Previous news story    Next news story

Miniature wide angle lens under development at UCSD

By dpreview staff on Sep 27, 2013 at 21:40 GMT

Researchers at UCSD's Jacobs School of Engineering are working on a miniature wide angle lens by taking advantage of the benefits of spherical lenses. At just a tenth of the size of a traditional wide angle lens, a spherical lens can create wide angle images without chromatic aberration or loss of resolution at corners. The challenge is capturing the lens' spherical projection on a flat sensor. The team have overcome this by using optical fibers fused to the rear of the lens to relay light to electronic sensors.

Researchers were initially concerned that focusing the lens would require the optical fibers to move back and forth, but found that the system boasts large enough depth of field that movement wasn't necessary. Early tests suggest the lens is capable of rendering images of very high resolution - 0.2 milliradian. That translates to 0.011 degrees resolution, or around 7800 lines per height. 

A prototype lens from UCSD's engineering school. The miniature lens is one tenth of the size of the Canon EF 8-15mm f/4L pictured beside it, yet would boast a resolving power of 0.2 milliradian. However, because of its use of optical fibers and monocentric lens, it's not likely to be used in an interchangeable lens system.

The project's engineers describe the design as using monocentric lenses with a bundle of glass optical fibers aligned to the lens' surface in order to relay the light collected by the lens to electronic sensors. Thanks to the lens' spherical design, resolution is (in theory) the same across the image field, making possible a kind of 'zooming without moving parts' anywhere in the field of view without encountering loss of resolution or off-axis lens aberrations.

The UCSD team is currently working with a design for a 12mm focal length with a 5 megapixel sensor and a 30 megapixel prototype. Next, they're hoping to build a system giving an 85 megapixel image.

It seems unlikely that this kind of optical system would be used by smartphones as the school's press release suggests, and more likely that its 'pan and zoom with no moving parts' capabilities would be put to use in surveillance equipment. The project's DARPA funding suggests that the lens will find a home in military applications, where a very wide angle of view rendered without a loss of resolution would be genuinely useful. It probably won't appear on a camera near you anytime soon, but it's an interesting proof of concept.

Via: Engadget, Source: UCSD Jacobs

Comments

Total comments: 72
saralecaire
By saralecaire (7 months ago)

That's cool, now engineer a miniature full frame sensor, miniature DSLR body with all the bits, now gnomes can take up photography too!

0 upvotes
yabokkie
By yabokkie (7 months ago)

it's for ultra wide angle only (> about 100 degrees).

Comment edited 4 minutes after posting
0 upvotes
wkay
By wkay (7 months ago)

Yes, our retina is spherical and we perceive straight lines as straight.

0 upvotes
morepix
By morepix (7 months ago)

Well, maybe what we call "straight lines" are actually circles. Who knows?

0 upvotes
forpetessake
By forpetessake (7 months ago)

That also means if they can grow (or etch) micro-fibers right on the silicon, they can create sensors corrected for curvature and also improve wide angle sensor performance.

0 upvotes
Roland Karlsson
By Roland Karlsson (7 months ago)

Strange comparison. Comparing the size of the fix focus, fix focal length experimental lens to a commercial Zoom lens. A zoom lens that is for an SLR camera with large flange distance, i.e. a retro focus lens. A zoom lens that can both focus and change aperture.

And then say it is 10 times smaller. 10 times smaller in what dimension? Length, width, lens front area, volume?

Comment edited 53 seconds after posting
2 upvotes
yabokkie
By yabokkie (7 months ago)

a comparison to fool the fools.

well, a spherical focal surface design does make lens much smaller, and the story may look less false-sounding if teller can explain it in a honest way. youthful vigor can back fire and lose credit.

Comment edited 10 minutes after posting
1 upvote
Paullubbock
By Paullubbock (7 months ago)

Uh, the test lens is pictured next to the wide angle lens for comparison. (to the left)

Comment edited 27 seconds after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

it's compared with a wrong lens which destroys the credibility of the article in the first second.

a better comparison may be wide angles for large format cameras. they are large but with large apertures, too, and they don't have the long flange-back issue of SLRs. but still we cannot adjust focus with the new lens, not with bloody efforts.

it will get bigger and bigger, more and more expensive, and lower and lower quality as we add features to it.

