0

I know that to get better image quality often requires aspherical lenses, which are expensive to make, so at some point it becomes non-economical to have a lot of aspherical elements in a camera lens.

However, let's suppose that a photographer wanted to commission a custom lens from one of the companies that make such lenses. The cost of this would be in six figures, maybe $150,000 if we add in lens design costs, something like that. Would it be possible to significantly outperform available stock lenses by doing this? By "significantly outperform" I mean achieve a full stop of additional light transmission while holding equal to or better to a comparable lens in sharpness, distortion, vignetting and chromatic aberration.

For example, take the Zeiss Batis 25mm f/2 Sony FE as the comparable lens, which DXO mark calls the highest scoring wide angle they have tested. Would it be possible to commission a custom lens that have similar performance, but be full T-stop faster?

Clickety Ricket
  • 1,854
  • 1
  • 18
  • 35
  • "Would it be possible ..." is simply to vague and unless something is a physical impossibility then the answer will always be yes. I don't even see a way to ask "Could a custom 25mm f/1.4 be designed to match the sharpness of the stated lens" would be any different. – StephenG - Help Ukraine Jun 13 '20 at 02:30
  • Anything is possible with enough money. But to get faster than that lens you're looking at tens, if not hundreds, of millions of dollars for development and production of a very limited number, not hundreds of thousands. You'd probably also need to commission a cart to carry it around on. – Michael C Jun 13 '20 at 06:25
  • I suspect the cost for designing and testing a lens is significantly higher than what you are thinking - the companies that make lenses get the cost down as low as they do (even though it may still seem high) by making and selling lots of copies of that lens, not just one. – twalberg Jun 13 '20 at 12:52
  • I don't know what would make you think it takes $10-100M to produce the first quantity of a lens, but you can reduce that to $10-100k. – Brandon Dube Jun 13 '20 at 20:20
  • @BrandonDube Yes, for more pedestrian lenses. For a lens that performs as well as the 25mm f/2 Batis and is a full stop faster (e.g. 25mm f/1.4), it's a little more complicated as you should be well aware. Either the elements are huge and need more correction because of their larger diameters, or some amazing new groundbreaking technology needs to be developed. Developing technology that is at, or even just past, the state of the art is much more expensive than applying existing technology to a mundane application that doesn't stretch any limits. – Michael C Jun 19 '20 at 14:20
  • Could you walk me through how you plan to spend $10-100M per unit on an initial quantity of lenses? Doubling the size still fits in a 77mm filter size (see: Canon, Nikon etc) -- if you release that, crossing 100mm "spikes" cost, since the maker must move from 4 to 6" interferometers. Crossing 6" costs much more, but you won't do that for a 25/1.4. – Brandon Dube Jun 20 '20 at 16:57
  • @BrandonDube I don't have plan, just a dream. I am just trying to figure how much of lens design is economics of scale versus cost cutting to garbage. If creating really good lens is just a matter of grinding asphericals, and I can one-up classic lens for paying for that grinding, then that is the win I am looking for. If making a good lens is a 500k proposition, then obviously is it not feasible (at least for me). – Clickety Ricket Jun 21 '20 at 05:07
  • My comment was directed to Michael C. I'm not sure they understand just how much money $10M is compared to the cost of materials an time. Directly to you, there is no concept of economy of scale in lens design. Design is an early phase of a project before you have made anything. You can use aspheric optics to produce better lenses, but they aren't a magic bullet. You will pay $5-10k in startup costs for each one you want in your design. You also don't just grind lenses, you polish them. You need that both to reach the form tolerance and get rid of roughness. – Brandon Dube Jun 21 '20 at 16:31
  • The lens you want is basically a cine lens. Panavision one-offs are about 70k for a spherical (meaning: not anamorphic) lens. If you want the level of performance in the 25/2, perhaps 100k for additional aspheres. It will be huge, heavy, and manual focus. – Brandon Dube Jun 21 '20 at 16:32

3 Answers3

2

Your question boils down to "is the 25/2 Batis the pinnacle of optical design; is it not possible to do any better." The answer is no; it is a local optimum of performance size and price.

Some companies that will make custom lenses for you include Panavision, Caldwell Photographic, and JML Optical among enumerable others. You can expect to pay $10-$100k per unit for the first quantity in a "mirrorless camera" sized range. To increase the aperture by a stop will likely more than double the length and diameter.

Brandon Dube
  • 2,445
  • 14
  • 24
1

When you are the NASA you can have lenses made to your specs, in particular the Zeiss Planar 50mm f/0.7. But if this lens also had a noted career in cinema, its original purpose was more reconnaissance/surveying.

There are also lens that Hasselblad made for the Apollo missions. But if one was designed for this from the ground up (the Biogon f-5.6/60), other lens were mostly modified to carry bigger markings and be usable with thick gloves. But the Biogon had a commercial and terrestrial career afterwards, so even the NASA didn't carry all the development costs.

Of course the USAF/CIA also had their lenses designed for them for use in "spy planes" such as the U2, A12 and SR-71...

xenoid
  • 21,297
  • 1
  • 28
  • 62
  • In the wiki article you linked for the Zeiss Planar 50mm, it says that only 10 of the lenses were made. If it was such a great lens, why didn't people make more of them or clone them? – Clickety Ricket Jun 13 '20 at 15:00
  • Terribly expensive, and very hard to use (I'll let you determine the Depth of Field at f/0.7) – xenoid Jun 13 '20 at 17:22
0

Lens design is always a compromise and typically you can't improve something without degrading another aspect of the performance, but since you specifically said:

achieve a full stop of additional light transmission while holding equal to or better to a comparable lens in sharpness, distortion, vignetting and chromatic aberration

then Yes, you can!

And you don't even have to spend time designing a new lens (which can always fail if the goal is too ambitious)

Instead, build a scaled up version of the original lens. By the laws of geometry, it will retain the same optical properties, except that everything will be just bigger - which means we also need a bigger sensor to cover the enlarged image circle.

The focal length increases but it is compensated by a bigger sensor, giving the same 35mm equivalent focal length (and the same field of view).

The f/ratio remains the same, but the amount of light increases with proportion to the geometrically larger aperture, so each of your scaled up pixels of your scaled up sensor receives more light so it behaves as if the ISO was multiplied by 2 (for +1EV).

Yes, the answer is a bit stretched, as it also calls for a new camera, but it seemed you might be interested in theoretical considerations, not just in ordering a custom built lens ;-)

szulat
  • 5,059
  • 1
  • 20
  • 31
  • Okay, that makes sense. So, let's say someone wanted to do this, then they would have know how the original lens was constructed to upscale it. Is that easy to do, or is it difficult to reverse engineer a lens? – Clickety Ricket Jun 13 '20 at 14:58
  • If you can produce a state of the art lens, you are probably able to reverse engineer your competitor. But then you are serious enterprise who wont break patent laws? I don't know... and there is a similar question here: https://photo.stackexchange.com/questions/65126/are-there-any-secrets-in-lens-design – szulat Jun 14 '20 at 00:14