Posted by Larry Fischer on March 30, 2010 at 5:06 PM
Why are camera lenses round?
I presume the early "figuring", done by grinding, led to round shapes.
I remember in my younger years a rifle scope made by Redfield which featured a rectangular lens.
My questions:
1) Would not this fit the sensor better?
2) Figuring problems could be outdone by computer grinding.
3) Are there mathematical and physical limitations to try this?
4) Other issues?
Thinking outloud.
the fisch
Probably because it is cheaper to make a round lens, let it cast a round image then let the sensor/film do the "cropping" than it is to make the lens, then cut off the parts that aren't' really necessary. Your scope's rectangular lens was probably a spherically ground lens cut down to a rectangle.
This is not at all my area of expertise, so maybe someone who really knows will respond.
Pete D.
Posted by Dan Borzynski on March 30, 2010 at 6:21 PM
Pete's answer makes the most sense.
Pre-computerized grinding and polishing machines are far easier to build with a circular lens - you can spin the lens so no one spot is favored over another. Cutting the afterwards adds another step and another chance to damage the lens.
There may be other mechanical considerations. Focusing is normally done by rotating the lens elements on a helix. The aperture diaphragm is generally a circular mechanism.
I wonder why lens hoods are round. I've seen rectangular hoods on large video cameras. Would rectangular hoods be more effective than petal style hoods?
I think it would also be very difficult to construct a zoom lens square.
I also imagine that there would be all kinds of problems with the way a square lens would defract the light. How on earth would they get the light to bend correctly?
Posted by Jim Bullard on March 31, 2010 at 9:22 AM
Everyone has pretty much covered the practical reasons to which I'll add that there is no real advantage to a rectangular lens since image formation is circular in nature. The reason some lens shades are rectangular or 'flower shaped' is that they protect against flare on portions of the round lens which the film/sensor doesn't see anyway. To make a rectangular lens you'd either have to cut the top, bottom and sides off a round one or figure out how to accurately grind a lens without those parts. Either way is more difficult and some lenses focus by having an element or elements rotate within a screw type collar that moves them in and out. You couldn't do that with a square or rectangular lens. And then there is the problem of creating a square or rectangular diaphragm.
Many years ago (in Internet terms) I had a website that dealt with B&W film photography. It is now defunct but some portions of it remain on the servers that were owned by the ISP I had at the time and this "sharelesson" http://www.northnet.org/jimbullard/CoC.htm I wrote on the circle of confusion might help understand image formation . Only one of the links to other pages at the bottom works. I'm not sure why they removed only parts of the site from their server which now belongs to another ISP. I've tried to get the new owner of the server to remove the rest but they ignore my requests. One of these days I should put up a new site with updates of the rather crude illustrations.
Posted by Larry Fischer on March 31, 2010 at 10:07 AM
All good answers and none of which I would argue. Jim B., I have visited your site about the "circle of confusion". Good stuff.
The focusing issue seems to be the most applicable. A rifle scope is basically fixed.
I think moden lens manufacturing with computer generated math could make a rectangular lens but I don't know if the result could be made better.
The response earlier about damaging the lens product by cutting a spherical lens to rectangular makes sense.
No doubt if a lens could be manufactured with equal or better application it would have been done. There is no price limits to cinematography lenses and that technology would have dribbled down to a small audience in other photo realms, albeit at a great price.
I was just thinking out loud and you all have brought out several points that make sense.
I don't have the link, but there was a blurb on DPreview last week about a company that is developing a round image sensor.
Posted by David Bostedo on April 4, 2010 at 12:28 AM
As far as the round sensor goes, I believe that was just an April Fools joke It was supposedly made by "Rokton" which doesn't exist anywhere on the web except for stories about the round sensor. And DPReview pulled the "press release" down.
Anyway, round sensors sufffer from the same problem as square lenses - they would be inherently more expensive. Space on wafers where chips are made is pricey, and since round shapes don't fit together edge to edge, there would be a lot of wasted wafer real estate - which would mean much more expensive sensors.
I've always thought a square sensor would be great. We go through a lot of trouble to shoot vertical or horizontal, but in this digital age, it is so easy to crop before final output that I think it would be cheaper to just shoot square and crop later.
Discussion
9 repliesI presume the early "figuring", done by grinding, led to round shapes.
I remember in my younger years a rifle scope made by Redfield which featured a rectangular lens.
My questions:
1) Would not this fit the sensor better?
2) Figuring problems could be outdone by computer grinding.
3) Are there mathematical and physical limitations to try this?
4) Other issues?
Thinking outloud.
the fisch
This is not at all my area of expertise, so maybe someone who really knows will respond.
Pete D.
Pre-computerized grinding and polishing machines are far easier to build with a circular lens - you can spin the lens so no one spot is favored over another. Cutting the afterwards adds another step and another chance to damage the lens.
I wonder why lens hoods are round. I've seen rectangular hoods on large video cameras. Would rectangular hoods be more effective than petal style hoods?
I also imagine that there would be all kinds of problems with the way a square lens would defract the light. How on earth would they get the light to bend correctly?
Many years ago (in Internet terms) I had a website that dealt with B&W film photography. It is now defunct but some portions of it remain on the servers that were owned by the ISP I had at the time and this "sharelesson" http://www.northnet.org/jimbullard/CoC.htm I wrote on the circle of confusion might help understand image formation . Only one of the links to other pages at the bottom works. I'm not sure why they removed only parts of the site from their server which now belongs to another ISP. I've tried to get the new owner of the server to remove the rest but they ignore my requests. One of these days I should put up a new site with updates of the rather crude illustrations.
The focusing issue seems to be the most applicable. A rifle scope is basically fixed.
I think moden lens manufacturing with computer generated math could make a rectangular lens but I don't know if the result could be made better.
The response earlier about damaging the lens product by cutting a spherical lens to rectangular makes sense.
No doubt if a lens could be manufactured with equal or better application it would have been done. There is no price limits to cinematography lenses and that technology would have dribbled down to a small audience in other photo realms, albeit at a great price.
I was just thinking out loud and you all have brought out several points that make sense.
Thanks.
the fisch
Anyway, round sensors sufffer from the same problem as square lenses - they would be inherently more expensive. Space on wafers where chips are made is pricey, and since round shapes don't fit together edge to edge, there would be a lot of wasted wafer real estate - which would mean much more expensive sensors.
I've always thought a square sensor would be great. We go through a lot of trouble to shoot vertical or horizontal, but in this digital age, it is so easy to crop before final output that I think it would be cheaper to just shoot square and crop later.
Pete D.