Impossibly fast 50mm?

_T_ · Sep 21, 2020

I don't understand why we can produce 35 pound 600mm lenses with front elements so large we can't use filters at the front of them, but we can't have a 50mm lens with an aperture of f/0.5 and an entrance pupil of 10cm.

Why is that? Will people not pay for that? Is it difficult to engineer?

It would have shallow depth of field wide open but would have over 1m of depth of field at a 10m focusing distance. It wouldn't be unusable. And we shoot with shallower depth of field in macro.

What am I missing here?

jjphoto · Sep 21, 2020

_T_ said:
I don't understand why we can produce 35 pound 600mm lenses with front elements so large we can't use filters at the front of them, but we can't have a 50mm lens with an aperture of f/0.5 and an entrance pupil of 10cm.

Why is that? Will people not pay for that? Is it difficult to engineer?

It would have shallow depth of field wide open but would have over 1m of depth of field at a 10m focusing distance. It wouldn't be unusable. And we shoot with shallower depth of field in macro.

What am I missing here?

If Stanley Kubrick was still around and there was the opportunity to have one made I'm pretty shure he'd be up for it (famously having adapted the 50mm F0.7 NASA Zeiss lenses for Barry Lyndon (crap film (coming from a big Kubrick fan))).

BradS · Sep 21, 2020

Hmmm... I'm thinking that the cost and physical size and weight might make such a beast prohibitive / not feasible.

Of course, if you really wanted one, and had a shit ton of money to devote to the project, you could contract the design and manufacture of such a thing...and then sell 'em at a huge profit!
A market that is not being served, is a business opportunity.

reddesert · Sep 21, 2020

- Yes, it's difficult to engineer.
- In fact, I think f/0.5 is a theoretical limit; it would be literally impossible to make a faster lens that imaged sharply off-axis. (Don't ask me to prove this right now, but see the section on "Numerical aperture versus f-number" at https://en.wikipedia.org/wiki/Numerical_aperture to dig deeper.)
- Its depth of focus (like depth of field, but at the film plane, not the image plane) would be absurdly thin.
- That would place strong constraints on how flat the camera held the film, accuracy of focusing system, etc. You'd learn all the ugly inaccuracies that are within tolerances at f/1.4.

To expand on point 3, the typical 35mm circle-of-confusion for acceptable sharpness is 0.03 mm = 30 microns. At f/2, that means you have a depth of focus of +/- 60 microns. But at f/0.5, you would have a depth of focus of +/- 15 microns just for acceptable sharpness (not even critical sharpness). The thickness of most roll films is about 100-120 microns, and the emulsion layer is maybe 20 microns thick? So any bend in the film would cause problems, and at an f-ratio as fast as 0.5, you would actually have to start worrying about whether you're focused on the front of the emulsion layer or somewhere in the middle. IOW, be careful what you wish for.

NB23 · Sep 22, 2020

I’m not sure what you are trying to say besides the theoretics, and you obviously overlook the fact that there is an image, a photograph to look at, just for what it is and not for what it is not.

And something would indeed be in focus, even if thin. And the ouf of focus characteristics will have its own story.

And the whole image would always be interesting to look at. Interesting because it would always exude the singular DNA of the lens, no matter how thin the focus, badly corrected aberrations or distorted. It is not a quest for perfection.

reddesert said:
- Yes, it's difficult to engineer.
- In fact, I think f/0.5 is a theoretical limit; it would be literally impossible to make a faster lens that imaged sharply off-axis. (Don't ask me to prove this right now, but see the section on "Numerical aperture versus f-number" at https://en.wikipedia.org/wiki/Numerical_aperture to dig deeper.)
- Its depth of focus (like depth of field, but at the film plane, not the image plane) would be absurdly thin.
- That would place strong constraints on how flat the camera held the film, accuracy of focusing system, etc. You'd learn all the ugly inaccuracies that are within tolerances at f/1.4.