Comment edited 3 times, last edit 12 minutes after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

it may be an interesting idea to design a lens-sensor module for SLR and mirrorless mounts. it can get power from the mount and can use wireless to talk with the camera if it's too difficult to have high speed data through electrical contacts (a hybrid GXR-QX, liveview only for SLR).

0 upvotes
Frank_BR
By Frank_BR (7 months ago)

An alternative to using a bundle of optical fibers is using a curved sensor. The planar process is the primary process by which monolithic integrated circuits are built, but nonetheless curved integrated circuit, including image sensors, can also be made. The article below from Stanford University is just about curving monolithic silicon for image sensors:
http://apl.aip.org/resource/1/applab/v92/i9/p091114_s1?isAuthorized=no

"… a spherical lens can create wide angle images without chromatic aberration or loss of resolution at corners."

I believe there is an imprecision here. A "spherical" (monocentric) lens is intrinsically free from lateral chromatic aberration, but not necessarily from longitudinal CA.

0 upvotes
Just another Canon shooter
By Just another Canon shooter (7 months ago)

Also, when you project the image in the spherical sensor on a plane (to display it on, well, a flat monitor), the LoCA away from the center is magnified, together with the other aberrations.

0 upvotes
yabokkie
By yabokkie (7 months ago)

a reason why spherical focal surface is because we are very bad at optics but quite good at digital. also aberrations away from the center will become less visible seen from the center.

Comment edited 6 minutes after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

on a 8k display (viewed 110 deg wide), angular sizes of pixels at far left and right are only about 0.6 times as big as center.

0 upvotes
Roland Karlsson
By Roland Karlsson (7 months ago)

Spherical lens? That's the normal stuff we always have had. If we want to be more exotic, we talk about aspherical lenses.

The lenses are monocentric, i.e. it is a number of spherical shells.

Comment edited 4 minutes after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

maybe the auther will call all spherical and aspherical lenses we have "flat" ones (flat focal plane).

Comment edited 1 minute after posting
0 upvotes
JLim22
By JLim22 (7 months ago)

high resolution from such a small lens is great, but why are they comparing 1/3.2" sensor and lens to a full frame slr camera and lens?

perhaps they should really be comparing with the lens inside of a gopro in terms of size?

0 upvotes
yabokkie
By yabokkie (7 months ago)

it's valid to compare what the lens can do in terms of field of view, aperture (and transmission light gathering capacity), and aberrations/resolutions as a photographic lens.

it doesn't make sense to compare weight and size with EF8-15/4L for it's not designed for EF or any other SLR mount. it's not even mirrorless but rather a data-only "mount", like Ricoh GXR (physical connector) or Sony QX (wireless).

Comment edited 3 times, last edit 4 minutes after posting
0 upvotes
AlanG
By AlanG (7 months ago)

They are only sampling from a very small area of this lens to show what is possible. This is like cropping in on a full frame sensor's image. Thus the comparison is valid.

Future models will have many more fiber optics feeding into a number of sensors and will cover a wider field of view by assembling the images from several sensors. The concept is that the area of all of these sensors added up will produce a narrow view, normal, or wide angle image from this tiny lens that outperforms current full frame cameras and lenses.

As for comparing this to the Canon lens, we'll have to see how large and heavy it gets once all of the fiber optics bundles and sensors are added. The only connection between that and the rest of a "camera" will be electrical so any kind of mount could be used to attach them.

Comment edited 3 times, last edit 5 minutes after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

the design makes interchangeable lens very difficult if possible (no hope for SLR). and one will have at least make it an interchangeable lens before comparing with one.

but maybe we will find it inside a Mirokia 4080 with 160MPix, though the 20mm diameter will make it quite thick.

hope Oly can find a way to do 5-axis stabilization for a m4/3" version of the lens.

Comment edited 2 times, last edit 9 minutes after posting
0 upvotes
AlanG
By AlanG (7 months ago)

Well of course it is nothing like an interchangeable lens on a DSLR. But why can't you compare the images from them? E.g. one total imaging system compared with another.

