I know it's possible (JVC's doing it in their 16mm adapter), but I have no idea how it would work. Any one know? I'm not about to try and make a purely optical adapter, I'm just very curious how the concept works--since it seems impossible.

I was spending a few months trying to create an optical design to increase the relative apperture and thus decrease Fstop rating. A few lenses in the front of the camcorder optics to increase the size of the opening pupil in comparison to the focal length (1/Fstop = opening pupil/focal length).
In my point of view, the optical adapters must be some kind of variance of that in order to get shallow DOF.

On the other hand, to build an optical adapter with deep DOF and 16mm lens field of view is quite easy.

JVC`s adaptor is basically a static DOF adaptor that doesn`t even flip the image. Sure it might have a relay without iris and less light loss.

T

But does it use some intermediary screen?
If I understood Matthew's question correctly, he was asking if a direct optical solution without any projecting element is possible.

It is my understanding (maybe my misunderstanding) that the JVC is purely optical--no focusing screen at all. It allows for 16mm PL mount lenses, retains their depth of focus and optical characteristics, loses next to no light, and is purely optical from what I understand.

Of course, it also only works with 16mm, not 35mm, doesn't flip, and is over $4,000. But I can't figure out how it works. Sure you can just put the lens on without a focusing screen, but then you don't get the shallow depth of focus or even the ability to focus.

Well, with a complicated lens assembly, it could "convert"

Let's say you got a 16mm PL mount lens: 25mm F2.0

Converted to 1/3" this should be roughly 14mm. If you retain your field of view, your maximum aperture opening will be approximately twice as high (F1.0). This is because opening pupil stays equal, effective focal length halves.
Then you would indeed have a far more shallow DOF.

Does it optically work? I don't know. I have tried exactly that with no success. But I am very willing to share my gathered knowledge with anyone who wants to continue trying, as long as it stays public domain.

I believe the new JVC's flip the image "in-cam" so any lens adaptor can be used. They are trying to target the indie crowd and this is one of their new features.

I may be confused by the initial question, but here's what it sounds like:

Does the JVC 16mm Lens adapter give the same FOV, DOF and T-stop as a 16mm lens on a 16mm film camera?

The answer is no. It gives the same FOV, DOF and T-stop as that particular lens...on a 1/3" camera. When people talk about "converting" a lens from 1/3" to 2/3" or 35mm (academy), they're talking about equivalent FOV. A 25mm lens is a 25mm lens, period. It may look very telephoto on a 1/3" camera, but it will look wide angle on a 35mm film camera. BUT, the focal length does not change! The Field of View (FOV) does change, drastically.

The 16mm adapter for the HD-100 is a mechanical adapter to make the mount type as well as the flange-focal distance compatible between 16mm lenses and the HD-100 body. The reason for this is to expand the number of possible lenses you could use with the system, especially to include extremely fast film-style primes. I won't go into the difference between film-style lenses and ENG-style lenses, there's a lot of info about that.

In terms of DOF, the quest to get 35mm-like DOF on 1/3" cameras requires understanding of what factors affect DOF. Image size (1/3", 35mm, s16, etc), focal length, aperture, and focal distance from lens to subject. Now,with a 50mm lens at T/4.0 18" from the subject on 35mm, the DOF will be pretty damn shallow. Take that same lens, and change nothing except the camera...to the HD-100 (1/3" chips). All of a sudden, the image appears to be "zoomed WAY in." If you take a still from the 1/3" camera, and a frame from the 35mm (cropped to the same composition as the still frame from the 1/3") you'll have images which look very similar. They will have the exact same DOF. So what's the problem? On a 1/3" chip, the 50mm lens has a very very narrow FOV. So the natural solution is to switch to a wider lens. But switching to a wider lens also changes one of the factors of DOF.

So the long answer is no, there is no straight "optical" adapter that magically turns a 1/3" chip into a 35mm sized chip.*****

The Mini35, M2, Letus, etc... all use the same principal to achieve the shallow DOF look. They all present the camera with an image resolved on a ground glass that's approximately the size of a 35mm frame. If you photograph that image on the ground glass, and compare it to a frame from a 35mm camera using the same lens, same T-stop, and same subject distance, you will have essentially the same FOV AND DOF. That's the trick - focusing the lens on "film" that's the same size as 35mm film, and then shooting that image. The reason that the HD-100 has a flip setting is to make it work easier with such adapters, or lenses whose nodal point is different than the ENG-style lenses.


*****Well, kinda. Theoretically, you could make a lens with a T-stop SO small that it has the same DOF as a 35mm lens with equivalent FOV. But current technology ain't there yet, and may never be.

So the long answer is no, there is no straight "optical" adapter that magically turns a 1/3" chip into a 35mm sized chip.*****



I think the definitive Dof Adaptor may use a fiber optic taper attached to the sensor. Have a look...:

http://www.proxitronic.de/prod/fo/ekoppeln.htm
http://www.proxitronic.de/
http://home.comcast.net/~mwaltuck/Tapermag/index.html
http://www.us.schott.com/fiberoptics/english/products/healthcare/imagingfiberoptics/fusedcomponents/tapers.html
http://www.grand-illusions.com/acatalog/info_58.html
http://www.plastecs.com/fiber%20optics.htm
http://www.o-eland.com/Instrumentation/faceplate.htm

In the last link you can find inverters for image flipping if necessary.

