Note: due to a lot of stuff going on these days I haven’t been able to do much for the Blog. I decided to publish this article unfinished. When you read it, keep in mind that most paragraphs are subject to change and further tests and results will be added over time.
Dec 12.2015 – Update: Vignetting, Bokeh Balls, Transmission and flaring tests have been added.
A great many people dream of super fast 50mm lenses but will never be able to afford one, or they simply can’t justify the cost and for a good reason.
Even on the market for vintage lenses you often pay 300$ upwards for a lens faster than f/1.4. If you want an f/1 (or f/0.95) lens, you will most likely have to spend somewhere in the 4 digit area.
On the other hand, if you’re happy with a 50/1.4, you can get some of these for as little as 50 bucks or less, depending on your luck and the place where you go for shopping.
So, where does the fascination for these super fast lenses come from? After all, f/1.2 is only half a stop faster than f/1.4.
Note: I was previously under the misconception that f/1.2 was only 1/3 of a stop faster than f/1.4, because my camera told me so. (the camera is set to increase aperture in third stops and does so by going f/1.4 – f/1.2 – f/1.1 – f/1.) In fact, mathematically speaking it’s actually closer to a half stop difference at 0,44 something.
However, as it turns out, the lens really is only 1/3 of a stop faster, therefore I decided to leave my title be (scroll down to find out more about that).
I believe that there are several explanations for this phenomenon:
Super fast primes are usually rare and very expensive and thus add to the “status” just like an expensive car, watch, handbag, etc. would.
Some super fast primes are based on very exotic, experimental lens designs, where the manufacturers left the standard glass formulas (7 elements in 6 groups for example) behind to venture into more complex territory. The rare Canon EF 50mm f/1L USM – yes, there was an f/1 lens – has 11 lenses in 9 groups, and the Noctilux 50/0.95 has 11 lenses in 8 groups just to name two. Some people might just consider these lenses a collectible or want them for their special properties.
Regarding the above mentioned special glass construction, it is likely that the super fast lens will render an image differently and might actually do a better job. As an owner of the Canon 85/1.2L II and someone lusting for the 50/1.2L I can assure you that there definitely is something to the “special rendering” argument.
Not everything needs a reason. If you’ve got enough money, why buy the cheap lenses? Some people need 20 lenses to be happy, or they believe they’d be happy then, and others might prefer to invest more money in a single lens.
What am I going to compare here?
With my Nikon F2, I directly went for the Nikkor 50mm f/1.4 Ai because I wanted something good and affordable. The Ai-S version has no benefits on an old film camera and is optically identical anyways.
The 50mm f/1.2 Ai-S, which is still being produced and sold by Nikon, was always on my wishlist, in case I would fall in love with film photography and/or suffered a major GAS attack. This review over at photographylife was another argument for me to get the 50/1.2 Ai-S.
Well, here I am and here are both lenses next to each other, ready for a merciless battle. In the end I will definitely sell the loser.
Nikon Nikkor 50mm f/1.4 Ai vs. Nikon Nikkor 50mm f/1.2 Ai-S
My 50/1.4 is an Ai version, the 50/1.2 is an Ai-S (the Ai-S works better on modern (D)SLR’s for aperture control but nothing else was changed.) which you can see on the colours of the aperture scale. If the smallest aperture (f/16 in both cases) is coloured in blue, it’s an Ai lens, if it’s coloured in orange, it’s Ai-S
The Nikkor 50mm f/1.2 is very light for such a fast aperture lens (430g with adapter and caps vs. 310g for the f/1.4), it’s also not much larger than the 50mm f/1.4.
The 50/1.2 has a large rear element that you should be very careful with when ataching and detaching the lens from the camera. The front element is also larger, of course, as is the focusing ring.
The 1.4 is in mint condition and the 1.2 is in a great condition too, but it has some scratches on the paint. Both have perfect glass and clean aperture blades. They share the same 52mm filter thread and both lenses are multi-coated but it seems as if Nikon changed the formula slightly:
When you go from infinity to minimum focus distance, both lenses will extend a little, however the front element does not rotate.
Focusing and setting the aperture works smoothly. While the focus ring on the f/1.4 could be called “perfect”, the one on the f/1.2 turns a little too easily for my taste. There’s another issue that adds to that, which is also the only real drawback I could find regarding the 50/1.2 so far:
The incredibly short focus throw. It measures only about 90° which is a shame really, especially considering the 50/1.4 with its focus throw of >180°. One would think that a manual focus lens at such a high price point would come with an improved focusing mechanism.