Consider that there are also a lot of uses for photo imaging besides DSLRs and general photography. They mention right in their paper that one goal is to have smaller cameras and thus smaller UAVs (Drones.) Why do you think this is a DARPA project? Even now individual soldiers can carry and launch small recon aircraft and they also have rc controlled ground based devices that can creep into a building and look around.

This just happens to be posted here at DPReview but don't jump to the conclusion that it immediately has to be relevant to how you currently take photos.

Comment edited 2 times, last edit 7 minutes after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

> why can't you compare the images from them?

if you didn't read my post above, read it now please.
(why it's small is because there are things it cannot do)

Comment edited 2 minutes after posting
0 upvotes
AlanG
By AlanG (7 months ago)

Try to understand that this is a prototype design just to prove concept. You are questioning things that are not relevant to the item at this point in time.

They simply are showing that if you take a narrow area from their lens (which is all they can do at the moment) it produces a better image than that same area cropped from a larger sensor using this much larger Canon lens.

Comment edited 2 minutes after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

please read this again:
why it's small is because there are things it cannot do.

the spherical focal surface does contribute to smaller size but the comparison is still very unfair for the lens will put on a lot weight if designed for SLR, so no valid comparison.

you should know that we trade off a lot of things for interchangeable lens mounts, from size to image quality.

Comment edited 4 times, last edit 6 minutes after posting
0 upvotes
AlanG
By AlanG (7 months ago)

Forget the camera mounts as this will never be used on an SLR. Simply look at the size and complexity of the glass elements in the two different lenses and study the resolution this smaller optical formula can achieve with fewer aberrations. A spherical design is much smaller and much simpler than a reverse telephoto wide angle design. By the time they add the fiber optics and the sensor it will be larger but should still be much smaller than a comparable DSLR lens/24x36 sensor combination.

They posted very clear illustrations of this on one of the links. It is a new way to create a lens/sensor imaging system. There is no point trying to shoehorn it into your SLR solution.

Comment edited 3 times, last edit 3 minutes after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

> forget the camera mounts

I agree that turn back to the reality is always the easiest thing to do.

Comment edited 30 seconds after posting
0 upvotes
DonM999
By DonM999 (7 months ago)

An interesting concept here; using bundled fiber optics as a "lens".

0 upvotes
yabokkie
By yabokkie (7 months ago)

comparing with an SLR lens is the rubbish part of the story.

0 upvotes
Francis Carver
By Francis Carver (7 months ago)

Wow, a new, small lens. And only from Caliweirdia!

0 upvotes
CyberAngel
By CyberAngel (7 months ago)

My next Lumia,,,

0 upvotes
Michael Ma
By Michael Ma (7 months ago)

UCSD is the best school ever!

0 upvotes
tjbates
By tjbates (7 months ago)

Fascinating subject. One of the main reasons why FF has become less desirable is the size and weight of lenses.
A different design approach to the lens/ sensor relationship could help miniaturize lenses.

0 upvotes
PLAMBERT
By PLAMBERT (7 months ago)

I was put off a Nikon D4 and zoom lenses due to the total weight of the outfit so bought a Leica M9 which weighs much less. No zoom lenses really, just a range of fixed focal lengths by Leica and others. Buy a lot of lenses and you get a very heavy outfit! I could manage with 4 or 5.
Philip

0 upvotes
lazy lightning
By lazy lightning (7 months ago)

dpreview is beginning to sound a lot like a DSLR Anonymous meeting.

Hi, my name is Bob and I used to own a DSLR. I found myself looking at a FF DSLR in BestBuy yesterday, you know, for times when I really need the image quality but didn't succumb to the temptation. I've now been DSLR free for 15 days but I just have to keep taking it day by day.

4 upvotes
HowaboutRAW
By HowaboutRAW (7 months ago)

PLAMBERT:

Well it’s not like this engineering school project is likely to be widely available in the camera market in say 5 years.

Then the D4 and the M9 are significantly different cameras.

One is incredibly good in lowlight, with a fast AF, and a staggering frame rate, but doesn’t have optically good lenses–unless you use Zeiss manual focus lenses.

And the other, the M9 of course, has staggeringly good lenses, is smaller, but is not good in lowlight, and has very slow buffer.

