Hyper-speed lenses

[Paul Bruening]

I have been thinking about ideas from other threads:

 

http://www.cinematography.com/forum2004/in...mp;#entry211461

http://www.cinematography.com/forum2004/in...showtopic=33127

http://www.cinematography.com/forum2004/in...showtopic=33034

 

The Barry Lyndon quote in David's post made me wonder about front element factors in lens design. As well, the distance from the rear element to the film plane. Is it correct for me to assume that a larger front element gathers more light because of its greater surface area? Is it correct for me to assume that rear element proximity effects light value, therefore a shorter distance is brighter than a longer distance (it is this way with bellows adapters on lenses. More length=less light)?

 

So, let's say you could use the ideas of converting a larger format, fast lens down to a smaller frame size like Techniscope and get a nearly X.5 reduction. Then you pile on the idea of using a larger front element to gather a greater amount of light. Then, pile on the idea of moving the rear element of the lens closer to the film plane (closer on a rack-over than a reflex, of course)... Assuming all those ideas could be designed into one lens and allowing for some compromise of light loss from those bigger front elements: How fast could a lens be?

[Paul Bruening]

Hey Mods,

 

I put this in the wrong category. Would someone be so kind as to move it to "Lenses"?

 

Thanks,

Paul

[Paul Bruening]

You know, "shutters built into the camera works" are a left-over from the pre-digital age. You could build the lens with the aperture plate on it and run the rear element almost right up to the film plane and digitally control the shutter as a built into the lens kind of thing. Arri puts the back half of the film sandwich in their mags. Why not put the front half into the lens? This would help with the tighter tolerances of FFD, as well.

[Paul Bruening]

AND, if the front element being large is purposed to gathering more light, why does it need to be a biconvex thing? Why not put a thin, diopter shaped lens (concave-convex) on the front that does nothing more than gather huge amounts of light and transmit it to the normal front lens element. That would be lighter in weight. Each front element would have to be ground to match the lens and fitted with an appropriate barrel that would still be heavy. But, it would be lighter over-all than a whopping-big, biconvex, front element. Faster, as well.

[Paul Bruening]

Take a moment to think of what you could accomplish with an f0.20 lens that still had a DoF of f1.2? You could shoot an entire movie of urban, night exteriors with no light set-up. You could shoot interiors with only end table lamps. AND you could do this all on slowish, high resolution film stocks.

 

... if only.

[Paul Bruening]

Could you put a kind-of reverse LCD projector type sheet at the back of the rear element? Let's say the last surface of the element was a plano-face. A micro-thin, transparent LCD sheet could be placed between the element and the micro-flange-focal-distanced film plane to provide reflex viewing of the image. If that could be done reasonably, then that would solve the reflex mirror getting dumped (postulated in a previous post). All of these ideas put together would make a pretty complex and expensive lens. But, with a zoom this fast with all these ideas involved, you could still get all the goodies but buy fewer lenses. Or, the hell with it... sell meth and buy a whole set of primes within a week.

[Paul Bruening]

I don't have a clue if there is even any kind of transparent LCD, sensor, type thingie. But, if there was, could you use it to actually accelerate, add-to or multiply light? It would be a sort-of "gain" device in the lens but for film applications.

 

As you might have guessed, I love film.

[Paul Bruening]

I am film's bi-otch.

[Paul Bruening]

An LCD senso-gain-filter could act as reflex viewer, gain device and digitally processing filter. You could spontaneously tone map the image. Let's say you are in a daylight exterior and the sun is causing a harsh 3.5 stop contrast ratio on your leading lady's face. You could dial in the amount of ND filtration on her bright side only and eliminate the need for any of those pain-in-the-ass 12K HMI fill lights. With a powerful enough CPU, you could vary the tone mapping to change the ND factor at each pixel sight and have a variation of filtration depending on the amount of light. Much of what you can do in post you could do during actually shooting. Luminance, color timing, artistic variations.

 

Damn. What if?

[Paul Bruening]

I'll call it [i}Hypersmart Lens[/i]. It could gain the dark sides and filter the bright sides. If you were shooting a night exterior with a single point of light. You could bring down the high contrast by brightening the dark side and darkening the bright side.

[Guest Glen Alexander]

AND, if the front element being large is purposed to gathering more light, why does it need to be a biconvex thing? Why not put a thin, diopter shaped lens (concave-convex) on the front that does nothing more than gather huge amounts of light and transmit it to the normal front lens element. That would be lighter in weight. Each front element would have to be ground to match the lens and fitted with an appropriate barrel that would still be heavy. But, it would be lighter over-all than a whopping-big, biconvex, front element. Faster, as well.

