Awesome Comparison: Bokeh Shootout between 50mm 0.95 to 400mm 2.8
For my final installment in this series of awesome bokeh shootout, I attempt to answer an age old question that has troubled many before me. Does size matter? In other words, do telephoto lenses render better bokeh or do fast normal lenses?
To answer this question, I assembled a bunch of lenses, ranging from 50mm to 400mm, to see which of the lenses render bokeh the best. The lenses in this shootout include:
Leica 50mm f/0.95 Noctilux
Leica 90mm f/2.0 APO Summicron ASPH
Nikon 135mm f/2.0D DC (set at normal)
Nikon 135mm f/2.0D DC (set at defocus)
Canon 70-200mm f/2.8 USM II (set at 200mm)
Leica 105-280mm f/4.2 Vario
Canon 400mm f/2.8 USM II
Canon 400mm f/2.8 USM II with a Canon Extender 2x (so 800mm)
The camera body used for this test is the Sony A7 MKII. I didn't used a Leica, Canon, or Nikon camera body, because I have already established that the variance in digital capture isn't significantly different, on my previous bokeh shootout. Besides, I wanted to compare these lenses on a level playing field. Plus I've been on a Sony A7 adapting mood lately.
With regards to process of conducting the shootout... in hindsight, I should not have done it around sunset. As you can see from the sample images, the background light changed throughout the course of my shootout. Ideally, I should have done it later at night, when the background lights are more interesting, but timing issues did not make optimization possible.
Also in retrospect, I should have tried to shoot the subject in a more uniform fashion. Now some images have the subject closer to the frame while others further. Though in my defense, at 400mm to 800mm, my office wasn't long enough to do that properly. At 800mm, the subject was forced to stand all the way back to my office window before I was in distance to get focus.
As for the methodology of this shootout, it was pretty basic. Just get the lead eye in focus, and the rest - or rather the bokeh - would fall in place.
So onto the sample images of the shootout.
Leica 50mm f/0.95 Noctilux. The background rendering is more or less the same as in my previous bokeh comparison.
Leica 90mm f/2.0 Summicron. At 90mm, the bokeh looks as pleasant as the Noctilux.
Nikon 135mm f/2.0 DC. At 135mm, the background appears to be getting more blurred out than the Noctilux. But this is largely a factor of blowing up the background, which in turn reduces the background area for the lens to render bokeh.
Nikon 135mm f/2.0 DC - Defocused Used. With the defocus control used, the bokeh gets a bit more of a blur than without it.
Canon 70-200mm f/2.8 USM II set at 200mm. This is the standard for most portrait rendering.
Leica 105-280mm R Vario Elmarit set at 280mm. The result of this lens is most interesting. The bokeh is pleasant, but the overall image is hazy and lacks contrast. It almost reminds me of the Canon 50mm f/1.0.
Canon 400mm f/2.8L USM II. At 400mm, the image really doesn't have much background to render into bokeh. What little variance the background sky once had appears to have all melted away into a single mass of tonal variation.
Canon 400mm f/2.8L USM II with a x2 extender, set at 800mm. At 800mm, there's even less background to render into bokeh. As a result, the background has become one solid mass.
So final verdict? Does size matter? It sure does. What you can't see in this writeup is me lugging around a Canon 400mm f/2.8 USM II lens. It's humongous. And it's pretty impossible to shoot hand hold for an extended period of time. As for the 50mm f/0.95 Noctilux, for the bokeh to be optimized, you would need to be shooting between 3ft to 6ft away from the subject. Further than that, and the bokeh is no longer at its best.
The surprise of the group is the Nikon 135mm f/2.0D DC lens. Although I had no doubt that the bokeh would be pleasant at 135mm - with or without defocus control deployed - I didn't realize just how nice it was, when compared to the Leica 50mm f/0.95 Noctilux. Admittedly, I think if I shot the subject with the Noctilux closer to camera, the bokeh would have just melted away, and gave this Nikon lens more competition. But for the price, this Nikon 135mm f/2 DC lens is outstanding and is certainly no slouch compared to the Leica glass.
The black sheep of the group is the Leica 105-280mm f/4.2 R Vario. On first examination, the hazy low contrast rendering may put off some. But when you look closely at the image it renders, the bokeh is creamy, while the subject's lead eye is sharp. The issue with the contrast and saturation can be resolved on post in Lightroom - albeit at great effort.
Image from the Leica 105-280mm Vario edited in Lightroom. Although it is possible to edit in post, the color balance and saturation is so awkward, that I would recommend against using this lens. But if you don't mind the extra work, I do think that this lens does produce a unique rendering, with a very pleasant bokeh, combined with a good combination of sharpness and softness to produce complementary rendering of skin.
I think that the take away from this shootout is that the bokeh from the industry standard Canon 70-200mm f/2.8 USM II isn't significantly inferior to the bokeh of the other lenses. It does a good job in both isolating the subject, and rendering a pleasant background bokeh. And as big as the lens is, the lens is still manageable. And if you're a Sony A7 user, the autofocus works pretty well on both the Metabone and the Sigma adapter.
So does size matter, when it comes to bokeh? Honestly, it depends on just how much you want to isolate the subject. At 50mm, you still have the option to keep a good part of the background into the frame for the purpose of rendering bokeh. But at 400mm, since you're tightening the shot significantly, there's really not that much background to render bokeh. This makes the process of rendering a nice background bokeh to be difficult, though not necessarily impossible. In fact, it can look very amazing - just not in the image that I posted.
As you go up in focal length, the bokeh is likely going to get worse, only because you have so little background variation to work with in rendering something pleasant. And honestly, there's no need to use such an extreme focal length just for the sake of rendering bokeh. If I've proven anything with this test, 200mm is more than sufficient... and I suspect that 300mm, which I didn't test is the optimal focal length. Perhaps I'll include a 300mm lens, once I get my hands on one.
A side note on post processing of the images. I have tried to retain as much of the original raw images as possible, with the exception of tweaking the exposure. I did that so that the lighting of the images between the different version lenses would be as uniform as possible. The purpose of doing that is to make the comparison easier. Also, the white balance has been tweaked, because I wanted it to look the way I see the light in my office. Last, I did a slight rotational crop on some of the images, for the purpose of composition. Since I'm not testing corner sharpness, I don't think I'm taking away from the authenticity of the comparison images.
And a big special thanks to Oxana!
