Cine lenses on 2/3" HD CCD camera

C.R. Caillouet wrote:

>The desire to mount cine lenses on HD cameras comes up often on this list. Would you care to weigh in with your experiences in this area or to give some technical background on the problem. ...........

This is a very complex subject, but the simple answer is that, for any given final image size, the image from a 2/3" CCD camera requires more than twice the magnification than the same image from a 35mm film frame. This means that the performance of almost all of the imaging characteristics of the 2/3" CCD lens must be twice that of its cine equivalent to achieve the same image quality. This is particularly true of Modulation Transfer Function (MTF). MTFs are usually represented as % response at a particular frequency in cycles per millimeter (line pairs/mm). For instance 80% MTF at 20 cycles per millimeter would be considered an excellent cine lens performance. However, at 40 cycles per millimeter that same lens would only achieve about 60% MTF. Because a cine lens is designed to produce an image in air, and a lens on a CCD camera has to produce an image through a thick color beam splitting prism, a relay lens is required to try to compensate for optical aberrations introduced by the prism.

MTFs of image forming components are multiplied together to achieve a final system MTF. Eliminating comparisons with other lens manufacturers, if we assume a Panavision Primo=81 cine lens with better than 80% contrast at 20 cycles/mm is combined with an equally good relay lens, the combined effect would be a lens with 64% MTF (Modulation Transfer Function) at 20 cycles/mm and 40% at 40 cycles/mm. To put this in perspective, a Primo Digital=81 lens designed specifically for the 2/3" CCD format has more than twice the frequency response of the best Primo cine lens or better than 80% at 40 cycles/mm (I must point out that the Primo Digital=81 lens only covers an 11 mm image diagonal and the cine Primo covers a 27 mm image diagonal).

It is possible to use a cine lens of about 100 mm or greater on a 2/3" CCD camera without a relay lens. However, to minimize the spherical aberration introduced by the prism, the lens should be stopped down to at least f5.6.

There were many other factors which influenced our decision to design a completely new series of lenses for cine style applications of 2/3" CCD cameras. One of the most important of these was depth of field control. Since depth of field is also proportional to magnification, a 10 mm lens at f1.9 will have approximately the same depth of field as a 1.85:1 35 mm format 21 mm lens at f4. All of the new Primo Digital lenses are designed to produce better than 80% MTF at 40 cycles/mm at full aperture. Which, depending on the lens, varies between f1.45 and f1.7.

In conclusion, even an excellent cine lens, combined with a theoretically perfect relay lens, will produce a relatively poor image on a 2/3" CCD camera when compared to even an average quality video lens.

John Galt,

Senior Vice President, Advanced Digital Imaging

Panavision 6219 De Soto Avenue

Woodland Hills CA

John,

I had the pleasure to work with the 11-1 digital primo recently and have to fully agree with your comments regarding the optical quality of this lens. I only wish that Panavision will be awarded an Oscar for their brilliant design work on the lens and viewfinder. Can the digital primo be used on Panavision 35mm film cameras? Or is it only for the F900. I also worked with the modified Canon wide-angle HD lens on this last shoot.

Are the Canon's being used only until Panavision releases their own wide-angle lenses? Besides the rear mount, modifications were there any other optical improvements or changes made to these Canon lenses?

What options are forthcoming for hand held work? Is there any chance of an on board zoom motor with rocker zoom control which will allow for hand holding the camera identical to the ENG lenses. I understand this goes against film style operation but I must say that for quick handheld and mobile operation this is an ideal method.

Another item I find unique about Panavision's EC system is every camera I have worked with comes standard with the CA-901 HD-SDI camera adapter which I find invaluable to work with instead of dealing with only one analog output.

Tony Salgado DP SOC

I own the Arri adaptor ring for use on a B-3 mount. You can mount any lens with an Arri ring (sorry can't remember the name of the Arri mount). I purchased it when I got my 700A a few years. The image looks GREAT, however the downfall is, it requires about 40 minutes to do the conversion. I have only used it twice in three years, luckly just enough to pay it off (it goes for about $2,000) I have NOT used it on my 900...as of yet.

for what it's worth...

Chris W HD Camera Support Los Angeles, CA - Orlando, FL

John Galt (PANAVISION) wrote a reply to the question of C.R. Caillouet:

>In conclusion, even an excellent cine lens, combined with a theoretically perfect relay lens, will produce a relatively poor image on a 2/3 " CCD camera when compared to even an average quality video lens.