So the two cameras aren’t exactly in the same category.

0 upvotes
AlanG
By AlanG (7 months ago)

Nikon does not have "optically good" lenses?

If this truly works as they say, I would not be surprised to see this concept developed fairly quickly for some applications.

Comment edited 4 minutes after posting
0 upvotes
HowaboutRAW
By HowaboutRAW (7 months ago)

AlanG:

"Good" yes, but nothing close to Leica or Zeiss. So I should have written: "doesn't have optically excellent".

Hope that clarifies things.

Don't bother with DXO scores for lenses.

0 upvotes
yabokkie
By yabokkie (7 months ago)

> "Good" yes, but nothing close to Leica or Zeiss

good it doesn't stand besides rubbish.

0 upvotes
Just another Canon shooter
By Just another Canon shooter (7 months ago)

Unfortunately, to keep the light gathering ability, the lens/light relationship is what matters, and you cannot collect the same light from a smaller area.

Comment edited 11 seconds after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

> same light from a smaller area

yes if it's aperture area.

0 upvotes
AlanG
By AlanG (7 months ago)

A larger version of the same lens could have more fiber optics to pipe more light to more sensors. Or it could send the light to the same number of megapixels but they could each be of a larger pitch and thus gather more photons in a given time period.

0 upvotes
yabokkie
By yabokkie (7 months ago)

more light has to come in from the aperture in the first place.

people don't understand sensor surface has no say here. they should at least see it through "Speed Booster" and it's really hard to not see it through "fibers".

the sensor area is simply irrelevant.

Comment edited 2 times, last edit 11 minutes after posting
0 upvotes
groucher
By groucher (7 months ago)

There seems to be no reason why a spherical section sensor couldn't be manufactured. There would then be no need to alter the spherical projection by such a complicated method. Wonder if it's been tried?

1 upvote
Revenant
By Revenant (7 months ago)

But wouldn't the projection then have to be digitally "mapped" to a flat image post-capture?

0 upvotes
AlanG
By AlanG (7 months ago)

I question if there is any existing manufacturing technology to make a spherical sensor. And if such an item could be made, what would it cost?

0 upvotes
stevens37y
By stevens37y (7 months ago)

If you have a method how to do it you should immediately have patent for that.

0 upvotes
Photomonkey
By Photomonkey (7 months ago)

All very nice but will they be wether-sealed and have a red stripe?

2 upvotes
RichRMA
By RichRMA (7 months ago)

Aspherical, ASPHERICAL!!!

0 upvotes
yabokkie
By yabokkie (7 months ago)

aspherical means planar here.

0 upvotes
Joe Pineapples
By Joe Pineapples (7 months ago)

Antenna engineers have been using spherical lenses since the 1960s at least...

0 upvotes
Chris Noble
By Chris Noble (7 months ago)

How fast is that lens? f22 or so? Not impressed. Resolving power is a solved problem for most photographers.

0 upvotes
AlanG
By AlanG (7 months ago)

That kind of design could be very fast.

2 upvotes
Chris Noble
By Chris Noble (7 months ago)

I don't think so. A big spherical lens would be impractical.

0 upvotes
AlanG
By AlanG (7 months ago)

A 12mm focal length only needs to be 6mm in diameter to be f2. By comparison, look at how large that Canon lens is to achieve f4. And the Canon lens is a fisheye lens. A rectilinear design would be larger.

If you follow the link you'll see the lenses they use are sort of like marbles.

And while we do have high resolving systems today, the idea is to get better resolution in much much smaller packages. If you filmed videos from a quadcopter as I do you'd appreciate the advantage of miniaturization. Just look at what the tiny GoPro has done for that field.

Comment edited 4 times, last edit 8 minutes after posting
8 upvotes
zoob
By zoob (7 months ago)

Follow your logic a little, here.

Does F-stop really relate to the ability for a lens to collect photons? Hmmm. Can the F-stop in a smartphone lens truly be compared to Canon EF 8-15mm?

0 upvotes
Andy Westlake
By Andy Westlake (7 months ago)

The researchers' paper shows an optical design for a 12mm F1.7 that's 20mm long and weighs 16g.

