“It is a truth universally acknowledged,” Jane Austen might have said, “that Thorsten Overgaard is utterly besotted with the Leica 50mm f0.95 Noctilux lens.”
Anyone who’s ever seen or read Thorsten’s pages will know it, and will know why: the 50mm Noctilux, with its very wide f/0.95 aperture, focuses sharply — if you’re sharp enough to focus it — at whatever distance you choose, but everything else is blurred, nicely separating what you’ve focused on from its dreamily blurry surroundings — although you may have to use a bit of extra sharpening afterwards on your computer to get absolutely bitingly sharp definition.
That concentrates your attention. When you look at the results — on just the one thing, or person, which is crisp and sharp and makes the rest of the picture look as dreamy as, well, as a dream of buying a £9,500-lens (or perhaps £7,000 on Amazon) when you’ve still got payments to make on everything else.
That £9,500 Noctilux 50mm f/0.95 — let’s just call it f/1 — isn’t quite so special as you might think, though.
Just double its focal length to 100mm, and knock off two stops from f/1 (f/1.4, f/2) for an aperture of f/2, and you’ll have the same amount of background blur, but you’re now talking about the Leica-screw-thread Canon 100mm f/2 or the easily adaptable (for ‘live view’ cameras) Olympus 100mm f/2.
Incidentally, the current newest Canon version, in EOS SLR fit, costs only £454 new on Amazon UK as I write this (and £225 on eBay). But I’m actually talking about the older Leica screw-fit Canon lens, made for Canon’s copycat versions of the original pre-M, Barnack-style Leicas.
Some pretty Canon LTM (Leica Thread Mount) 100mm f/2 lenses are on eBay — as I write this — at £410 including a leather case and strap, but the normal eBay price, with no box or case, is about £250.
£9,500 divided by £250 is 38 — that’s one thirty-eighth of the price of a 50mm Noctilux. And can you tell the difference in these photos below? Look at the backgrounds in particular. One of them was taken with the Nocti, one with a screw-fit Canon 100mm f/2, and one with a screw-fit Canon 50mm f/1.2. Click on each pic to enlarge:
The above example was taken at 50mm at f/0.95 on ISO 400 XP2 Super, 1/500th, rangefinder focusing on Leica M3: a slightly ‘soft’ face, because the focus is a bit ‘off’, even with the high-accuracy M3 viewfinder. But film is more forgiving than digital, because the emulsion has some depth to it, and so can compensate for a little misfocusing. As a result, this is less off-focus than it would be on digital M cameras. Very pleasant appearance — looking similar to results from the infra-red-sensitive digital Leica M8. XP2 Super film, unlike other film or digital cameras, performs better with over-exposure, because it gives a finer-grained, more detailed result, without ‘whiting out’. This picture would have had even finer grain by overexposing it at, say, 1/250s.
By comparison (above), the old Leica-screw-thread Canon 50mm f/1.2 is slightly contrastier, similar shallow depth-of-field and bokeh, 1/250s. It’s a slightly grainier result, because, although using slower shutter speed, less light seems to have reached the film (..pause here for a discussion about why it should be grainier even though the same amount of light was supposedly captured, but it’s probably due to ‘mist’ in the lens reducing light transmission, and being further from the incoming light of Aperture’s shop window). Incidentally, the minimally-coated Canon works nicely on film but may give flare and poor-contrast on a digital sensor, because the shiny glass-covered sensor can reflect incoming light back into a poorly-coated lens, where it bounces around, with contrast-reducing results. Film, however, has a light-absorbent ‘anti-halation’ rear-side layer, which pretty-much prevents this.
Another comparison (above): Old Canon screw-fit 100mm f/2, 1/125s. It produces a similar out-of-focus background blur, nicely sharp and contrasty — but grainier still (the argument continues about why that should be, but to some extent due to the extra extension of the lens during close focusing, and being farther still from the window).
With an auto-exposure camera, like the M7, or a digital camera, this could be automatically compensated for, and the camera may choose a suitably slower shutter speed. Ilford confirms that XP2 Super uses ‘grainy’ silver technology when shooting. But the silver’s all removed during development/bleach/fix, and the negative is converted into its final dye image). Note that each of these three lenses gives a similarly blurred background bokeh: the 50mm Noctilux at f/0.95, the Canon 50mm at f/1.2, and the Canon 100mm at f/2.