Carlos,
That's just what I was thinking. It would explain why the adapter only handles 16mm lenses, as a 35mm sensor-sized to 1/3" taper might be too bulky to integrate or difficult to manufacture. However from what I've read of these things they don't go much above 70% light efficiency. That's worse than typical DOF adapters on the market today...

Essentially you're replacing the relay lens with a fiber optic pipe. Same principal as mini35, m2, etc. The image is still formed on a plane in front of the sensor, and photographed through a lens assembly to the sensor. 6 of one, half a dozen of another. Two different approaches to the same solution. Technically, they're both "straight optical solutions." True, this version has no "GG" but the ends of fiber optics ARE GG's.

its na funny thing but I had suggested using fibre optic tapers and image inverters some six months ago but a number of people said that they were not suitable because there would be to much light loss, but it seems that fibre optrics have already been coupled with ccd's to deal with things such as low light image inversion, reduction and enlargment, for a while now.


told ya so
rob ( not so regular poster but scans dvi daily , this is a great community )

I replied to all this too soon, 70% light loss , sheesh . maybe I didn't told ya so..... teehee
rob

I replied to all this too soon, 70% light loss , sheesh . maybe I didn't told ya so..... teehee
rob
Robert, you´re wrong, it´s 70% light efficiency, no 70% light loss.
When you use an adapter with GG, the light goes throught the GG, the macro lenses, and the camera lenses (some lenses in the complete objetive). Each lens could have a 2-5% light loss. If you use the Fiber Optic system, you don´t need all these elements. I think the image could be even brighter, and there are no problems with image distortion, chromatic aberrations, vigneting, grain, motor noise... I think this is the cleanest solution for a 35mm adapter.

Jaron, I think the image is projected directly over the sensor, not over a GG, but I can be wrong.

We are already seeing adapters for less than $1000 that see negligable resolution loss, have very little light loss, and are improving a great deal with respect to ease of use and reliability. Even given those attributes, we are continuing with product developement to further refine our own implementation.

It would seem RED, which will allow the use of full format lenses with their inherent DOF properties, would be the logical next step up with respect to total camera/adapter cost.

True carlos, I mis-phrased. This seems like it would eventually be a FAR better solution than all the parts of a mini35. My point was simply that the image is formed by the primary lens on the front surface of the fiber optic. The individual optics themselves act like lenses to redirect the image to the chip. So instead of a single ground glass with a single lens to focus it on the chip, you have thousands of gg's and lenses. Either way, you're photographing an image formed in front of the chip.

We are already seeing adapters for less than $1000 that see negligable resolution loss, have very little light loss, and are improving a great deal with respect to ease of use and reliability. Even given those attributes, we are continuing with product developement to further refine our own implementation.

It would seem RED, which will allow the use of full format lenses with their inherent DOF properties, would be the logical next step up with respect to total camera/adapter cost.

I agree. RED has set a new price point that adapter people have to watch out for. MovieTube and adapters in it's price region are pointless investments when a full 35mm sensor camera system is only a grand or two away. Its development in that price range has made expensive adapter solutions like fiber optics moot.

I know it's possible (JVC's doing it in their 16mm adapter), but I have no idea how it would work. Any one know? I'm not about to try and make a purely optical adapter, I'm just very curious how the concept works--since it seems impossible.

It turns out it's possible to make a purely optical (no scattering intermediate) DOF adapter. In fact, the Frazier lens does exactly that - though it gets the opposite effect (deep depth of field), since that's what it was going for (it's used in a 35mm regime, where shallow DOF is fairly easy to get).

Essentially, an all-optical adapter would just relay the aerial image directly to the CCD/CMOS chip. To put it another way: imagine the focal plane of your prime lens -- a lens that exactly relays that image to your imaging device (film or CCD/CMOS) will give you precisely the lens properties of that original lens at the original focal plane.

Well, not quite -- the relay system will necessarily add its own aberrations, which in practice is why this hasn't been done yet -- in order to relay the entire image with any appreciable amount of light, you've got to use some pretty extreme optics that will necessarily add lots of aberrations (at one point, I decided that going from IMAX to a single 1/3" chip would require a relay system of about F/1.4, which I just can't build at home). In fact, that's why the Frazier lens (I suspect) only operates at T/8 or slower - it's the only reasonable way to have a lens system that complicated that doesn't make the image completely fuzzy (due to aberrations). (Keep in mind the Frazier lens is manufactured by Panavision, who know a thing or two about lenses, and they still keep it at T/8.)

This gets considerably more complicated when you're using 3-chip cameras, too. Because the beam-splitting prism introduces aberrations (lots of astigmatism, for one) in any non-perpendicular ray bundle, you've got to make your relay lens design telecentric in image space (and actually entocentric in object space -- check out wikipedia and edmund optics website). Not trivial, and certainly beyond me.