To expand on point 3, the typical 35mm circle-of-confusion for acceptable sharpness is 0.03 mm = 30 microns. At f/2, that means you have a depth of focus of +/- 60 microns. But at f/0.5, you would have a depth of focus of +/- 15 microns just for acceptable sharpness (not even critical sharpness). The thickness of most roll films is about 100-120 microns, and the emulsion layer is maybe 20 microns thick? So any bend in the film would cause problems, and at an f-ratio as fast as 0.5, you would actually have to start worrying about whether you're focused on the front of the emulsion layer or somewhere in the middle. IOW, be careful what you wish for.

BradS · Sep 22, 2020

NB23 said:
.....snip.... It is not a quest for perfection.

darkosaric · Sep 22, 2020

I think it is simply matter of economics - few would buy this expensive lens. But this could be an interesting project for home made lens, one can buy some old telephoto lens and try to rebuild some 50ish f0.5/0.7 from the parts?

RLangham · Sep 22, 2020

NB23 said:
I’m not sure what you are trying to say besides the theoretics, and you obviously overlook the fact that there is an image, a photograph to look at, just for what it is and not for what it is not.

And something would indeed be in focus, even if thin. And the ouf of focus characteristics will have its own story.

And the whole image would always be interesting to look at. Interesting because it would always exude the singular DNA of the lens, no matter how thin the focus, badly corrected aberrations or distorted. It is not a quest for perfection.

What he's saying is there will hardly be an image under most adverse circumstances.

Moreover think about every SLR you've ever used (because this would have to be focused with TTL viewing, hence SLR or mirrorless): how many of them could mount such a lens (in its most compact possible housing) without it blocking some controls or being interfered with by projecting parts of the camera body?

At the f/stop we're describing it really would be quest for perfection because only perfection would be adequate.

reddesert · Sep 22, 2020

What I'm saying is that a lens approaching the theoretical limit of f/0.5 - or even f/1 - is difficult to make, places strong requirements on the equipment it is used with (eg tolerances of the body, focus system, film flatness), and requires a lot of care from the user. And thus it could easily wind up with disgruntled users who complained that it made blurred images and wasn't sharp, even if that wasn't the fault of the lens. Any manufacturer would have to consider this. Most of the super-speed lenses near f/1 have been (IMO) expensive curiosities used for cine/TV and/or bragging rights.

BTW, the Zeiss 50mm f/0.7 made for NASA and Kubrick was designed to cover the 18x24mm frame of 35mm cine, not 35mm still, and it had a very small back focal clearance of 5mm so couldn't have been adapted to an SLR. (I'm only looking at optical diagrams, of course I've never seen one of these in real life.) See bottom of this page: http://www.klassik-cameras.de/Biotar_en.html (partially in German).

RLangham · Sep 22, 2020

reddesert said:
What I'm saying is that a lens approaching the theoretical limit of f/0.5 - or even f/1 - is difficult to make, places strong requirements on the equipment it is used with (eg tolerances of the body, focus system, film flatness), and requires a lot of care from the user. And thus it could easily wind up with disgruntled users who complained that it made blurred images and wasn't sharp, even if that wasn't the fault of the lens. Any manufacturer would have to consider this. Most of the super-speed lenses near f/1 have been (IMO) expensive curiosities used for cine/TV and/or bragging rights.

BTW, the Zeiss 50mm f/0.7 made for NASA and Kubrick was designed to cover the 18x24mm frame of 35mm cine, not 35mm still, and it had a very small back focal clearance of 5mm so couldn't have been adapted to an SLR. (I'm only looking at optical diagrams, of course I've never seen one of these in real life.) See bottom of this page: http://www.klassik- kameras.de/Biotar_en.html (partially in German).

It would follow then that too many variables would have to be perfect for the shot to turn out--critical focusing, the tolerances of the camera, optical formula... I think it very likely that without a huge company trying very hard to perfect such a lens the effect would be neither interesting nor artistic.

blockend · Sep 22, 2020

Lenses wider than f1.0 are more popular in smaller formats, perhaps because they are much less expensive to produce. For example, a 58mm 0.95 Noct Nikkor in Z mount costs £8,299, whereas m43 (and presumably half frame or 110 if they were still around) lenses of the same aperture in various focal lengths can be had for £700 - £1000. That said, a TTArtisans 50mm 0.95 in Leica M-mount costs £600.

