-ke) may be altered by the other aberrations that are present wide open. As some have said, you would be better I think to use one or two stops down from max when shooting so close. Anyway thanks for showing. The effect is, as you say, omini-presently awful.</p>

[bob_bill]

<p>I think Steve nailed it. In the real world, I rarely take such a shot with that much oof in the foreground. But one other commonality of those shots is bulls- eye of the subject. I wonder if you composed using rule of thirds if that would eliminate the even amount of space around the subject that could contribute to the pattern being more obvious. Or, perhaps switch to Nikon. Just kiddin.</p>

[q.g._de_bakker]

This will be equally visible in a blurred distant background, so avoiding much close foreground will not help. Compositionally, the workaround is to have featureless things in the out of focus areas, such as a blue sky. Anything else will show the swirl, the more finely textured, the more obvious the swirl will be.

[bob_bill]

<p>Converted this to b&w and still don't see any semblance of swirl. </p><div></div>

[sarah_fox]

<blockquote>

<p>What is going on here?</p>

</blockquote>

<p>It's a consequence of the clipped shape of the light paths at oblique angles through the lens. Consider this photo, taken with a 100/2 at f/2.0 on a 5D:<br /> <img src="http://graphic-fusion.com/portraituresample.jpg" alt="" width="325" height="406" /></p>

<p>Note that the OOF lights in the background have a circular bokeh pattern towards the center of the frame and a clipped, oblong bokeh pattern towards the edge. Hold up your unmounted lens, and rotate it through different angles while you observe the light pattern through it, and you will see the same shapes. As you tilt the lens, the spot of light streaming through the lens is clipped on one side by one edge of the front element and on the opposite side by the opposite edge of the rear element.</p>

<p>Anyway, if you have an image with a more complicated OOF background than the simple point sources in my photo, the blurred pattern will form the "swirls" at the edge. One of my little experiments illustrates how the shape of the aperture impacts the bokeh:</p>

<p>http://graphic-fusion.com/beyondbokeh.htm</p>

<p>It should be readily apparent from this how an oblong aperture (at oblique angles) would create the "swirls" towards the edge of the frame.</p>

[studio460]

<p>As I understand it, the phenomena is caused by a kind of "vignetting" inside the lens. I had initially hesitated to buy the (excellent) Sigma 150mm f/2.8 due its characteristic "swirly bokeh," but when I tested the rest of my fast-aperture Nikkors, I was able to generate similar results. Conclusion: All lenses at large apertures exhibit this effect to some degree (i.e., your lens is fine).</p>

[studio460]

<p><img src="http://studio460.com/images/Sigma%20150-V.jpg" alt="" /><br /> Sigma 150mm f/2.8 OS @ maximum aperture on a Nikon D800E.</p>

<p><img src="http://studio460.com/images/Nikkor%2085-V.jpg" alt="" /><br /> AF-S Nikkor 85mm f/1.4G @ maximum aperture on a Nikon D800E.</p>

[bob_bill]

<p>Sarah, great article. Will have to try it. </p>

[tom_mann1]

<p>That is indeed a wonderful article, Sarah. Very innovative.</p>

<p>I never looked over your website before this, but I also liked your article on selecting a wedding pro, and reading a bit about your background. It's no wonder that we tend to have similar responses in photo.net threads over the years.</p>

<p>Best regards,</p>

<p>Tom</p>

<p>PS - BTW, the Nikon 35/2.0 is another great producer of swirly bokeh. See attached.</p><div></div>

[laur1]

<blockquote>

<p>It's a consequence of the clipped shape of the light paths at oblique angles through the lens.</p>

</blockquote>

<p>I am aware of this effect, but I'm not sure if it accounts for all the swirl you can get, because some lenses seem more prone to produce swirl than others.</p>

<p>So I'm wondering if this has to be combined with other optical defects to make it easier to come across. For example, the shape of the outer highlights is very clean in your sample, but if you look at Tom's last sample, the shapes look more deformed in corners and they aren't just elongated along the circumferences, but seem somewhat stretched along the radii as well.</p>

<p>I'd actually like to see a crop of the lower left corner - Tom, could you post it?</p>

<p>For a counter example, check this image - click on it to get a larger version and then examine the lower area - the swirl shows under the kid but it disappears towards the corners because they get closer to the focal plane. But then, at that position in the frame where the swirl shows, I would not expect much clipping of the highlights. This is the Helios-44 on an APS-C camera, btw. And I've compared it with another lens of similar focal length, the Cosina 55/1.2, but that one never gave me the appearance of a swirl, even though its faster speed might suggest it's a better swirl factory.<br>

<a href="https://www.flickr.com/photos/laurentiucristofor/13207867393"><img src="https://farm4.staticflickr.com/3704/13207867393_3271e4b6bb.jpg" alt="" width="335" height="500" /></a></p>

 

[tom_mann1]

<p>LC: "I'd actually like to see a crop of the lower left corner - Tom, could you post it?"</p>

<p>When I get home late tonight, I'll be happy to do so.</p>

<p>T</p>

[JDMvW]

<p>Now that the serious answers have been given.</p>

<p>Have you considered the possibility that the mass of the dog was causing gravitational lensing? http://hubblesite.org/hubble_discoveries/10th/photos/graphics/slide36high.jpg</p>

<p>BTW, this is not <em>entirely</em> off-topic .</p>

[tom_mann1]

<p>It does look a bit like that, doesn't it, JDM. :-)</p>

<p>T</p>

[lex_jenkins]

<blockquote>

<p>"...Cosina 55/1.2, but that one never gave me the appearance of a swirl, even though its faster speed might suggest it's a better swirl factory."</p>

</blockquote>

<p>If the image circle of that lens was designed for 35mm film coverage you'd need to test it on full frame film or digital to see the effect. APS and smaller sensors/film formats crop out the periphery where most flaws, or in the case of specialty lenses, desired soft focus effects, are visible.<br>

<br>

The occasional renewal of interest in Petzval and other special effects lenses often fails to account for the image circle and original intended coverage. Used on 35mm film or smaller sensors, some classic special effects lenses pro merely look mushy and uninteresting because the softer periphery is cropped out. These older lenses were designed for large format film. As with many things in photography - "sharpness", printing DPI, depth of field, etc. - these effects are relative to the entirety of the photograph and viewer's subjective impression. Viewed in the intended context, a lens with soft edges may still appear adequately "sharp" in the center - but only if the viewer can see the entire image. Crop out and enlarge only the center, and it looks mushy.</p>