Ultimately, a truly optical solution would be so gnarly that you're probably better off going with a RED camera or similar, even if the DOF adapter had large economies of scale (which most of these solutions do not).

Next post: fiber optic tapers.

Next post: fiber optic tapers.

First, the good news: not only is it possible, it's widespread. Tapers cost as little as $500 (or even less), there are lots of different manufacturers, and it's a reasonably mature technology. In fact, many night-vision devices use image tapers.

And that's the other good news: the reason it's used in night-vision is that a fiber optic taper actually gives you a light *gain*. Yes, the cladding material in the taper absorbs about 30% of the incident light. However, the total flux at the entrance face (less those losses) is transmitted to the exit face -- which is usually glued to a sensor. When you squeeze a given amount of light into a smaller area, the intensity increases as the reciprocal of the ratio of the areas -- so a 3x taper will actually give you 530% light increase (1:9 area ratio, x 70% transmittance). You could gain a few stops, and have the DOF of a larger sensor, to boot. (Incidentally, this is why larger chips tend to have better low-light performance: a single pixel measures total incident light; with a larger area receiving the same intensity -- and therefore greater flux -- the net effect is more total photons and a higher effective ASA. A similar analogy holds for fast, large-grain films versus slower, less-grainy film.)

Okay, so why isn't it being done in D-cinema? There are a few problems with this pie-in-the-sky calculation. For one, the taper itself only accepts a certain amount of light. Not intensity, but angle: each taper has an inherent 'acceptance cone' outside which light will simply be absorbed by the cladding (this number is independent of the 70% figure). Effectively, each cone has a working f/number, and faster lenses will not see any increase in light (nor apparent DOF: you'll be locked into the maximum acceptance F/# of the taper -- it's almost as if the individual fibers in the taper are functioning as their own apertures). Furthermore, the steeper the cone (larger image plane), the smaller that acceptance cone. There's really no free lunch, any light gain you think you might get results in a smaller angle to pour light onto your sensor.

As for image-inverters, forget it. They have a so-called 'Mae West' configuration (big at both ends, tiny in the middle) that accentuates this problem. Besides, if you're gluing this thing to a sensor, you don't *need* to flip the image -- it will be flipped once by the taking lens, and once by your camera's circuitry.

It gets worse: the acceptance cones are all (necessarily) perpendicular to the taper surface (in a flat cone, that is). So while fibers in the center of the image may get a fair amount of light (if the working F/# of the taper is lower than the F/# of the lens), the fibers on the outside of the image probably won't - their acceptance cones may miss the lens completely. You can (sort of) get around this with a field lens, but now you're introducing aberrations. Note that this isn't actually a problem if your taking lens is telecentric in image space -- like any lens designed for 3-chip cameras must be (so you could conceivably use a taper on a small chip video camera and mount a lens for a larger chip video camera, and enjoy a modest light gain). But it's very problematic for anyone trying to use cinema or still lenses on their video camera (in fact, I believe that making still lenses telecentric is exactly what the Canon EF adapter for the XL series does -- though it does nothing to change effective sensor size).

Add to this the mechanical problems of working with a taper: In order to get the benefits of the taper, you have to physically glue it directly to the sensor. Any dust or bubbles, and you're stuck with a permanent image defect. This also means you can't use tapers with 3-chip cameras, since the CCDs are all hidden behind the beam-splitting prism. Also, most commercially available tapers have as much as 2-3% distortion, which may or may not be acceptable to you.

That's not to say using a fiber optic taper isn't possible, or even undesirable -- it's just not a magic bullet. I am tempted to get a single-chip HDV camera, hack off the lens, glue a taper onto it and build a lens mount (along with field lens) for using still camera lenses. It would look neat, I think, and be a handy tool to have. But it won't replace other solutions, and besides, that money is probably better spent on the RED camera (I do have a deposit down, and I don't think there are going to be better solutions for less $).

Sorry for the long posts, but that's my two cents. If you want clarification on anything I posted, just drop me a line at first underscore last at yahoo dot com, though it may take me a while to reply as that's not my regular e-mail address (it's a handy pre-filter for spam). Or, just reply to this post.


Cheers,


Ryan

PS - I'm not a lens designer, optical engineer, or anything of the like. Please take these posts with a grain of salt, I'm just hoping to spare people a little time. If this stuff interests you, I highly recommend starting with Eugene Hecht's Optics, which I believe is in its 4th edition now. It's not cheap (about $100 on Amazon), but a lot of libraries have a copy (I cut my teeth on my public library's copy, a 3rd edition).

Does anyone here remember optical printing in the film title houses. I saw one such machine and it captured a projected image in a large condensor. The condensor I believe was from an enlarger. This in turn was filmed frame by frame by another camera. They used selsyn motors that kept all in frame synch.

Ryan, this info is really interesting. Thanks so much. No need to apologize for the long posts; I really appreciate all the great info.