Conversely, the larger the format the easier it is to get extreme subject separation, especially with camera movements.

AgX · Sep 22, 2020

reddesert said:
Most of the super-speed lenses near f/1 have been (IMO) expensive curiosities used for cine/TV and/or bragging rights.

There were such lenses made for technical reason as screen photography.

Ko.Fe. · Sep 22, 2020

Lens like this is going to be useless. At f0.5 here is no DOF for practical photography.
This is why Canon now makes tele lenses with F8 maximum aperture.

Ko.Fe. · Sep 22, 2020

blockend said:
Lenses wider than f1.0 are more popular in smaller formats, perhaps because they are much less expensive to produce. For example, a 58mm 0.95 Noct Nikkor in Z mount costs £8,299, whereas m43 (and presumably half frame or 110 if they were still around) lenses of the same aperture in various focal lengths can be had for £700 - £1000. That said, a TTArtisans 50mm 0.95 in Leica M-mount costs £600.

Conversely, the larger the format the easier it is to get extreme subject separation, especially with camera movements.

Here is 50 0.95 lens for Nikon Z and Canon RF as 600 USD lens. Probably same giant seagul factory made.

removed account4 · Sep 22, 2020

too rich for my blood, and seems not many people would buy it.
I mean a .95 noctilux costs 12grand+ ( new ) and is more of a cult accessory for some people.
not saying there aren't people who aren't cultists who use it but just sayin'

https://en.wikipedia.org/wiki/Noctilux

what were you hoping to shoot with it, portraits? still lives? landscapes ? there are ways of achieving selective focus and bokeh&c without a super fast lens like that, that don't cost 12grand+...
have fun !
John

film_man · Sep 22, 2020

As said the DOF would be so thin that to get a photo of what you intend to do you'd be pretty much limited to tripod shots of static and very flat objects. Pretty bland and useless really if you ask me. It is hard enough shooting handheld a 50/1.2 lens and getting what you intend in focus.

Nodda Duma · Sep 22, 2020

Your thread title says it all. A 50mm lens with a 100mm aperture provides a paraxial f/# of 0.5, but the working f/# is significantly slower due to cos^4 law roll-off even on-axis, where f/# is usually specified. At f/0.5 the on-axis marginal rays intercept the image plane at 45 degrees, at which angle relative illumination is down 25% for the marginal rays due to cos^4 law. Quick and dirty estimation tells me that, on-axis, the effective speed is thus actually around f/0.9-f/1.0, but with exponentially worse image quality compared to an actual f/0.9 lens due to aberration dependency on f/#. The same rough calculation says Kubrick’s f/0.7 lens was effectively f/0.85 or so (marginal ray illumination is down to about 47% on-axis)

Off-axis it’s even worse, as at some point the ray angles exceed Brewster’s angle (about 54 degrees for digital camera cover plates, 56.5 degrees for gelatin) and simply reflect off the imaging media, where they then bounce around inside the camera and then contribute to veiling glare and loss of contrast. Effective f/# off-axis would drop to f/4 or less even without vignetting.

An *effective* f/0.5 cannot be achieved, as Brewster angle and cos^4 limits effective f/# to 0.7 or so..it’s an asymptotic approach iirc.

So there you go. Note this limit has nothing to do with technology to fabricate / assemble, practicality of using a lens, nor even image quality. Technology and practically do not limit design work. Laws of physics, however, cannot be broken and create a true limit.

Once you can follow what I wrote, you will no longer have to say you don’t understand why they don’t make a 50mm f/0.5 lens.

