In this article I would like to present my results for the bokeh comparison I did with my 50mm roundup.
For those who did not read my overview, here’s the list of contenders:
Yongnuo EF 50mm f/1.8
And, because the guys on dpreview once showed a terrible set of portraits to compare the Canon EF 85mm f/1.2L II USM with the Zeiss Otus 85mm f/1.4 that mainly resulted in a discussion about bokeh, I wanted to include my beloved portrait lens too. (At least for some apertures)
Canon EF 85mm f/1.2L II USM
Why do we get Bokeh?
Most lens reviews recommend to use lenses “stopped down” to achieve better pictures. I beg to differ! Why would one carry along a massive chunk of glass – the Canon 85L weighs in at 1 kg – just to use only a small amount of it?
Sure, the sharpness usually isn’t great when shot wide open but photography generally isn’t all about sharpness!
The reason for fast prime lenses not only lies within their capability to be used in low light situations while still being able to keep the iso sensitivity at an acceptable level (for me that’s below iso 800 on any Canon DSLR while I do sometimes allow my Fuji X100s to go up to iso 3200 if it’s necessary for the greater good.)
No, fast primes also offer something that most zoom lenses will never allow you to do. The capability to isolate your subject from the background. To make it pop, give the image that 3D-look and feel. Also, with a fast prime you usually get a look that the common smartphone sharpshooter won’t be able to compete with. You have much more room for creativity with background shapes and colors and don’t have to worry too much about possibly distracting stuff.
The ability to isolate the subject depends on several factors that are interdependent:
– Size of the sensor used
the larger the sensor, the smaller the resulting depth of field and therefore the more background blur
– Aperture size
the larger the aperture (small f-number), the smaller the resulting depth of field and therefore the more background blur
– Field of View
A result of the focal length and the distance to the subject. The longer the focal length and the shorter the distance to your subject the smaller the depth of field will be. This might be the most important thing.
– Subject-Background distance
The further away the background is, the more it will be blurred.
My test is based on a rather short distance – 60cm – to the subject (with most lenses in my roundup the minimum focus distance is 45cm while some could go as close as 35cm) and twice that distance from subject to background. The further away the background gets the more it will be blurred and the larger the bokeh balls would be.
What does the test show?
While it’s important to investigate color rendition, sharpness, contrast, light transmission and handling of a lens, to me one of the most important aspects of fast primes is their ability to isolate the subject from the background and even more important how they render the out of focus areas (the bokeh).
This first test is primarily focused to show how the out of focus highlights (the bokeh balls) look. That gives us an idea as to how the bokeh could look in a setting with harsh contrasts in the background (imagine leaves of a tree).
Another thing that would be nice to see is how the lenses fare in “real life” conditions. Some of my specimen are said to produce “nervous bokeh”, the Helios can do “swirly bokeh” and some are said to render backgrounds in “painterly bokeh”. This test, unfortunately, won’t show any of that but I will hopefully find some time and a suitable subject to make another set of tests outdoors. (It does, however, consume a lot of time to make sure that the exposures and the settings are identical.)
This picture I took with the Color-Ultron a few days ago shows nicely how the background can be blurred to such a “painterly” effect. Shot wide open at f/1.8.
While the subject isolation is rather great and the sharpness wide open is astonishing for a handheld shot of a moving bumblebee all those tiny details on the lavender are rather distracting to the viewers eye. I was about 50cm away from the bumblebee and the lavender bush measured somewhere around 1m in depth.
How did I take the pictures?
Those who know about Dustin Abbott and his amazing lens reviews might already know that my bokeh test was inspired by him. (Scroll down in his review of the new Canon 50mm f/1.8 STM to find his bokeh test setup and a comparison with the old nifty-fifty and the SMC Takumar)
The setup is pretty simple to be honest.
– I set up a black background and hung a simple LED chain (from IKEA for 2 bucks, including batteries) in front of it.
– At a distance of 120cm I set up my couch table, which I’ve built myself, as a platform for my subject.
– On the couch table – which has a top plate made of glass sitting on the dark wood – I placed my subject, a bookend I inherited some time ago. It’s a beautifully crafted globe with a nice latitude scale around it (of course I’ve got two of those globes). I used the number 50 on the scale to set my focus with all the lenses I tested. To the right side of the globe I installed a white reflector to cast some light on the rear side of earth.
– Another 60cm (almost 2 feet) further away I set up my tripod with the Canon 1Ds mk III. I chose this distance for reasons of space limitations, to get reasonably big bokeh balls and because my backdrop isn’t large enough. For the pictures I took with the 85L I stepped back a little further, trying to get the same field of view.