So where’s the myth come from, that you have to pay £9,500 to get beautiful background blur?
It’s come from people not thinking. It’s come from believing that “I have to buy a 50mm f0.95 because I’ve seen pictures which show such gorgeous blur” and not thinking that “I could get a lens twice as long — 100mm — but two stops slower at f/2 and get the same results”. And I would pay only one thirty-eighth the price. Oh, and step back two paces.
“Ah, but I want to use the lens on my Leica, not on a Canon!” Yes, but that old Canon 100mm f/2 does fit a Leica. It was designed and built to fit on a Leica. “B-but it’s an old lens”. Mmm, so are those classic Dual-Range Summicrons. And so’s the terrific 1968 Leica three-element 90mm f/4. So’s the incredibly sharp re-issued collapsible Leica 50mm f/2.8 Elmar. We talk about “classic” lenses, and the Canon 100mm f/2 certainly is a classic.
The above image with the 50mm f/0.95 manual-focus £9,500 Noctilux is shot into the light, and focused with the Leica M10-P’s rangefinder. Plenty of contrast-destroying flare (light bouncing around inside the lens) and inaccurate rangefinder focus (see the magnified close-up below).
In the image below we see the manually-focused 50mm f/0.95, £450 Zhongyi Speedmaster ‘Dark Knight’ shot into the light, on a Sony A7RII. Plenty of contrast-destroying flare (light bouncing around inside the lens) but accurate focus (see the magnified close-up, bottom). Pretty-much otherwise indistinguishable from the Noctilux.
Auto-focused (below) — at equivalent to 50mm f/3.5 — small-sensor £150 Sony super-tele HX300 bridge camera with Zeiss-branded built-in zoom. Plenty of contrast-destroying flare (light bouncing around inside the lens) around the sky, but far better contrast (due to the smaller aperture and Zeiss ‘secret sauce’ lens coating) away from the sky in the foreground. Sharp focus, but ‘grainier’, blobbier results in the shadows (see magnified close-up) due to much smaller, less capable sensor and electronics.
A bit of portraiture
Then again, there’s the marvellous ‘bokeh-meister’ modern Olympus 75mm f1.8 for micro-four-thirds cameras, and the Canon 85mm f/1.2 for EOS (also available for Canon’s newest R cameras), and Canon’s terrific old FD breech-lock 58mm f/1.2. Now there’s a lens! These all give wonderful background blur, just like — or better than — Leica’s Noctilux.
Looking at some of those in a bit more detail:
The Olympus 75mm focal length is halfway between the Noctilux’s 50mm and the screw-fit Canon 100mm. With an aperture only one stop slower than the Nocti — because it has a 50% greater focal length than the Leica lens — it’ would give similar out-of-focus background blur with an aperture of f/1.4. Let’s see which of these pics is from the 50mm Nocti at its widest aperture of f/0.95 (call it f1), and which is from the Olympus 75mm at its widest aperture of f1.8 … and which is from the Canon EOS 85mm at f/1.2, the wedding photographers’ ‘Jesus lens’. (Click on each of them for larger)
Above: The Olympus 75mm f1.8 on Olympus E-M1 camera ..a bit too bitingly sharp?
Now the Leica Noctilux (above), rangefinder-focused on M10-P . Inaccurate RF focusing; it really needs the Visoflex EVF or live view instead. (Although intended focus was on the nose and glasses, only the earrings, beyond, are reasonably sharp because of rangefinder inaccuracy.)
Above, the Canon EOS 85mm f/1.2 on Canon R. Similarly blurred background to the other two (but notice the really shallow depth-of-field on the farther shoulder). With a greater focal length than the Nocti, the 85’s slightly smaller aperture of f/1.2 nevertheless gives about the same fuzziness as the blur of f/1, or f/0.95, with a 50mm lens.
So, it isn’t just an f/0.95 lens which gives ‘creamy’ dissolving backgrounds. It’s a combination of focal length (the longer the better) and relatively wide aperture (the wider the better).
