The cost of Hi-Def lenses

A. J. deLange · November 27th, 2005, 12:09 PM

Bill,

Let's keep in mind that the goal of HD is to approach the quality of 35mm movie film. The movie frame has a diagonal of about 30 mm. The sensor on the XL H1 has a diagonal of about 6 mm. To produce an image of equal quality, thus, the XL H1 lens must have resolution about 5 times better than that of a 35 mm movie lens. As I mentioned in an earlier post this requires reducing aberrations to close to the diffraction limit. To do this in a 20:1 zoom lens for under $10K represents a tremendous engineering accomplishment. Twenty or 30 years ago it would be impossible to design let alone make such a lens. In those days rays were traced a handful a day. With modern computing thousands of rays can be traced per hour. Take a picture with any one of the old lenses you mentioned using fine grain film. Then cut out a 3.6 x 4.8 mm section of one of the negatives and blow it up to 24 x 18 inches (a smallish TV by today's standards) and compare the results to the XL H1 clips which have been posted here. You will find that those old lenses are not as hot as you might have thought. Are they OK for what they were designed for? Yes indeed but the resolution requirements for a lens using a 35 mm or 5 cm "sensor" are not nearly as stringent as those of a lens using a 3.6 x 4.8 mm sensor.

Nate Weaver · November 27th, 2005, 01:07 PM

A.J. speaks wisdom.

Can you tell us more about raytracing for optics design? I know what raytracing is technically, but I didn't know it was used to model lens designs before production.

Bill Anderson · November 27th, 2005, 02:06 PM

A.J I think you misunderstand my point. The reference I made about the quality of the still lenses I own(ed) including the fixed lens was to highlight the inequities in price, relative to quality, and not to compare them to today's best optics. But I stand by my zoom lens comparison to the XL series lenses. These price/quality inequities are still prevalent in today's market. Still lenses costing a fraction of "HD" lenses perform as well when attached to a video camera.

"A zoom lens for under 10K represents a tremendous accomplishment"

I couldn't agree more, but just imagine how much greater an accomplishment it must have been to create the 700$ 7megapixel point and shoot that I own, the one that knocks the socks off the 10K "HD" lens!
If you desire to compare the old with the new, I see little point in shooting a portion of film relevant to the sensor area, what you might want to try is fixing both lenses to the digital camera, and compare that way. And that is also what you might want to try with a decent contemporary still lens and an "HD" lens, and consider the price tag. The chromatic aberration on the XL2 stock lens is laughable, at any price. But it's probably no worse than other SD optics. I'm not trying to be antagonistic here, but is this the best they can do, or is it all they're willing to do, or is that our money's worth?

Chris Hurd · November 27th, 2005, 02:55 PM

Quote:

Originally Posted by Bill Anderson

Still lenses costing a fraction of "HD" lenses perform as well when attached to a video camera.

I'm sorry but no they most certainly do not. How do you record a slow creeping zoom while using a still photo lens on a video camera, where there's no zoom motor at all? How do you get a wide angle shot with a still photo lens on a video camera, when there's at least a 7x field of view magnification? How do you pull focus with a a still photo lens on a video camera, where there's no geared ring for a follow-focus rig? How do you control zoom, focus and iris remotely from the tripod pan handle when a still photo lens lacks the capability for these most basic remote lens controls? And this is just for starters.

There's a tremendous amount of difference between a still photo lens and a video lens, and most of this difference lies in the fact that a video lens is designed for the movement of the lens (focal length plus focal plane and more) while the camera *is recording.* The requirements of a video lens are different from photo, the design is different from photo and certainly the cost to manufacture is different as well. I'm more than a little surprised to find that some people expect that the price of a video lens should match the price of a photo lens... it's an apples to oranges comparison. They are not at all the same thing. The reason for the drastic disparity in pricing should be obvious to those with the most basic understanding of the differences in optical and mechanical requirements for video recording vs. still photography.