It makes me wonder how the JVC system is going to work if it indeed functions wide open, however. I guess we'll see. I didn't realize the Hylen system was purely optical (I should have assumed as much but the "cartridges" always remined me of ground glass), but f8 seems like a tough restriction, since you already get a relatively deep depth of focus at that f stop and f8 requires a lot of light. The fiber optic tapers seem like a cool experiment, but I have no idea how it's possible for their minimum effective f-stop to not only alter the transmission of light from the taking lens, but also the effective depth of focus. Insane.

Still, 2.8 is as fast as the dvx at full telephoto, so I bet there's something to be done there...not by me, though. Oh well. Makes me wish I'd listened to my mom and taken physics classes in college....

So I probably should've looked at the JVC adapter more closely before posting.

Jaron Berman suggests that it's just a mount adapter - letting you use pro lenses, but cropping the 35mm- or 16mm-sized image circle to the 1/3" of the JVC's chips.

Pictures of the adapter suggest it's actually a purely optical adapter, like you were wondering. The hz-ca13u looks far bigger than a mount adapter should (or could) be; instead, it's probably a system akin to the Frazier lens I mentioned earlier.

If it's like the Frazier system, it would work like this: the taking lens (35mm or 16mm PL) projects an image onto an image plane. Unlike ground glass adapters, that image plane is just air -- the ray bundles converge at the normal back focal distance, and then diverge beyond it. That divergent light is then collected by the hz-ca13u and refocused onto the JVC's chips.

Or, to think of it in reverse, the adapter images the JVC's chips at the image plane of the PL mount lens. Same effect; the image ends up on tape with the optical properties of the original taking lens (plus any aberrations of the adapter, which are probably pretty minimal).

This may sound impossible - in fact, if this works, why would you need an intermediate scattering plate (the ground glass) in 35mm adapters at all?

The rough answer is, you don't. If you could photograph the spot where the ground glass normally is without ground glass, you'd see the precise image, without any grain or haziness. This image is called the 'aerial image.' However, this would only work on-axis; as you move off-axis, the ray bundles diverge and -- close enough to the edge -- actually miss the second lens entirely, resulting in extreme vignetting. The ground glass scatters the light, so that some of it ends up going towards the second lens. (In fact, for all those people with condenser lenses who have minimal light loss, it would be fun to try focusing the system then *completely removing* the ground glass - if the system is set up properly, it would actually approximate what I imagine the JVC adapter is doing - though homemade rigs might still show some vignetting.)

Adapters like the Frazier lens use a field lens to redirect these ray bundles towards your second lens (exactly the function of the condenser lens 'sandwich' that some 35mm adapters call for). Because the field lens is very close to the image plane of the taking lens (sometimes, the image plane is actually inside the field lens itself, though this puts any lens defects in the field lens in focus in the final image), the field lens doesn't actually add much convergence to the ray bundles -- it actually acts more like a variable prism that redirects edge rays into the lens of the second camera (commonly called a 'relay lens,' it relays the original image onto your final image plane).

By the way, some lens tests actually use a microscope to examine the aerial image as though it were a real object -- that way you can directly see the effects of the lens (basically you focus the microscope as though it were looking at the image of the taking lens). Same principle as above -- and you definitely see the image of the taking lens, that's the whole point.

So... I guess the short answer is yes, it's absolutely possible. Since JVC has real optical engineers, I'd believe that it actually works as advertised. The relay adapter may add its own aberrations (as I said, it's not an easy process, particularly for using with 3-chip cameras), but since they're professionals, the result is probably, well, professional looking.

And if it works as advertised, I'd personally see no reason to buy another GG adapter again (sorry, P+S Technik).

Note: it wouldn't work for both 35mm and 16mm -- if it is indeed a purely optical relay lens, it probably crops 35mm lenses to a 16mm size (then relays them to the 1/3" chips).

As a personal note, I was at CineQuest 2005 in San Jose and I heard reps from Panavision, JVC, Lockheed Martin (they make super-high-end optics for the military) and others talk about new developments...

Now, I didn't know that much about optics back then, but I asked the Panavision guy some questions about both the P+S Technik adapter and the Foveon (stacked CMOS) chip. Even at the time, I felt he dodged the questions.

Basically, I asked him if there was any interest in optically enlarging the chips to get the image characteristics of a large-chip camera with the manufacturing economies of small-chip cameras (the Dalsa rep had *boasted* that they throw out 92% of the chips they manufacture). The Panavision rep rather curtly replied that solutions like the P+S Technik would never have acceptable quality.

At the time, I assumed you needed a scattering plate, but of course that rep knew otherwise: Panavision's Frazier system does exactly what we're describing here, relaying an aerial image onto a physically distinct film plane.

Now, I'm not a big fan of the indie-inferiority complex (I've got it bad sometimes, I'll admit), but I can't help but think there's a marked disrespect at some companies for the low-ticket end of the market, particularly for D-cinema. I think that's a mistake, but hey -- maybe JVC has scooped Panavision on this one by getting to the lower-cost, higher volume segment of the market.


Okay... I think I've gone way beyond my two cents.

Matthew - thanks for the mention of the Hylen system, it was completely off my radar.

By the by, that Hylen system isn't just an optical defocus - it looks like they do the equivalent of selective scattering for the defocus (almost like a ground glass with a small segment that's not ground).