-Jason

guangong · Sep 22, 2020

If memory still serves me, the Canon 0.9 lens did not have a very good reputation and was not a market success because pics not that good and lens just too large to be practical. I haven’t seen one in person since about 1970. High prices most likely due to collectors’ interest rather than photographers’.
However, I found this thread fascinating. Kudos to all contributors!

NB23 · Sep 22, 2020

A lot of useless talk about perfectness at f0.5.

If you don’t have enough perfectness to your taste at f0.5, I’m sure that this state of the art lens would have an aperture ring.

pentaxuser · Sep 22, 2020

Ko.Fe. said:
.
This is why Canon now makes tele lenses with F8 maximum aperture.

So Canon now restricts all its zooms to a maximum of f8? At the Wimbledon tennis championships each year you can always tell the Canon sports photographers from the Nikon ones due to the light grey lenses of Canon v the black ones of Nikon and yet the Canon ones look to have similar max aperture to Nikon and judging by their size they all look to have more than f8 as a max aperture

pentaxuser

AgX · Sep 22, 2020

Ko.Fe. said:
Lens like this is going to be useless. At f0.5 here is no DOF for practical photography.

And as Reddesert pointed out, even if the theoretical DOF is sufficient for the purpose, still it must be at all possible to achieve it by correct focusing.

Dan Fromm · Sep 22, 2020

reddesert said:
What I'm saying is that a lens approaching the theoretical limit of f/0.5 - or even f/1 - is difficult to make, places strong requirements on the equipment it is used with (eg tolerances of the body, focus system, film flatness), and requires a lot of care from the user. And thus it could easily wind up with disgruntled users who complained that it made blurred images and wasn't sharp, even if that wasn't the fault of the lens. Any manufacturer would have to consider this. Most of the super-speed lenses near f/1 have been (IMO) expensive curiosities used for cine/TV and/or bragging rights..

Red, consider the humble Canon 310XL super 8 cine camera. It has an 8-24/1.0 lens. Guess focus; autoexposure, but not TTL which costs light. I have several. Within their limits they produce much better footage than my slightly more expensive Beaulieus with, respectively, 6-66/1.8 and 6-70/1.4 Schneider lenses.

Canon sold many 310XLs and variants.

ic-racer · Sep 22, 2020

I use a f 0.9 13mm Switar on my Non-Rex Bolex H8.
It is a little difficult to use at f 0.9 because, since the shutter speed is fixed at 1/50 sec, the lighting conditions have to be somewhat fixed at a certain level. That camera does not have an adjustable shutter.

H8 REX has an adjustable shutter but fastest lens in that system is f1.4,

Chan Tran · Sep 22, 2020

_T_ said:
I don't understand why we can produce 35 pound 600mm lenses with front elements so large we can't use filters at the front of them, but we can't have a 50mm lens with an aperture of f/0.5 and an entrance pupil of 10cm.

Why is that? Will people not pay for that? Is it difficult to engineer?

It would have shallow depth of field wide open but would have over 1m of depth of field at a 10m focusing distance. It wouldn't be unusable. And we shoot with shallower depth of field in macro.

What am I missing here?

Perhaps simply making the large elements won't make the lens fast. I don't know for sure but there may be some reason. Also is there a demand for it? When I said demand I mean if you make it can you sell many of them? But there is this lens
https://petapixel.com/2013/08/06/carl-zeiss-super-q-gigantar-40mm-f0-33-the-fastest-lens-ever-made/

Ko.Fe. · Sep 22, 2020

pentaxuser said:
So Canon now restricts all its zooms to a maximum of f8? At the Wimbledon tennis championships each year you can always tell the Canon sports photographers from the Nikon ones due to the light grey lenses of Canon v the black ones of Nikon and yet the Canon ones look to have similar max aperture to Nikon and judging by their size they all look to have more than f8 as a max aperture

pentaxuser

Canon makes all kinds of tele lenses. No restrictions. Canon L lenses are white or black. But neoprene covers could be any colors.
I'm not sure Wimbledon was covered with any of RF tele lenses I have mentioned.

Impossibly fast 50mm?

Privacy & Transparency

Privacy & Transparency