The Olympus 75mm f/1.8, incidentally, costs new, as I write this, $750 US or about £645 on Amazon UK.
£9,500 divided by 645 = about 14-and-a-bit, so the punchy, contrasty, extremely-fine (too-much detail, perhaps?) Olympus costs about one-fourteenth the price of the Noctilux.
I know it’s not all about price, and I know you’re not going to put an Olympus micro-four-thirds lens on a ‘full-frame’ Leica, but here’s what I’m pointing out. I’m trying to wean you off thinking that the Nocti is the only lens to give you lovely bokeh blur, and I’m trying to wean you off thinking that you have to pay £9,500 for that look, because you don’t.
The current new price, by the way, for the Canon wedding photographers’ 85mm f/1.2? Amazon says about £1,400 or less. The even newer ‘RF’ version for Canon’s latest, silent mirrorless R full-frame cameras is £2,266 on Amazon UK, but the older EOS version works perfectly well (see photo above) with Canon’s EOS-to-R adaptor.
In the above image, the Leica 50mm f/2 Apo-Summicron offers a smaller maximum aperture, so provides slightly greater depth-of-field than the f/0.95 Noctilux, thus the tree and cars look slightly sharper and less blurry than with the Noctilux photograph. There is less bokeh, but look at the sharpness of those glasses — see magnified version below — and the teeth.
Now for the Noctilux at f/0.95, best focus with the external off-the-sensor-sensing Visoflex electronic viewfinder on M10-P. Blurrier tree, but not quite such sharp and detailed glasses frames.
The Noctilux at f/0.95 again, but auto-focused with a Techart adaptor on a Sony A7RMkII. It’s no sharper (after all, same lens, same aperture) but much faster focusing.
Above, the Zhongyi Speedmaster 50mm f/0.95 on Sony A7RMkII: almost identical result to the Noctilux; maybe the glasses are a teeny bit sharper — see the magnified version.
Why get hung up on the Leica myth? Everyone nowadays makes aspherical lenses, so just because a Leica lens has the magical ‘ASPH.’ on the lens bezel, that doesn’t mean that other makers’ lenses aren’t aspherical (to be sharp right out to the frame edges). Kodak made aspherical lenses for their dinky little Disc cameras back in, er, 1980 — forty years ago.
Just because one breakfast cereal company prints “0% Mercury” on its packaging doesn’t mean that all other cornflakes do contain mercury. If one lens maker advertises that theirs are “100% Actual Glass”, does that mean that others’ lenses aren’t? And do you then start worrying that your own lenses are plastic? And why should that be a worry anyway?
Just because a lens has the magical ‘f/0.95’ on it doesn’t mean that it doesn’t have vignetting, or that it really provides twice the aperture of f/1.4, or that it’s free from colour aberrations, or that it actually can be focused using a rangefinder. From Leica’s own technical data sheet, the 50mm Nocti vignettes (reduces light at the corners of the frame) by about eighty-five per cent of the central brightness when it’s used at f/0.95 .
It is difficult to see it on this graph from Leica, because they’ve used such terribly faint dots to reveal — or hide? — that diagonal line running from 100% transmission at top-left down to about 15% at the lower right edge of the diagram (where it’s marked ‘0.95’). I’ve even enhanced the diagram slightly so that you can just about see it. So you get something like f/0.95 light transmission only at the very centre of your photo at f/0.95 and get only about one-sixth of that at the edges. Sorry. And you’ve paid £9,500 for that.
Above, Noctilux deer photo; look at the severe darkening at the lower corners (you can’t see the darkening at the upper corners, as the light’s so very bright there). Below is an enlargement of the central animal from photo above; not as sharp as one would hope from a £9,500 lens, even though focused through the bulky, bulbous, awkward-but-accurate, afterthought Visoflex add-on electronic finder.
Above, the same image, but with focus improved, using Topaz Labs’ ‘Sharpen AI’ program — not for any sharpening, but for its focus adjustment! Look at those horns now!