I'd guess (based on the still images at http://www.panavision.com.au/News/HylenScreenShots/HylenScreen15.htm) that it has nothing to do with defocus -- the bokeh suggests that the blurriness is actually the product of two convolutions: a first (true defocus) convolution with the taking lens's bokeh, and a second convolution caused by scattering at an intermediate. In fact, based on the chromatic aberration of the secondary bokeh, I'd say it's actually done with diffractive scattering (based on the reddish tinge to the blur circles).

The relatively sharp-edged delineation between in-focus areas and 'blurred' areas suggests this intermediate is either close to the final image plane (which is difficult) or at/near an aerial image. (In which case, it's exactly analogous to the Frazier lens and possibly the JVC adapter.) That's also how they could superimpose a graphic: by placing it at the aerial image plane.

There's a whole category of (somewhat arcane) lenses that allow you to insert objects at the aerial image -- called, unsurprisingly, aerial lenses. I think this would have been handy for process shots or extreme close-ups with a deep depth of field; you don't hear much about them these days. (On a side note, it was suggested that Frazier used shots taken with an aerial lens in his application for a patent on his eponymous lens -- one of the considerations the court used in overturning the patent, by the way, which is an interesting read despite being invalidated).


Just another view from the outside.

Oh, and thanks for the thread, Matthew -- I've lurked around these forums for a while before finding a topic that got me going.


Cheers,


Ryan

It makes me wonder how the JVC system is going to work if it indeed functions wide open, however.

Well, there's nothing preventing you from operating a relay at low f/#s except aberrations, so if JVC can engineer a lens without aberrations, there's no reason why not (my initial skepticism results from trying it in my garage, and I doubted home-brewed setups would have the ability to diagnose and correct the aberrations... though I'm starting to come around).



The fiber optic tapers seem like a cool experiment, but I have no idea how it's possible for their minimum effective f-stop to not only alter the transmission of light from the taking lens, but also the effective depth of focus. Insane.

It's not that the taper alters the light from the taking lens -- it only accepts some of the light (the remainder veers off into the cladding of each invidual fiber, to be absorbed). The reason it changes the depth of field is that the physical front surface of the taper is larger than the original chip: http://www.edmundoptics.com/onlinecatalog/displayproduct.cfm?productID=1599&search=1

And they actually have a minimum acceptance cone, usually expressed as numerical aperture but mathematically equivalent to f/# (used in this way, they're two ways of talking about the same thing). Basically, you can imagine each individual fiber projecting a cone directly out of the face of the taper; this cone is the sum of all possible directions that would lead to the CCD on the other side. Light originating outside of this cone isn't totally internally reflected by the side walls of the individual fiber optic, instead veering off into the cladding.

There's more information in this .pdf (I'll try to attach it - no promises). Not sure where I got it originally, probably from a fiber manufacturer.


And don't worry about the physics. This stuff is so specific, I doubt it's covered in most general physics courses. (I was an English major, anyway.)

It would appear Panavision's Hylen lens is covered by this U.S. patent: patent #: 5,649,259

(http://patft1.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.htm&r=1&f=G&l=50&s1=5649259.PN.&OS=PN/5649259&RS=PN/5649259)

Unfortunately, I can't get the images to load up in my browser. Maybe I'll order a physical copy of the patent, if I'm feeling ambitious.

And not that they're required to follow the patent to the letter, but based on the abstract it would seem I'd be wrong about the diffractive scattering. I dunno - I still trust my eyes.


Cheers,


Ryan

Does the JVC 16mm Lens adapter give the same FOV, DOF and T-stop as a 16mm lens on a 16mm film camera?

The answer is no. It gives the same FOV, DOF and T-stop as that particular lens...on a 1/3" camera.

I just noticed this... is this JVC adapter nothing more than their version of Canon's EF Adapter? Because that thing is a borderline useless POS.

I just noticed this... is this JVC adapter nothing more than their version of Canon's EF Adapter? Because that thing is a borderline useless POS.

No, it's a pretty complex optical system that does what a mini35 does without ground glass. At least, that's what seems to be the case.

Ryan, I thought you might find this interesting: http://www.smsprod.com/products/lenses/angenieux4.html Apparently, JVC based their adapter at least somewhat around it. T1.5, pretty optically complex, and converts 35mm to 2/3''. Looks sweet! (I think it's also 30 grand.) Can you make anything out from the small diagram below in regards to how it works?

http://www.cine-one.com/pdf/angenieux_hdconverter.pdf

So cool.

If you look at the pdf, there's a relatively high resolution image of the thing. It has 12 lens elements in 10 groups, which would seem to imply that it's complicated as hell, but the average prime lens (relay lens) has what...6 groups or something and the DVX's zoom has about twice that...So maybe not.

The rear looks like some sort of relatively complex relay lens, and the front looks like any other static adapter...there's clearly a condenser behind the imaging plane and seems to be a piece of glass there on which the image is focused (but it's not ground glass, apparently, as Angenieux has stated). Trying this type of design on my own, however (adapter minus ground glass) results in an "areial image" with the same depth of focus as the 1/3'' camera itself...