This is simply not true. John Galt's reply combines some true and false statements and completely overlooks the one decisive aspect of the whole concept. Read on!

It is in fact true, that a cine lens cannot be attached directly to a 3-chip CCD-camera, mainly because of the thick beam splitting prism. The design of the cine lens assumes, that the optical path between the lens and the film consists of air, not glass. In an electronic camera the optical path contains a beam splitter prism made of optical glass with a much higher refractive index compared to air. This would cause severe spherical aberration with a directly attached cine or photo lens used wide open.

There is another image degrading effect: color shading caused by the pupil of the cine lens appearing too close to the camera, so that the light beams going to the top and bottom edge of the frame pass through the beam splitter prism not rather parallel to the optical axis, but at very oblique angles.

However, with his considerations about MTF John Galt is not on safe ground. Simply multiplying the MTFs of single components to calculate the resulting MTF is definitely incorrect with real photographic lenses. Here is a striking example: In real world optics the resulting MTF of an 8-element lens is certainly not equal to the multiplied 8 single MTF values of its elements, or is it?

Now, let's take a closer look at the relay lens. It has to perform two essential tasks. 1. (This one was completely neglected by John Galt) In order to maintain the angle of view of the attached cine lens, it has to reduce the field size from 35 mm cine (approx. 27 mm diagonal) to 2/3" CCD (approx. 11 mm diagonal) which goes by a reduction factor of about 11:27 which equals 1:2.5. In the process of this size reduction the spatial frequencies of corresponding structures are increased by this same factor of 2.5x. Thus, the 20 Lp/mm structure at the output of the cine lens will be transferred into a 50 Lp/mm structure at the output of the relay lens. At the same high MTF, which is clearly over 90% in case of a Zeiss Ultra Prime lens and the Zeiss/Angenieux CLA 35 HD .( Yes ninety not just 80! And is has been measured with actual production units, not just theoretically calculated.)

2. It has to compensate for the refractive index situation in the CCD-camera and has also to direct the light rays in a way that avoids color shading.

It is a basic law of optical imaging, that the product of field size and numerical aperture is constant in front and in the rear of a lens (this is by the way simply the optical expression of the universal law of energy preservation). From this follows, that a cine lens + relay lens combination with f-number f/1.4 utilizes the cine lens only at f/3.5 where its performance is even better than wide open.

In terms of depth-of-field, a cine lens + relay lens combination with the same angle of view and the same f-number as a TV lens delivers exactly the same results.

In conclusion, a well-conceived and well-made relay lens is able to preserve the image quality of excellent cine lenses, and to deliver their well-known image quality to the large screen. This is achieved even though it requires, due to the small size of the CCD, to deliver high spatial frequencies at high MTF values which are not normally utilized in 35 mm cine.

Helmut Lenhof

Product Manager,

Digital Cinema Lenses Carl Zeiss D 73446 Oberkochen

Helmut Lenhof wrote :

>In conclusion, a well-conceived and well-made relay lens is able to preserve the image quality of excellent cine lenses, and to deliver their well-known image quality to the large screen.

Helmut,

Can we infer from your statement that Zeiss has such a relay lens available for sale and (sorry for the blasphemy) does it work with cine lenses other than those made by Zeiss?

Thanks for the contribution.

Vision Unlimited/LA ...searching for the right tool for the job...

Helmut Lenhof wrote :

>In conclusion, a well-conceived and well-made relay lens is able to preserve the image quality of excellent cine lenses, and to deliver their well-known image quality to the large screen.

John Galt (PANAVISION) wrote :

Gentleman :

I long to be able to see, with my own eye, the merits of both of these two statements from the eminent experts at Panavision and Zeiss. However, until Panasonic releases their AJ-HDC27V and / or Sony releases a board to "flip" the image on the HDW-F900, either of which will allow for the use of the Zeiss/Angenieux CLA 35 HD adapter with Zeiss Ultra Prime 35mm lenses on 2/3" CCD cameras in a vertically proper way, it will be difficult for anyone outside of a laboratory to evaluate the merits of these two statements.

Seeing is believing.