Above is a similar shot with Leica 50mm f/2 Apo-Summicron lens. There is no extra focusing done afterwards, straight out of the camera. (But that’s not surprising when you look at Leica’s MTF charts for the Noctilux and for the 50mm Apo, below.)
Above, the old Canon 58mm f/1.2. Now that’s what I call sharp and contrasty, and background-blurry!
Above: Similar, but unadjusted shot with Olympus OM 100mm f/2 (on a cheap n’ cheerful ‘K&F Concept’ OM-to-Leica-M-mount adaptor), focused through M10-P add-on ‘Visoflex’ electronic finder.
Above, the MTF charts at widest apertures for the 50mm f/0.95 Noctilux (left) and the 50mm f/2 Apo-Summicron (right). The nearer the line is to the top of the chart, the sharper and more contrasty the image will appear.
Of course, digital Leicas automatically adjust the edge brightness (somewhat) using in-camera software to brighten-up dark corners. That’s one of the reasons for those 6-bit codes on modern Leica lenses. But you don’t get that extra brightening when you use a Noctilux on a film camera, of course. It doesn’t have the electronics inside to correct peripheral light fall-off. Making the photo look bright right across the picture at f/0.95 is done with a digital camera’s electronics; not with the glass which you’ve just paid for.
And we haven’t even talked about the venerable, and distinguished, and inexpensive, beautiful background blur, Leica-screw-mount Nikkor 105mm f/2.5.
I borrowed a friend’s Noctilux for these photos, and used it on my M10-P and M9 and M7 and M3. And it didn’t focus accurately using the rangefinder on any of them. It worked best on the M3, because the old 1954 M3 has the highest-magnification viewfinder of all of them, which made it a bit easier to see whether it was — approximately — in focus, or not.
A few years ago I’d had to send back my then-new M9 to Leica at Solms twice to get it adjusted to match my lenses — because sometimes a rangefinder camera and rangefinder lenses just do not mechanically match up properly (the travel of the in-lens focusing cam and the travel of the separate focusing cam which is inside the camera may not be quite in sync; that’s been an interchangeable-lens rangefinder problem since the beginning) and the wider the lens aperture (and longer the focal length) the more accurate that focus-matching needs to be.
That’s why the Leica 135mm f/2.8 lens comes, or came, with its own magnification goggles, to help get really accurate focusing, and why the original (film) Leica CL wasn’t meant to be used with anything longer — or with a wider aperture — than its 90mm f/4 lens. Sharp rangefinder focus could not be guaranteed beyond those parameters.
Above: Wilting flower bouquet — I mean ‘bokeh’ — from the £9,500 Noctilux at f/0.95
Darn it; can’t focus it precisely with my (accurate) M10-P rangefinder.
Ditto, above, £9,500 but better focused with the £400 add-on Type 020 Live View Visoflex in (just about) focus. Total of lens plus EVF: £9,900 (not including the camera). Of course, the EVF might not be needed if the lens and camera were correctly matched together, at Wetzlar, for about £350.
Above: Same again, but with the £450 Chinese Zhongyi Speedmaster 50mm f/0.95. It’s slightly sharper than the Noctilux at this aperture and distance. Otherwise it’s indistinguishable. That’s £9,050 indistinguishable.
Above: Identical bokeh, but shot with the smaller, lighter £800-£900 Voigtländer 75mm f/1.5. I couldn’t move further back; I was stuck against the wall. Identical to the Noctilux and Speedmaster bokeh.
Above, encore, but with Canon’s excellent old £300 FD-mount 58mm f/1.2, plus a £130 Novoflex Canon-FD-to-Sony-FE adaptor on a Sony A7S. (My Canon-FD-to-Leica-M adaptor hadn’t yet arrived): more contrasty, and a slightly cheaper and much lighter option than the 50mm f/0.95 Speedmaster. Slightly more pronounced ‘rings’ around distant out-of-focus highlights, though.
Same again, but with a current Canon EOS 50mm f/1.2. It’s the sharpest overall focus of this bunch of lenses but less with yellow saturation than the others, I reckon, and so the flowers don’t stand out so much from their background, even though saturation was turned up two notches, as on the other cameras, on this Canon R.