A. J. deLange · November 27th, 2005, 05:00 PM

Nate:

In ray tracing one places a point on the object side of the lens and then models a ray of light leaving that point which strikes the lens at some point on its surface. Considering the curvature of the surface at the point it strikes, the angle at which it strikes and the index of refraction on can calculate how much that ray is bent and thus the angle at which it travels inside this first element of the "lens". I put "lens" in quotes because the first surface the ray encounters is likely to be a coating, not the lens itself but the principles are the same. In any event the refracted ray is extended to the next place where the index of refraction chances such as the surface another coating, the surface of the actual lens, the rear surface of the lens (if there is no coating) etc. The process is repeated for each place where index of refaction changes up to the last coating of the last element of the lens. From that interface the ray is extended to the image plane. Where it intersects is the image of the original point in object space. The process is now repeated for other rays from the same point which strike the lens front element at different places. Were the lens perfect all extensions of rays from the same object point would strike the image plane at the same point. In fact they don't and so the designer makes adjustments to his design to correct the "aberations" to get the tightest grouping he can. It is sort of cut and try but the trying is done in the computer rather than on the optical bench thus saving untold numbers of man hours and wasted material. That this must be done for all apertures and all focal lengths of which the lens is capable AND for all colors of light in the visible spectrum and that these lenses require dozens of surfaces should help you to appreciate the magnitude of the task.

Glenn Chan · November 27th, 2005, 06:59 PM

It's probably a combination between economics and technical difficulties. If you look at the price of video equipment in general, you pay exponentially more for small increases in quality.

Mass production also makes a difference... look at the low price of 3CCD cameras due to volume. The manufacturers perhaps hold back on a few features to avoid cannibalization of their more expensive product lines. If you compare Avid to Final Cut, they are pretty close in terms of quality. Yet Avid costs a lot more because it has customers willing to pay its price.

HD lenses are also pretty new and the manufacturers are trying to recoup their R&D costs. Or they put out sub-par lenses at a low cost (i.e. low-end Fujinon, Canon zooms, the JVC lens).

A. J. deLange · November 27th, 2005, 07:23 PM

Bill,

The fact that you consider a $5K HD lens (that's my guess as to what the XL -H1 lens will be worth on the street) an inequity relative to the lens on a $1K point and shoot camera means that you do not value the things the $5K buy and that's OK. These include all the things Chris mentioned and the things I tried to explain as well. I doubt that the point and shoot has AIS or a 20:1 zoom range or coatings of the same quality (if it even has glass elements) or the aperture of the XL lenses or the build quality or the built in neutral density filters or the back focus adjustment. I also doubt very much that the sensor is 1/3". If it were they wouldn't be able to get away with leaving out the stuff that we are forced to pay $5K for. It's pretty clear to me that if Canon could get adequate performance for $500 bucks worth of plastic they would do so and perhaps charge us $1-or $2K for it putting the XL-H1 at a price point where they would sell lots more of them than they are going to at the current price point.

I'm sorry you can't grasp the point of my little gedenken experiment. I actually did it tonight using a scanned 35 mm image and the result was illustrative of the point I am trying to make: the big lens quality requirement driver (ceteris paribus) is the size of the sensor. The smaller it is the tougher the lens designer's job becomes especially if all the goodies Chris mentioned are on there as well.

BTW I do use a 105 mm Micro Nikor with the XL2 (and did so with the XL1s) as well and have certainly been pleased with the results. But this was designed for one thing: sharpness at close range. It does not zoom. It does not have AIS. It does not have a built in relay system. While it may not have the resolution of the XL-H1 lens it certainly does have enough for the XL2. I look forward to trying it on the XL-H1.

One final question and that is with respect to the "laughable chromatic aberation" of the XL2 lens. I spent a fair amount of time this evening looking at XL2 video and I can't find any. I think chromatic aberation looks like color fringing around bright objects. I do see color smear from the chroma subsampling but nothing which would suggest chromatic aberration which should surround objects. The color smears I see are always to the right of the transition to which they are attached. Does anyone else see chromatic aberration? Under what conditions?

Ray Van Eng · November 28th, 2005, 03:34 AM

Instead of a stock zoom lens that is often cited for poor quality, I wish that the manufacturers would offer a lens that only has a modest zoom range and maybe a small aperture as well so they can be made cheaply and still offers great quality for the user. Something like a 28-105mm zoom for a 35mm still camera (which many professionals realize that that is all they require for something like 85% of their work.). Anyone who wants to have an exotic wide angles zoom or extreme tele would have to pay for it.

Bill Anderson · November 28th, 2005, 10:24 AM

Chris, no one has suggested a video lens should cost the same as a
a still lens. The last time I shot motion picture film stock, HD, oh, and stills, I had no difficulty discerning between objectives.
This is not an "apples and oranges" scenario, it is an investigation
into the cost/quality ratio of optics for a given purpose, and
even taking into account the obvious unique mechanical functions of a video lens, I still see no reason to shout hallelujah when a lens that can resolve for a "HD" frame costs 15K- nor when lenses touted as the next coming can barely resolve for a 700 or so pixel dimension without showing chromatic aberrations unworthy of the humblest still lens.