Getting the focus spot-on is therefore much more difficult than with the f/1.4 and the wider focus ring cannot compensate for that.
Setting the aperture is basically the same for both lenses. You’re stuck with full stops – yes, full stops only! – but the 50/1.2 has an unmarked stop between f/1.2 and f/2. I assume that this is f/1.4.
I have an adapter for both lenses and my adapters are chipped, so that I can set focal length and aperture (in A-mode and M-mode) and most importantly that I get a focus confirm sound and visual confirmation in the viewfinder. This helps a lot when you and your subject aren’t stationary and it’s very reliable too.
Now, let’s get to the tests, shall we?
Light transmission / colour cast
First off, I couldn’t find any noticeable colour cast in both lenses with the 50/1.2 being a little closer to the Color-Ultron.
I wanted to investigate the difference between f/1.2 and f/1.4 and so I set up a little something to do just that.
Instead of using my fluorescent lamp I used one of my flashes (Yongnuo YN560-IV) because it was standing there anyway.
The camera was set up 1m away from the target and I only used a tiny crop from the card (in the centre of the frame) to compare the Histograms in Lightroom without any measurable vignetting affecting my test.
I set the Nikkor 50/1.4 at f/2 as reference (0 EV) and compared the other images to that. The Nikkor 50/1.2 at f/2 is 0.1 EV brighter. What I found really pleasing is the fact that both lenses upon opening the aperture to f/1.4 actually produce an image that is exactly 1 stop brighter (I lack the precision in my Histogram to make super-precise measurements).
However, when I open up the 50/1.2 further to f/1.2 there is the surprise: the image only brightens by approximately 0,3 to 0,35 EV.
Contrary to the mathematical solution that would suggest that f/1.2 is half a stop faster than f/1.4, in this case we obviously only get 1/3 of a stop more light through the lens.
I will repeat this short test with fluorescent light to check my results of course.
A very common feature with photographic lenses is the so-called vignetting. It’s usually more pronounced the smaller the lens is and the larger the aperture.
The explanation is not so simple (see this Wiki article for more information):
The glass has to spread the light gathered evenly on the sensor area. However, with large apertures and small lens bodies (to save size and weight) there will always be some optical vignetting added to the pixel vignetting only occurring in digital cameras. The amount of natural vignetting is more common in wide-angle lenses, though.
One can easily remove vignetting in most cases, because it’s even around the frame and most modern lenses have a camera profile in Lightroom that will correct for distortion and vignetting. (Some manufacturers won’t even let you chose whether to leave vignetting in the image or not. Yes, Fuji, I’m talking to you!)
Vignetting with portrait photography can be a desired effect, because it increases the focus on the subject.
However, with landscapes, architecture and especially astrophotography vignetting is very nasty because it results in an unevenly lit image. (In the case of astrophotography it will also require you to brighten the image and thus induce more noise.)
I shot this comparison about 1m away from a white screen (home theatre) with a bare YN560 IV flash on either side of the camera. Because of the high gain of my screen, I had to place the flashes several metres away at minimum power and set the camera to iso 50 to prevent a washing out at f/1.4.
Here’s the comparison:
First of all, at f/2.8 vignetting can’t be considered an issue anymore. Even at f/2 the vignetting is quite neglectible unless you have a perfectly even subject.
Looking at f/1.2, the Nikkor 50/1.2 Ai-S performs quite well in my opinion. The vignetting only amounts to 2.2 EV in the far corner.
Vignetting is reduced to ~1.7 EV upon stopping down to f/1.4. That’s only barely worse than the huge Sigma 50mm f/1.4 Art lens (1,66 EV). And the Zeiss MILVUS 50mm f/1.4 sports a whopping 3,1 EV vignetting wide open.
The Nikkor 50mm f/1.4 Ai has much stronger vignetting, with a falloff of 2,3 EV at f/1.4.
If you wonder why the 50/1.2 looks brighter at f/1.2 than at f/1.4, the reason is pretty simple:
I couldn’t reduce the flash power further and was already at iso 50, so the image is about 0,35 EV brighter.
For a more thorough look at the vignetting, check out my comparison article here. This article covers vignetting for all my vintage lenses and also provides a chart with measured numbers.