 

Pick up and read Borne and Wolf, or at least the lens maker equation.

 

If you put a big massive lens up front, say 8.5", most are designed to have a focal length around 1000mm. You need an initial lens with a massive curvature to get down to something like 50mm, without have a telescope for a lens. massive curvature tends to separate the wavelengths of light so that don't focus at the same point/region. Massive curvature is also a lot harder to maintain tolerance over the surface.

 

Go ahead and get a celestron 10" or 14" lens and try to build a 50mm F0.2.

[georg lamshöft]

The physical f-stop limit is f0.5, which is one stop faster than the famous planar used by Kubrick.

[Simon Wyss]

The physical f-stop limit is f0.5, which is one stop faster than the famous planar used by Kubrick.

The theoretical limit is f 0.5, in practice one ends at f 0.53 where the light is just grazing the front element. F 0.7 is already very tricky to compute. An alternative is the mirror lens.

[Guest Glen Alexander]

The theoretical limit is f 0.5, in practice one ends at f 0.53 where the light is just grazing the front element. F 0.7 is already very tricky to compute. An alternative is the mirror lens.

 

Yes, i looked at designing a curved mirror but the manufacture said they couldn't maintain tolerance over the curvature and didn't want to quote. .5 limit is only for 'simple' convex or concave lens. with a parabolic lens, it is possible.

Edited November 19, 2008 by Glen Alexander

[Simon Wyss]

Never despair! There is Mr. Kouchatchki a few kilometers from where I live who has ground and polished his own concave mirror from pure aluminium. He says it is better than glass. Now he observes the stars with the best reflector on earth. In case of interest I can provide you with the address.

[Paul Bruening]

Never despair! There is Mr. Kouchatchki a few kilometers from where I live who has ground and polished his own concave mirror from pure aluminium. He says it is better than glass. Now he observes the stars with the best reflector on earth. In case of interest I can provide you with the address.

 

Can he grind glass elements?

[Simon Wyss]

I'd have to ask him. How much time will you give me ?

[Guest Glen Alexander]

These are the people I contacted about my design

 

http://www.jmloptical.com/

 

They are willing to do one offs for a reasonable price if they can meet your specs.

[Guest Glen Alexander]

Never despair! There is Mr. Kouchatchki a few kilometers from where I live who has ground and polished his own concave mirror from pure aluminium. He says it is better than glass. Now he observes the stars with the best reflector on earth. In case of interest I can provide you with the address.

 

how can he guarantee surface smoothness? what are his tolerances? what does he use for such precise measurements? does he have a flatbench? how does he quanitify comatic aberrations?

 

i would be somewhat hesitant, aluminum doesn't have a crystalline structure and there is NO such thing as "pure" aluminium, there are always grades, 6061, 5051, 5356, medical grade, etc,.....

 

there are methods to reduce the surface roughness but one guy by hand seems like it would take a long, long time. why doesn't he put a silver finish on his mirror?

[Simon Wyss]

Man, you're eating me up with your questions.

 

He is a hobby astronomer and sole hand worker, has no instruments, trial and error until he reaches the result. On the JML flash intro I read High Speed Manufacturing. There you are: either a solitaire which takes time or a company eager to set up series.

 

I cannot cope with your interest in entering manufacture details. Seems to me that you have to know the adventure for yourself. Most certainly you can find an optical specialist in your country. If you'd insist I should politely ask you for the dimensions before we discuss surface roughness.

 

So long

[Guest Glen Alexander]

Man, you're eating me up with your questions.

 

He is a hobby astronomer and sole hand worker, has no instruments, trial and error until he reaches the result. On the JML flash intro I read High Speed Manufacturing. There you are: either a solitaire which takes time or a company eager to set up series.

 

I cannot cope with your interest in entering manufacture details. Seems to me that you have to know the adventure for yourself. Most certainly you can find an optical specialist in your country. If you'd insist I should politely ask you for the dimensions before we discuss surface roughness.

 

So long

 

 

thanks but now the lens/mirror is virtually unmakeable unless you want to deal with horrific color separation issues.

[Paul Bruening]

I'd have to ask him. How much time will you give me ?

 

No rush at all. Just something to ask when you're around him for your own reasons.

[K Borowski]

You have to keep in mind that super-fast glass is going to have very poor DOF and probably very poor MTF as far as maximum resolvable LP/mm. All lenses have the tendency to photograph quite poorly wide open.

 

So isn't F/0.7 fast enough? It was fast enough for Kubrick, and he had basically 2 1/3 to 3 stops less speed to work with . . .

 

You could easily push '19 to say 1600 speed.