There's a long, very technical, and somewhat heated discussion about this today at cinematography.net. I'm still trying to understand it. Several lens designers as well as some DPs weigh in. It's also enlightening in the "Panavision's corporate culture doesn't get it, Angenieux's does" sort of way. (Or maybe they are right and smarter than everyone else.)

Here's the link http://www.cinematography.net/Pages%20GB/%5Bcml-hdtv%5D%20Cine%20lenses%20on%2023%20HD%20CCD%20cameras.htm

Good luck!

One thing Galt said that definitely did resonate with me was, "Unfortunately, in practice, a relay lens only works well when designed for a particular prime lens."

That may have something to do with the exit pupil size and why some Cine lenses in fact do not work well with the Ang-Zeiss CLA 35. It must also be why the Zacuto adapter has a different relay to go with each lens.

Good luck to anyone who tries to home-build one. I can barely get my oscillator to consistently work!

Matt et al -

So I read the discussion on CML. Very interesting, and quite technical. I think I side with Angenieux ('optical relay adapters do exist') for several technical and non-technical reasons.

In this post: nontechnical reasons.

Here's my take on the situation - John Galt is correct, but he's talking about the wrong things.

Some background: I had the pleasure of hearing Mr. Galt speak at CineQuest 2005; he was there representing Panavision at the large-chip section of the "Day of the Cinematographer." He was talking about the Panavision Genesis camera; there were reps from Dalsa (representing their S35-sized single-chip camera) and NHK Japan (showing off a UHD, 32 Mpixel tv system - 8k, 120fps madness). But that's neither here nor there.

Mr. Galt and the Dalsa rep (whose name escapes me) had a remarkable disconnect with the audience. We're talking San Jose, CA - not Hollywood by any means. Most of the questions after the talk were from the low-budget regime, and I imagine most of the filmmakers there were too. What I took away from the talk: if you're low-budget, don't bother with Panavision or Dalsa. Ever. Mr. Galt also refused to engage questions about 35mm adapters, and dismissed the P+S Technik entirely (briefly citing the same technical assertions as in the CML thread, when he bothered to offer an explanation at all). I actually asked about 'optically enlarging the target plane,' basically what the Angeneiux and JVC adapter do, and he more or less dismissed the question.

Mr. Galt can't be blamed too much - for one, it was clear that the audience wasn't all that technical (myself included, at the time). Also, Panavision as a company really isn't interested in the lower-ticket items. (A colossal mistake right now, if you ask me - as distribution channels open up, it's not the large-ticket companies that will own the market, but the huge, huge base at the bottom.)

The real issue, I believe, is one of corporate focus. Mr. Galt's message was handed down from Panavision up on high. They focus on super-high-quality optics. They measure this quality by MTF, by lp/mm, by very technical means. And it's true, by those measures Panavision lenses are very good. But that's really not the whole story. By the technial specifications he cited in the CML thread, you're probably best off using whatever lens shipped with your camera.

As anyone on this thread knows, there's a lot more to the story than that. (And I'll get into that in the next post.)

Okay, technical reasons -

Let me distill the argument, if I can. John Galt (Panavision) said the MTF would be superior with a lens designed specifically for a 3-chip camera, rather than a 35mm lens + adapter. Basically, he's using resolution as the foundation of his argument.

The other reps - I think Cooke, Angeneiux, etc - never said he was wrong about that point. They just claimed the 35mm + adapter would be good, whereas he seems to think they would be unbearably bad. His arguments about why they would be unbearably bad don't hold much water.

The technical reason was the 'multiplying MTF' argument, mostly. Basically, one of the main causes of reduced resolution in an optical system is aberrations. When you have an intermediate scattering plate, you keep aberrations from building up on either side of the plate, but you do multiply their effect on resolution at each phase (like John Galt was saying, MTF of taking lens x MTF of scattering medium x MTF of relay lens). With an optical setup, you can actually build some correction into the adapter itself - if the MTF of the taking lens is degraded somewhat by, say, undercorrected spherical aberration, you can build some overcorrected spherical aberration into the relay lens. (The situation is vastly more complex, but the general principle is the same.) Mr. Galt also made some points about sensor size which are correct but a bit nit-picky.

So you may see some better performance with certain 35mm lenses - if the corrections that are built in are lens specific, then you may see increased aberrations (decreased MTF, lower contrast and lower resolution) with other taking lenses.

But we're actually splitting hairs here. If you're going to rent or buy lenses, you should do tests first. If real-world tests work for you, get the lens and forget MTF calculations; debates between lens techs are interesting but really far from relevant. When you go to a movie theater, the biggest impact on MTF etc isn't even the lens. It's the fact that the print you're watching is an nth-generation analog reproduction. You're seeing 2-3 or more layers of film grain superimposed on each other. You're seeing misregistered film, you're seeing aberrations in the printing optics, you're seeing flutter in the projector. You're seeing the resolution and color reproduction of print stocks.

In fact, I wonder to what extent this focus on lens quality is a holdover from the analog days, when you had to wring every last drop of resolution out of every step of the way, because the final product would be watered down by a dozen steps of analog reproduction? The fact is, even consumer-grade lenses on consumer HD cameras produce pretty sharp results. Not that it couldn't be better, but to the average viewer, it matters not at all.