November 27th, 2005, 01:07 PM	#17
Nate Weaver Wrangler Join Date: Oct 2002 Location: Los Angeles, CA Posts: 2,100	A.J. speaks wisdom. Can you tell us more about raytracing for optics design? I know what raytracing is technically, but I didn't know it was used to model lens designs before production. __________________ My Work: nateweaver.net

November 27th, 2005, 12:09 PM	#16
A. J. deLange Major Player Join Date: Jun 2004 Location: McLean, VA United States Posts: 749	Bill, Let's keep in mind that the goal of HD is to approach the quality of 35mm movie film. The movie frame has a diagonal of about 30 mm. The sensor on the XL H1 has a diagonal of about 6 mm. To produce an image of equal quality, thus, the XL H1 lens must have resolution about 5 times better than that of a 35 mm movie lens. As I mentioned in an earlier post this requires reducing aberrations to close to the diffraction limit. To do this in a 20:1 zoom lens for under $10K represents a tremendous engineering accomplishment. Twenty or 30 years ago it would be impossible to design let alone make such a lens. In those days rays were traced a handful a day. With modern computing thousands of rays can be traced per hour. Take a picture with any one of the old lenses you mentioned using fine grain film. Then cut out a 3.6 x 4.8 mm section of one of the negatives and blow it up to 24 x 18 inches (a smallish TV by today's standards) and compare the results to the XL H1 clips which have been posted here. You will find that those old lenses are not as hot as you might have thought. Are they OK for what they were designed for? Yes indeed but the resolution requirements for a lens using a 35 mm or 5 cm "sensor" are not nearly as stringent as those of a lens using a 3.6 x 4.8 mm sensor.

November 27th, 2005, 02:06 PM	#18
Bill Anderson Regular Crew Join Date: Jul 2004 Location: Ottawa. Canada Posts: 163	A.J I think you misunderstand my point. The reference I made about the quality of the still lenses I own(ed) including the fixed lens was to highlight the inequities in price, relative to quality, and not to compare them to today's best optics. But I stand by my zoom lens comparison to the XL series lenses. These price/quality inequities are still prevalent in today's market. Still lenses costing a fraction of "HD" lenses perform as well when attached to a video camera. "A zoom lens for under 10K represents a tremendous accomplishment" I couldn't agree more, but just imagine how much greater an accomplishment it must have been to create the 700$ 7megapixel point and shoot that I own, the one that knocks the socks off the 10K "HD" lens! If you desire to compare the old with the new, I see little point in shooting a portion of film relevant to the sensor area, what you might want to try is fixing both lenses to the digital camera, and compare that way. And that is also what you might want to try with a decent contemporary still lens and an "HD" lens, and consider the price tag. The chromatic aberration on the XL2 stock lens is laughable, at any price. But it's probably no worse than other SD optics. I'm not trying to be antagonistic here, but is this the best they can do, or is it all they're willing to do, or is that our money's worth?

November 27th, 2005, 05:00 PM	#20
A. J. deLange Major Player Join Date: Jun 2004 Location: McLean, VA United States Posts: 749	Nate: In ray tracing one places a point on the object side of the lens and then models a ray of light leaving that point which strikes the lens at some point on its surface. Considering the curvature of the surface at the point it strikes, the angle at which it strikes and the index of refraction on can calculate how much that ray is bent and thus the angle at which it travels inside this first element of the "lens". I put "lens" in quotes because the first surface the ray encounters is likely to be a coating, not the lens itself but the principles are the same. In any event the refracted ray is extended to the next place where the index of refraction chances such as the surface another coating, the surface of the actual lens, the rear surface of the lens (if there is no coating) etc. The process is repeated for each place where index of refaction changes up to the last coating of the last element of the lens. From that interface the ray is extended to the image plane. Where it intersects is the image of the original point in object space. The process is now repeated for other rays from the same point which strike the lens front element at different places. Were the lens perfect all extensions of rays from the same object point would strike the image plane at the same point. In fact they don't and so the designer makes adjustments to his design to correct the "aberations" to get the tightest grouping he can. It is sort of cut and try but the trying is done in the computer rather than on the optical bench thus saving untold numbers of man hours and wasted material. That this must be done for all apertures and all focal lengths of which the lens is capable AND for all colors of light in the visible spectrum and that these lenses require dozens of surfaces should help you to appreciate the magnitude of the task.