2 upvotes
AlanG
By AlanG (7 months ago)

"Does F-stop really relate to the ability for a lens to collect photons? Hmmm. Can the F-stop in a smartphone lens truly be compared to Canon EF 8-15mm?"

Of course if the F stops are the same the light hitting a given area will be the same. (Assuming comparable transmission ability of the lenses.) The difference is that this Canon lens can cover a larger sensor so it passes more photons in total over a given time period. (It is up to the sensor to collect these photons.)

Comment edited 2 minutes after posting
0 upvotes
Chris Noble
By Chris Noble (7 months ago)

Thanks Alan.

0 upvotes
AlanG
By AlanG (7 months ago)

Keep in mind that while this model is a very wide lens, they are only sampling the image from a very narrow area by using a single small sensor. This is like an extreme crop of an image. But once they are able to run the fiber optics out from a larger area of the lens, section by section, they will have a wider view being "projected" onto a number of sensors. That is the way they will be able to collect a lot of photons from such a small lens.

This is an incredible breakthrough for photography that is only possible by digital technology. At one time the technology for digitizing music and reading via a laser was cutting edge but it very rapidly moved to CDs and readers.

Comment edited 2 times, last edit 5 minutes after posting
0 upvotes
yabokkie
By yabokkie (7 months ago)

the focal lengths look like equivalent ones converted from angle of view, which is more important than focal length.

light gathering capacity is defined by two factors, aperture area and (solid) angle of view (nothing to do with sensor size or f-number).

Comment edited 3 times, last edit 4 minutes after posting
0 upvotes
Artistico
By Artistico (7 months ago)

Eyes are tiny and have very good resolving power. Seems like camera technology is closing in...

5 upvotes
Felix E Klee
By Felix E Klee (7 months ago)

I wonder if it's possible to connect a biologic eye to an electric circuit. ;-)

1 upvote
Keenorhian
By Keenorhian (7 months ago)

Eyes are quite crappy as an optical device. Brain does all the post-processing, that's why it looks good.
To have a good picture you have to have big lens. You can't do much about it, it is the law of physics. If still in doubt, just look at the best telescopes in the world - the bigger the better.

2 upvotes
Dheorl
By Dheorl (7 months ago)

Well they have a very good resolving power over a very small area. The edges are absolutely shocking though.

2 upvotes
AlanG
By AlanG (7 months ago)

More likely is the idea of having a "digital" eye and feeding that information directly into your visual nerves (if viable) or maybe directly into the brain. This would be to counteract blindness and could possibly have other uses. Cochlear implants work sort of like this for hearing.

1 upvote
BJN
By BJN (7 months ago)

" “Next year, we’ll build an 85-megapixel imager with a 120-degree field of view, more than a dozen sensors, and an F/2 lens – all in a volume roughly the size of a walnut,” Ford said."

I'd like to see the physics of how such a tiny lens can have a fast aperture explained. It sounds like there's no aperture at all as such, just direct piping of light to the sensor.

0 upvotes
AlanG
By AlanG (7 months ago)

When you say "aperture" do you mean an adjustable diaphragm?

What they mean by spherical is that it projects a spherical image rather than a flat image. If you follow the link they explain the design a bit.

Comment edited 3 times, last edit 9 minutes after posting
1 upvote
AlanG
By AlanG (7 months ago)

Now I'm not a lens designer but it seems to me that another advantage of using fiber optics to "pipe" the light is that you can use a bunch of small sensors rather than one large one.

2 upvotes
yabokkie
By yabokkie (7 months ago)

it's easier to have faster lenses for wider angles. why we don't have for SLRs is because there is a fvcking mirror box (or long flange back for some mirrorless mounts, like m4/3" and NX).

0 upvotes
forpetessake
By forpetessake (7 months ago)

If they do manage to make this system with millions fibers that would solve the curvature problem and redefine the way lenses are designed.

3 upvotes
calking
By calking (7 months ago)

And the net result will be forum heads like we have on DPR demanding that any new lens manufactured from here on be a walnut-sized lens with a .50 aperture (you know, because f2 is so "slow" and all...).

2 upvotes
Michael de Ruijter
By Michael de Ruijter (7 months ago)

I love technology!

1 upvote
Total comments: 72