I have to report that this gorgeous 50mm f/0.95 Noctilux does not focus accurately using the rangefinder of my factory-adjusted digital M cameras; and not even with the 85%-lifesize finder of my (film) M7, nor with the 90%-lifesize (film) M3 finder. Of course, it may focus perfectly on my friend’s own M camera(s), but I didn’t try it on his.
To try to get perfect focus I used the Nocti with Leica’s straight-off-the-sensor live view on the rear of the digital M10-P, and also with the add-on eye-level electronic Visoflex finder. With those, you get focus peaking — sharp coloured outlines on in-focus edges — and you can also get a really magnified close-up view, too, straight off the sensor, for — supposedly — even better accuracy. Except that with no built-in stabilisation, the magnified view bounces around in the electronic finder (and on the live-view rear screen) making it almost impossible to focus accurately at all, because camera-shake is magnified as well as the view which the lens sees!
So this is not really a fair revue of the Noctilux — without sending it back to Wetzlar with the camera(s) which it’s going to be used on to have the lens and camera(s) properly matched to each other. My cameras have already been adjusted, so I don’t want them fiddled with yet again, as they’re now set up to match my own various lenses (50mm, 75mm, 90mm). That’s the problem with traditional mechanically-geared rangefinders. They can get nudged out of sync or may not even focus perfectly when new straight out of the box.
I do like to be beside the seaside 🎵
Above: Noctilux at the seaside, f/0.95, focused with the M10-P rangefinder: a blurry mess. Notice the vignetting?
Above: Noctilux at f/0.95, but focused through the Visoflex off-the-sensor EVF: the best the lens can do wide open. Central people magnified.
Above, Noctilux at f/2 with the Visoflex — now that’s looking better. Crispness and contrast at last. Almost-central people, magnified.
The much smaller Leica 50mm f/2 Apochromatic: crisp, sharp and very contrasty at its widest aperture, but some vignetting, too, at its maximum aperture; rangefinder focus (there’s nothing the matter with my rangefinder itself), central people and dog.
Above: Voigtländer 75mm f/1.5 wide open, rangefinder focusing: crisp and sharp; not quite as much contrast as the other lenses deliver at f/2, but pretty close.
Just for comparison (above), a modern Canon 100mm f/2 at f/2 on Canon R. Now that’s what I call crisp and sharp, and note that the blurred owl sculpture at the centre has the same bokeh blur as the 75mm gives at f/1.5 and the 50mm at f/0.95. All these were shot at ISO 200 and 1/4000s, except this photo which was shot at ISO 100 and 1/2000s.
Best of the bunch
Years ago, when I first compared a 50mm f/0.95 Noctilux to a 50mm f/1.4 Summilux, I preferred the f/1.4. At average portrait distances, the smaller, lighter 1.4 kept nose, eyes and back of the head all in focus, whereas the wider 0.95 aperture meant that you could have nose and/or eyes and/or back of the head in sharp focus, but not all three. So the back of the head tended to be blurry, and blend in with the background instead of separating from it.
In Thorsten’s very sharp Noctilux photos, though, he tends to shoot sharply outlined people at greater than head-and-shoulders distances, and he tends to shoot three photos as a burst, and then choose the sharpest of the three and adjust that with software (and I have no idea what he does about cropping his photos). So we see sharp results. But I think, like most people, he tends to show only the sharp results . Who knew that Cartier-Bresson regularly shot several almost identical pictures and then chose, of those, just the one he preferred, until the Magnum book of his contact sheets was published? (So much for his — so-called, in translation — ‘decisive moment’.)
What I’m saying is that the ‘hit rate’ with a Noctilux is pretty low. Yet we generally see, online and elsewhere, only successful, accurately-focused results, because almost no-one publishes their missed-focus, out-of-focus, blurry failures. People display their successes, not their failures. So although all the Noctilux photos which we see in books or online do look nice and sharp and crisp and well separated from their backgrounds, there is no easy way of knowing how many unseen and ‘failed’ Noctilux photos there are which are blurry, unsharp, full of flare, and generally unsatisfactory. .Going by published results doesn’t mean a thing: what about all the unpublished results?
Above: Leica M9, unknown lens — probably an old screw-fit Canon.