Flaring and Ghosting
When shooting portraits outdoors in bright daylight, most of the time you either look for open shade or shoot against the sun so that the subject’s face is evenly lit.
I used my Jansjö lamp as a stand-in for the sun because the latter was unavailable this past few days.
A black background was the only other thing inside the frame so you can clearly see the flare patterns.
This flaring test is intended to show how the lenses compare with the sun inside the frame.
This image shows my bare background exposed the same way as all the others with added “sun” were. (you can see the reflection of the LED lamp in the upper right corner)
Now to the comparison:
I shot the scene at various apertures from wide open down to f/4.
White balance was not adjusted for the LED source but all the images have the same white balance applied.
Both lenses exhibit strong ghosting even when stopped down to f/4.
Flaring for the 50/1.4 does not really improve upon stopping down, but the sunstar becomes more defined. The 50/1.2 clearly profits from stopping down regarding the flaring but it suffers more loss of contrast throughout compared to the 50/1.4.
It’s clear that both lenses should not be used with the sun inside the frame, unless you aim for a special effect. In that case, I would prefer the 50/1.2 because the haze is more homogenously spread across the frame.
In a second round I wanted to see how the lenses fare with the “sun” slightly outside the image frame (namely in the corner of a 35mm fov corner).
The effect of the sun in this spot can not be diminished by the use of a normal lens hood, it’s still too close to the image frame. If the sun was further away from the frame corner, the use of a lens hood would protect the lens from flares and comes highly recommended.
Both lenses are not safe from flaring with the sun closely outside the image frame. What’s striking is the fact that the flaring grows stronger when the lens is stopped down (the opposite of what happened with the sun inside the image frame).
The faster lens loses this battle because it still causes some ghosting up to f/2. This – at least I believe so – is a result of the larger front element catching more light from the sides.
Special test: Starbursts
When you stop down the aperture a lot, diffraction at the aperture diaphragm will cause spots of light to look like stars (See this link for more info). I like to do this for nightscapes with city lights, or when I want to have the sun in the frame.
The starburst effect is greatly influenced by the number (and shape) of aperture blades and an even number in this case is not favourable unless you prefer your stars to be “simple” (look forward to the upcoming 50mm comparison where I did this for all my 13 lenses).
The Nikkor 50mm f/1.4 has 7 aperture blades which causes a 14-pointed star, whereas the 50mm f/1.2 comes with 9 blades to form an impressive 18-pointed starburst. And here’s what this looks like at f/16:
The Nikkor 50mm f/1.2 Ai-S even at f/16 suffers from ghosting and some flaring, but the star looks beautiful.
The f/1.4 Ai on the other hand has a better defined starburst and more contrast overall (almost no flaring) with only a little ghosting visible.
Keep in mind, however, that this effect is commonly used with weak light sources far away that will cause less/no ghosting and no flaring at all.
If you want to use it with the sun, it helps to partially cover it behind leaves or a building.
Here’s an example for that (although not free of ghosting):
Even though it’s getting old to show you the bright dots of light as a “measure to judge Bokeh”, they are still the easiest way to compare the shape of oof (out-of-focus) highlights.
What you can see is that, without rounded aperture blades, no matter the number of blades a lens has, they become visible upon stopping down. This often results in a “busy” background. Other factos that add to this impression are the cat-eye-effect, where oof highlights lose their round shape the closer you get to the image corners. This is a result of the large aperture opening and internal vignetting. Optical correction of lenses (aspherical or apochromatic elements, etc.) can also lead to a busy background, because they often create sharp edges or so-called onion rings inside the oof highlights.
Both tested lenses here have unrounded aperture blades and suffer from a pronounced cat-eye-effect, but other than that the Bokeh balls are evenly lit.
For a lack of space I shot this comparison at only about 50cm distance to my subject (the Zeiss Planar T*) with approx. 2m to the backgroung. A long tube with lights was used to create the Bokeh balls in front of my black backdrop this time.
Two things you’ll have noticed right away:
Already at f/1.4 the shape of the aperture of the Nikkor 50/1.2 becomes visible. The Nikkor 50/1.4 looks better there, because it has perfectly round highlights wide open.
The Nikkor 50/1.2 has a less pronounced cat-eye effect and looks better wide open (f/1.2 vs f/1.4) and also from f/2 on up with the advantage of having more aperture blades.