But the real reason adapters are of interest to me is the possibility of getting something more than resolution out of lens choice. Things like bokeh, diffraction patters. Working advantages to certain lenses (like focus pulling or even cost).

And in that regard, Panavision's solution isn't for everyone anyway. I'm not interested in renting lenses from them, not if I own a bunch of still lenses that I want to use (or even if I want to build some myself). And I'm not interested in what they're interested in - resolution at any cost. I think there's some interesting things about old lenses that I miss in new ones.

For example: ever look at the focus blur in modern movies (or CSI, for example)? Ultra sharp, dinner-plate blurs. Sharp edged and crisp and totally artificial. Neon-green lens flares in mile-long cascades.

Then look at street lights out-of-focus in your own eye, through scratched glasses or in the rain. Totally different. Interesting in its own right. The blur pattern in most people's eyes couldn't be more different than the perfect-disc blurs of Panavision.

And with a world of lenses already out there, why would I want to restrict myself to the five or six, or even ten that Panavision chooses to release? I could really care less about the ultimate MTF. I want the freedom to choose lenses that fit my needs, even if my needs are a little retro, or funky, or just different. Motivated partly by cost, but even more by aesthetic concerns.

I mean, how boring would it be if every digital production used the same lenses just because they had the highest lab resolution?

So I have to disagree with John Galt's main point (not any of his specific arguments): relay lenses aren't pointless. And as the CML thread more or less concludes, we'll see at NAB.

Before we get too excited about these adapters:

The Angenieux adapter is an f/1.4 system. It does get the DOF characteristics of a 35mm lens, but not at f/1.4. The maximum working aperture of the taking lens is restricted to f/3.5 (the image side - the 2/3" chip - is working at f/1.4). If you open the taking lens up beyond f/3.5, you're just adding light to the lens system that won't get picked up in the final image (and won't contribute size to the individual defocus blurs).

And with regards to the JVC adapter, the same holds true. There's a limiting aperture in the adapter that will affect the largest aperture you can work at. This will be true of all optical adapters (no scattering plate) due to the optical invariant. In fact, you get basically the same depth of field that you'd get with a really large-aperture lens designed for a 2/3" (or 1/3" chip).

So with the JVC adapter, you're probably limited to the DOF of a 16mm lens at a medium aperture. Not crazy-narrow. But at least you get the optical characteristics of a wide selection of professional lenses. And you get other (non-DOF) advantages, like diffraction patterns, flare properties, and the ability to play nice with focus pullers, matte boxes and other toys.

If you really wanted the full aperture of the taking lens, the relay lens would have to satisfy the condition:

1/magnification=(f/# taking lens)/(f/# relay lens);

In other words, if you shrink the image by, say, 2.5 (35mm to 2/3" chip), your working f/number must increase by the same amount (f/2.0 becomes the impossibly small f/0.8). Note that if you merely had a taking lens with the same f/number as the relay optics, you'd have the same small DOF as if you were shooting on a larger format at a more reasonable f/number.

But... you'd only have to correct the relay optics for one specific conjugate, so it's not as impossible as it seems.



Okay... time for lunch.




*I think these are the figures, but feel free to double check or second-guess*

2/3" = 11mm diagonal
1/3" = 6.5 mm diagonal
35mm cinema = approx 28mm diagonal
16mm cinema = approx 13mm image diagonal

Thanks Ryan,
you just made me feel even dumber-er.

I dunno, Jack - which is dumber-er: reading posts on DVinfo to get to the bottom line, or spending countless hours reading optics textbooks in your spare time? (I have yet to get paid for a smidgen of optics knowledge, and it doesn't exactly make for good cocktail party conversation.)

I just posted some footage from my HZ-CA13U test at Sundance.

http://www.dvinfo.net/conf/showthread.php?t=85007

Thanks Tim.

It could just be me, but the DOF doesn't seem that strong in that footage. Like the out of focus areas aren't defocused nearly as much. That's probably the 16mm speaking and I'm just used to 35mm...

Thanks Tim.

It could just be me, but the DOF doesn't seem that strong in that footage. Like the out of focus areas aren't defocused nearly as much. That's probably the 16mm speaking and I'm just used to 35mm...
I've been told that the equivalent frame size is somewhere around 16mm/Super-16, so the short DOF won't be as pronounced as you may find in 35mm.

However, the 16mm focal length lens did feel a bit wider than I would have expected if the same lens were mounted to a 16mm or even a super-16 camera.
As you can see from the still photos, I was very close to the subjects and still getting "head and shoulder" framing.
There is no official literature on this device yet, so I'm just going by the bits and pieces I can get out of the JVC guys at these events.

The big surprise for me was the almost complete lack of chromatic aberration. If figured there would be some sort of distortion with whatever optical elements they have in there, but the footage seemed sharp edge to edge with no color separation.

I've been told that the equivalent frame size was somewhere around 16mm/Super-16, so the short DOF won't be as pronounced as you may find in 35mm.