November 27th, 2005, 06:59 PM	#21
Glenn Chan Inner Circle Join Date: Jun 2003 Location: Toronto, Canada Posts: 4,750	It's probably a combination between economics and technical difficulties. If you look at the price of video equipment in general, you pay exponentially more for small increases in quality. Mass production also makes a difference... look at the low price of 3CCD cameras due to volume. The manufacturers perhaps hold back on a few features to avoid cannibalization of their more expensive product lines. If you compare Avid to Final Cut, they are pretty close in terms of quality. Yet Avid costs a lot more because it has customers willing to pay its price. HD lenses are also pretty new and the manufacturers are trying to recoup their R&D costs. Or they put out sub-par lenses at a low cost (i.e. low-end Fujinon, Canon zooms, the JVC lens).

November 27th, 2005, 07:23 PM	#22
A. J. deLange Major Player Join Date: Jun 2004 Location: McLean, VA United States Posts: 749	Bill, The fact that you consider a $5K HD lens (that's my guess as to what the XL -H1 lens will be worth on the street) an inequity relative to the lens on a $1K point and shoot camera means that you do not value the things the $5K buy and that's OK. These include all the things Chris mentioned and the things I tried to explain as well. I doubt that the point and shoot has AIS or a 20:1 zoom range or coatings of the same quality (if it even has glass elements) or the aperture of the XL lenses or the build quality or the built in neutral density filters or the back focus adjustment. I also doubt very much that the sensor is 1/3". If it were they wouldn't be able to get away with leaving out the stuff that we are forced to pay $5K for. It's pretty clear to me that if Canon could get adequate performance for $500 bucks worth of plastic they would do so and perhaps charge us $1-or $2K for it putting the XL-H1 at a price point where they would sell lots more of them than they are going to at the current price point. I'm sorry you can't grasp the point of my little gedenken experiment. I actually did it tonight using a scanned 35 mm image and the result was illustrative of the point I am trying to make: the big lens quality requirement driver (ceteris paribus) is the size of the sensor. The smaller it is the tougher the lens designer's job becomes especially if all the goodies Chris mentioned are on there as well. BTW I do use a 105 mm Micro Nikor with the XL2 (and did so with the XL1s) as well and have certainly been pleased with the results. But this was designed for one thing: sharpness at close range. It does not zoom. It does not have AIS. It does not have a built in relay system. While it may not have the resolution of the XL-H1 lens it certainly does have enough for the XL2. I look forward to trying it on the XL-H1. One final question and that is with respect to the "laughable chromatic aberation" of the XL2 lens. I spent a fair amount of time this evening looking at XL2 video and I can't find any. I think chromatic aberation looks like color fringing around bright objects. I do see color smear from the chroma subsampling but nothing which would suggest chromatic aberration which should surround objects. The color smears I see are always to the right of the transition to which they are attached. Does anyone else see chromatic aberration? Under what conditions?

November 28th, 2005, 03:34 AM	#23
Ray Van Eng Regular Crew Join Date: Mar 2004 Location: Vancouver, BC, Canada Posts: 25	Instead of a stock zoom lens that is often cited for poor quality, I wish that the manufacturers would offer a lens that only has a modest zoom range and maybe a small aperture as well so they can be made cheaply and still offers great quality for the user. Something like a 28-105mm zoom for a 35mm still camera (which many professionals realize that that is all they require for something like 85% of their work.). Anyone who wants to have an exotic wide angles zoom or extreme tele would have to pay for it.

November 28th, 2005, 10:24 AM	#24
Bill Anderson Regular Crew Join Date: Jul 2004 Location: Ottawa. Canada Posts: 163	Chris, no one has suggested a video lens should cost the same as a a still lens. The last time I shot motion picture film stock, HD, oh, and stills, I had no difficulty discerning between objectives. This is not an "apples and oranges" scenario, it is an investigation into the cost/quality ratio of optics for a given purpose, and even taking into account the obvious unique mechanical functions of a video lens, I still see no reason to shout hallelujah when a lens that can resolve for a "HD" frame costs 15K- nor when lenses touted as the next coming can barely resolve for a 700 or so pixel dimension without showing chromatic aberrations unworthy of the humblest still lens.