Here’s a comparison for the OOF-rendering capabilities of these two lenses:
The first set was shot at a distance of only 0,8 metres, which leads to rather pleasing, creamy bokeh.
The second set was shot at a distance of 1,2 metres, leading to many nervously shaped bokeh balls in the background blur, especially in the left side of the image:
Both lenses are capable of creamy bokeh when shot up close to maximize the background blur, but they both tend to nervous bokeh balls once the distance is increased. To me, the 50mm f/1.4 looks more pleasing despite the smaller maximum aperture. In addition to a less nervous bokeh, the 50/1.4 also delivers more contrast and sharpness.
The comparison was performed at a distance of approximately 1,3 metres. I decided not to test at the minimum focus distance anymore, because I rarely use my lenses up close.
At 1,3 metres the subject height filling the frame would be around 60cm vs a width of 90cm, which is still pretty close for portraits where you want more than just headshots though.
I took pictures with my subject text in the centre of the frame and at the outer border (see full images below). However, I did not do any tests for extreme corners, because those are irrelevant for my purposes.
When I take portraits, I focus on the eye closest to the camera using one of the corner AF points – I never focus/recompose with the centre AF point – to follow the rule of thirds as best as possible.
I used one YN 560-IV shooting through a 32″ umbrella and my white reflector on the opposite side of the text to spread the brightness nicely. (I do agree that this is not perfect and will certainly work something out for the huge lens comparison coming 2016)
All images were processed identically: (standard sharpening was applied, camera mode set to standard but no other changes to default settings in Lightroom)
Here’s the centre area (100% crops) from wide open to f/2:
The 50mm f/1.2 shows slightly more colour fringing (the glow in the oof area is coloured green/purple for the f/1.2), but it has more contrast and appears to be noticeably sharper at f/1.4 overall (even at f/1.2 it’s visibly sharper than the 50/1.4 wide open).
By f/2 both lenses are already very sharp and I really struggle to see any differences at all.
Regarding the results of the Bokeh balls comparison above, I’d suggest to shoot the 50/1.2 either wide open or stopped down to f/2, not at f/1.4.
Here’s the centre sharpness from f/2.8 to f/5.6
(50/1.2 Ai-S on the left, 50/1.4 Ai on the right)
I really can’t see any difference here..
Update – Dec. 14.2015
I just shot my sharpness comparison for the big 50mm comparison with all my lenses (here’s that article) and did so at only 80cm distance, because in my opinion the differences are easier to judge.
Here are the two Nikkor lenses pictured against each other:
I really failed the 50/1.2 wide open, but my camera screen doesn’t offer such a high resolution and I didn’t want to spend hours just to take the pictures. (all the rest of my images were better focused)
As you can see, the depth of field (DOF) at f/1.2 at only 80cm distance is really, really shallow (the text is printed at size 11) and I missed my focus by several millimetres. The focal plane is close to the word “was” but not spot on, so everything is either coloured green or magenta because of chromatic aberrations – which are really strong.
At f/1.4 I got the focus nicely on the word “power” which allows for a more precise comparison between the two lenses. The 50/1.2 clearly is sharper with more contrast, but this also leads to a more pronounced glowing of the colour fringing.
Stopping down to f/2 the same result as before can be seen: both lenses are incredibly sharp and contrasty and there’s hardly any difference to be spotted.
The same goes for f/2.8 and so on. That’s why f/4 and f/5.6 were left out of this comparison.
I rearranged my previous (first) setup to compare the border sharpness:
the reflector is not inside the frame anymore, but it’s very close to the text.
Wide open comparison:
Yes, the Nikkor 50mm f/1.4 Ai is a little sharper in the border areas, both lenses are quite disappointing though. Placing the models eyes in this area of the image would require a lot of processing to get a usable portrait.
The situation gets better upon further stopping down:
Incredibly, the 50/1.4 remains sharper down to f/5.6.
And, once again, simply to show what you can get out of an image with adequate RAW processing, here’s a 100% view from an image taken at f/1.2 with the Nikkor 50/1.2 Ai-S:
Obviously, I treated the chromatic aberrations, applied some sharpening, pushed clarity to +15, reduced the Highlights (and pushed the whites instead) and reduced the shadows and blacks slightly to gain some contrast. The contrast slider in this case was working against me and I had to leave it at zero.
I did not overdo any of the sliders, so this scenario can easily be applied to any real-world pictures too.