However, the 16mm focal length lens did feel a bit wider than I would have expected if the same lens were mounted to a 16mm or even a super-16 camera.
As you can see from the still photos, I was very close to the subjects (within 2 feet) and still getting "head and shoulder" framing.
There is no official literature on this device yet, so I'm just going by the bits and pieces I can get out of the JVC guys at these events.

The big surprise for me was the almost complete lack of chromatic aberration. If figured there would be some sort of distortion with whatever optical elements they have in there, but the footage seemed sharp edge to edge with no color separation.
Actually I noticed vertical red/green chroma aberration on a lot of the out-of-focus highlights. Take the first shot for example. The white bar on top of the car has a green edge on top, red on bottom. Looks like it pops up a lot in a couple other places too.

I must say the in-focus areas are extremely sharp though.

Ben's got better eyes than I do. Nice job, Tim & JVC!!

Haha. I guess every imaging system has its flaws. Overall it's great image, don't get me wrong :)

Actually I noticed vertical red/green chroma aberration on a lot of the out-of-focus highlights. Take the first shot for example. The white bar on top of the car has a green edge on top, red on bottom. Looks like it pops up a lot in a couple other places too.

I must say the in-focus areas are extremely sharp though.
Yes, as with any optical system there is always going to be some amount of CA. However, compared to what we have become accustomed to with 1/3" 3-CCD imagers paired with low-cost zoom lenses, this new system is amazing.

The new HD200 and HD250 also have lens shading capabilities, so this may also be a factor. I have no idea if the prime was programmed into the lens shading memory before I tested it.

http://www.distantfocus.com/projects/montage/

I'm just not impressed with the JVC unit. It's not going to be cheaper than the M2, SGpro, or Brevis. It doesn't work with inexpensive 35mm SLR lenses. The DOF isn't very shallow. And the CA is worse than the latest adapter footage I've seen.

On the plus side, it's sharp. It doesn't seem to lose much light (how many stops?). It's probably a breeze to setup. It looks solid and professional.

I haven't used the adapter, but I think I understand how it works. And it appears that the only light losses would be scattering and absorption in the intermediate glass, so I'd expect very, very little light loss - less than a third of a stop (roughly similar to the difference between the T-stop and the F-stop on a given lens).

Now, the real 'loss' is the fact that it only operates at f/3.5ish on the object side - so opening up your taking lens beyond that won't give you more light (at least not in your image; it'll probably give you a little more flare and image defects).

And I think it's a neat idea, but I agree it's not that revolutionary. HD200 + adapter isn't that much cheaper than RED. And as fond as I am of elegant solutions (like an optical-only adapter), there's no real substitute for a large imaging chip.

http://www.cinematography.net/Pages%20GB/%5Bcml-hdtv%5D%20Cine%20lenses%20on%2023%20HD%20CCD%20cameras.htm

From this article, it seems like the only problem with a purely optical solution (taking lens at flange distance >> relay lens at flange distance >> camera sensor) is the chromatic aberration that must be overcome by CCD sensors.

That said, would this purely optical solution be MORE possible on a 1/3" CMOS sensor?

Beam splitting prisms introduce undercorrected spherical and chromatic aberration, and can produce astigmatism at large angles. Correcting this is really the job of the lens, not the CCD. The normal solution is to make a lens system telecentric in image space, which means the aperture appears to be directly above each pixel (from that pixel's perspective), and the light ray bundles arrive perpendicular to the CCD/prism block.

Telecentric lenses are several orders of magnitude more difficult to design, and have some subjective compromises. They typically require many more elements, are difficult to fabricate and cost more money. And the image they produce looks worse: next time you're watching sports, check out the image blur in the background. The out-of-focus highlights should blur into a roughly polygonal shape with (usually) a weird, banded radial pattern inside. This is probably the result of residual zonal aberrations (exacerbated by the video camera's sharpening algorithms). But regardless of the cause, it's ugly.

A single-chip CMOS (or any single chip, for that matter) wouldn't require telecentric optics, so you've got a big improvement there. But, as I mentioned earlier, the relay optics in a non-scattering system would have to be awfully, awfully fast to give you the same DOF as in the 35mm regime (to get the DOF of f/2.0 in the 35mm regime, a 1/3" chip camera would need relay optics at around f/0.57, I think).

And while his 'multiplying MTFs' garbage sounds like hand-waving designed to cow Panavision's customers, I'll have to cede John Galt his main point in that post: in terms of sharpness and DOF, you might be 'best' off with a custom designed lens for that particular camera. In addition to shallow DOF, the f/1.4 optics at Panavision would gather a lot of light.

Of course, if you're Panavision, sharpness and DOF are probably your only concerns. If you're a filmmaker, you might have more artistic (what are the subjective characteristics of this lens?) and practical concerns (gee, I already own all these 35mm lenses that I like, and I don't have $10,000/day to spend on a camera package).

Oops... I found this great link:

http://www.dpreview.com/news/0210/02100402sensorsizes.asp

that explains CCD sizes a little better. Turns out that the relay optics would have to be more like f/0.40, which is similarly ridiculous.