The super-wide aperture of the 50mm f/0.95 Noctilux would have been relevant in film days — when the widest possible apertures let you shoot in extremely dim light (see the first pics at the top of this article). But with the very high ISO capabilities of current digital cameras it is really rather pointless, except for bragging rights, or as a possible investment, to sell at a higher price in ten or twenty years’ time.
It’s as if Dunlop (are they still around?) now make the world’s most expensive, and pretty good, car tyre (tire) inner tubes. But they’re an irrelevance, because nowadays we all drive tubeless.
A properly-adjusted Noctilux may be a good buy (but a Speedmaster is better value) if you shoot film in low light. With modern digital cameras, though, the camera will handle the low light, so paying extra for an f/0.95 aperture doesn’t make much sense. Use a smaller, cheaper, lighter 75mm f/1.5 or a 100mm f/2 instead.
Notice that with those seaside pictures, above, even though the Noctilux is correctly focused (using the add-on electronic live viewfinder on the M10) it still gives some blurry blue edges in the widest-aperture shots. It’s an ‘aspherical’ lens (for better sharpness right to the edges of the frame) but it’s not an ‘apochromatic’ lens, for focusing all colours at the same plane. Its designer, Peter Karbe, says in one of Thorsten’s interviews, that the lens’s blue or purple fringing hasn’t been eliminated at its widest apertures because that would have cost much much more. But the whole point of buying it is to use it at its widest apertures.
So the more modest-aperture f/1.5 and f/2 lenses here give far better results, especially when pointing towards the light, than the Nocti does. Sorry, but that’s the way it is. (Blue edges can easily be removed, of course, with most photo-editing programs, but they’re not automatically removed within Leica cameras, as, say, Nikon does it.)
My advice when buying a lens, and to get the ‘creamiest’ background – or foreground – with a 50mm, is to buy a cheap Canon EOS camera, buy a Canon 50mm f/1.2, and then carefully tear up and throw away seven and a half thousand pounds.
If you really need a 50mm with delicious blur for a full-frame modern Leica, then buy an old, clean, Leica-fit 100mm f/2 Canon lens, or (only if your Leica has live view, as there’s no rangefinder cam in the old Olympus 100mm OM lens), the brilliant 100mm f/2 Olympus OM lens and a £30 Leica-screw-to-M-bayonet or OM-to-M adaptor ring. And then – very, very carefully – tear up into very, very small pieces, and then put in the dustbin, eight and a half thousand pounds.
The smaller and neater the pieces which you tear the money into, the better will be the beautiful bokeh blur. I absolutely promise. Some people just don’t understand that it’s all in the tearing. The precision and dedication with which you tear up the money will make your background blur all the more spectacular!
That’s the essence of photography; take your time, think very carefully about what you’re doing, and then tear up the money very, very, slowly.
A little bit at the end…
Yes, I do realise that a lot of work, money and research went into designing — and manufacturing — the 50mm Noctilux f/0.95. Yes, I do realise that it has exotic glass inside and that producing aspherical elements isn’t as run-of-the-mill as just polishing ‘spherical’ glass blanks. I do know that the Nocti is expensive to make. I do realise that they’re hand-assembled, and it’s the centring of each element that’s so important, and that each one is tested before boxing and shipping.
I do know that their anti-reflection lens coating is far more capable than that of the old Canon screw-fit lenses. But what I’m saying is don’t get hung up on the mythical properties and supposed magic of the Noctilux. Advertising and PR is all about convincing people of the desirability of some product or other, but never mentioning its shortcomings.
I’m saying, “look around, see what else will give you pretty much similar results”. Don’t be blinded by the PR effulgences because there is more than one way to skin a cat. Or, for that matter, to take blurry-background pictures.
This article is very much an opinion piece and, of course, you may have a completely different view on the rôle of expensive lenses such as the Noctilux. I’d like to hear them and will respond robustly, as usual. I accept that the Noctilux is more akin to a high-end Swiss Watch in its appeal. It’s expensive and desirable as a luxury item. While I can argue that a humble Casio will do the same job not everyone will be convinced. So let’s hear your views.