– Behind the camera, camera left, I set up my 5’500K continuous light source and closed the window blinds to make sure that the lighting would remain the same for all contenders. The lamp points toward the rear wall in order to get diffused light. The globe is exclusively lit by that lamp (and to some extent the reflector next to the globe).
Because the lamp isn’t really strong I had to use rather long shutter speeds (2 seconds) and despite using the self-timer and live-view (for focusing and to reduce mirror-shock) I wouldn’t give too much about the sharpness of the latitude scale right now. I will do my sharpness comparison this week too.
Let’s start with the really fast ones shot wide open. In this comparison I only included the Fujinon, the Nikkor and the Takumar all shot at f/1.4 and the Canon 85L shot at f/1.2.
You instantly notice that I had to place the camera with lens differently for the Canon 85L and that even though the globe is smaller in the frame there’s also less of the background included. The reason for that is pretty simple: with the longer focal length (smaller field of view) you get a “compressed” view of the background.
You will also notice that the bokeh ball closest to the globe near the image center is perfectly round on all 50mm lens shots and also the one close to the glass surface remains rather round shaped whereas all the other balls already get a cats-eye shape the further out you get in the frame.
The reason for those cats eyes is again rather simple: the lens suffers from internal vignetting.
It’s even worse on the Canon where there is some rather harsh cutoff at the top and bottom of the balls. I still have no explanation for them.
To me all the bokeh balls in the three 50s look rather similar and I could not pick a personal favorite yet.
Comparison at f/1.8 or f/2
The next set of pictures was shot wide open for the lenses that start at that aperture (Yongnuo, Ultron, Pancolar at f/1.8 and Helios at f/2) and at f/2 for the other contenders.
Another thing you’ll realize is that the top row of 50s (shot wide open!) looks better than the bottom row. While there is optical vignetting at the borders and especially in the corners, the bokeh balls at least are nicely rounded.
In the lower set of pictures the Fujinon with its 6-bladed aperture fares pretty well compared to the Nikkor with 7 blades and the Takumars 8-bladed aperture does great but the Canon (also 8 blades) looks a lot better in my opinion.
Also, while the Canon has no more optical vignetting and the Fujinon got rid of most, the Takumar and the Nikkor still show strongly pronounced cat eyes.
Revenge of the Tessar at f/2.8
The following pictures were taken at f/2.8 with the Tessar 50mm f/2.8 now replacing the Canon 85L in the bottom right corner.
Yep, the Tessar has nicely rounded, though rather small, bokeh balls in the center and strongly pronounced cat eyes in the corners. The Helios (58mm) still renders rather large bokeh balls but also suffers from some optical vignetting while the others are mostly free now.
You can now clearly make out the 6-bladed apertures of the Fujinon and the Color-Ultron and the 7-blades on the Nikkor. The Yongnuo (7 blades), the Pancolar and Takumar (8 blades) still look rather good to me.
Leftover bokeh balls at f/4
To get bokeh balls at f/4 you need the background to be rather far away if your subject isn’t really close. All the lenses have now been stopped down by at least 1 stop. (the speedsters even by 3 stops)
All the other lenses display some kind of star bursts which I find rather unusual.
I’m impressed at how well the Yongnuo fares in this comparison. The bokeh balls can easily compete with the Pancolar and the Takumar and even the Helios.
The Fujinon and the Color-Ultron look better to me than the Nikkor despite it having one more aperture blade. That’s what surprises me the most.
Lastly at f/5.6
f/5.6 is probably the smallest aperture I would ever use these lenses handheld. Firstly, at f/5.6 you don’t really see that much through the viewfinder unless it’s really bright outdoors. For flash portraits this aperture is small enough and the same goes for group shots. If I were to shoot long exposures on a tripod, I would use live-view and might go down to f/8 or even f/11 if necessary but I haven’t done much of these so far.
Bokehlicious comparison at f/2.8
Last but not least, I took near-100% crops from the image centers of all the images shot at f/2.8 and put them together in an almost 36 MP spanning image.
To me the clear winners at f/2.8 are the lenses with 8 aperture blades (the Canon 85L, the Takumar, the Pancolar and the Helios) but also the Tessar looks rather good in the center (could be nice for APS-C or m4/3 cameras, especially considering its MFD of only 35cm!)
I really don’t like what the Nikkor does with the aperture shape so clearly pronounced compared to the Yongnuo. Even the Fujinon and the Color-Ultron look better to me.
When you look at the sharpness at f/2.8, I’d say that there’s nothing to worry about. Even the Tessar looks reasonably sharp wide open.
But the sharpness is subject of another set of tests I will do shortly.
Feel free to leave a comment if you’ve got some questions or would like to voice your own opinion.