Bob Zahn

Bob,

Since I can't see in the near future the ability to mount a Panavision lens on the HDC-27 camera I must agree that it will be difficult to compare image quality apples to apples unless Sony decides to offer an inversion board in the F900.

I have tested the Zeiss adapter and my only objection was the overall length in size in addition to the pricing.

I must say that my personal experience using the Panavision digital primo was that I was quite pleased with the optical performance.

On a side note the viewfinder Panavision developed is by far the best piece of engineering design and offers maximum flexability. I have now used the Sony and Abekas versions and by far Panavision's version is number one, with Sony 2nd and just forget about the Abekas.

Disclaimer: These are my own personal opinions and results may vary depending on the end user.

Enuff said,

Tony Salgado

Helmut Lenhof said; "In conclusion, a well-conceived and well-made relay lens is able to preserve the image quality of excellent cine lenses, and to deliver their well-known image quality to the large screen. This is achieved even though it requires, due to the small size of the CCD, to deliver high spatial frequencies at high MTF values which are not normally utilized in 35 mm cine."

***** Respectfully, I have some questions if I may :

Are you saying that Zeiss can make a relay lens (or any lens) that is loss-less? No light dispersion, no loss of MTF, unity?

How do you compensate for the differences / tolerances of each of the Primes that would be used with the adapter? i.e....

If you are not focusing on an intermediate surface (which would definitely cause a loss in MTF) how are you compensating for the different way each prime will insert light into the system? Wouldn't you have to build a relay lens for each lens you use in order to maximize performance and avoid lateral chromatic aberrations?

If the lenses are designed for a medium like film which has an emulsion thickness much greater then that of a semiconductor's surface and accordingly broader tolerances for image flatness, how is it that you are able to compensate for this or have you found that you even need to?

Would such a relay work with Cine Zooms as well as Primes, Zeiss doesn't currently manufacture any Cine Zooms do they? Who's Zooms would you make it compatible with?

I do work for Panavision in Digital but the questions are meant honestly.

sincerely Scott MacDonald

Excuse me, but wouldn't it be a better comparison between 16mm format lenses and 2/3" ccd cameras?

In terms of MTF and Resolution, I always understood that 16mm Format lenses were sharper because the image is magnified more times than 35mm when projected to the same size. That owing to physics, and costs of production, 35mm lenses are designed to cover the larger surface, and produce a certain Coc, while 16mm format lenses had a tighter Coc, but were less expensive to produce to this level as they had a smaller area of coverage behind the lens than 35mm.

Any thoughts? do the Relay lenses work for 16mm format lenses? Is it a more valid comparisson 16mm : 2/3 inch

rather than 35mm :2/3 inch.

Steven Gladstone Cinematographer -

Gladstone Films Cinematography Mailing List - Listmum

Better Off Broadcast (B.O.B.) New York City, U.S.A.

Hi CML :

Here at Cooke Optics we have been following with interest the discussion about the use of 35mm Cine to HD adapter optical units. I have passed along the emails to Jon Maxwell in our design department and below is what I hope is a clarifying technical analysis of what has been posted on this topic.

Thanx

Les Zellan

John Galt from Panavision has been corresponding with Charles Caillouet Jr and previously with Jason Rodriguez about relay optical adaptors and saying that "they don't work very well". Actually he has been putting it more strongly than that, and has said that a 35mm cine to HD relay "...will (inevitably) produce a relatively poor image...when compared to even an average quality video lens.".

Helmut Lenhof of Zeiss has responded with a neat summary of the technical situation, which says exactly the opposite.

Scott MacDonald at Panavision has asked whether Helmut Lenhof is really saying that Zeiss have a perfect relay adaptor? ( the detailed questions that he asks provide a useful focus for this discussion).

Charles Caillouet Jr at Vision Unlimited has also come back and asked whether Zeiss's adapter optic works with lenses other than Zeiss.

These are powerful, passionate and relevent claims, counter claims and questions and I am essentially writing to support Helmut Lenhof's view, to attempt to provide some answers for Scott MacDonald's and Charles Caillouet Jr's questions and to add a litle more to (hopefully) clarify the situation with regard to image contrast (that is "MTF", or "Modulation Transfer Function", which most engineers are happier with, rather than just "contrast", because what we are talking about is very little to do with stray light, which also effects overall image contrast, I'll use "image modulation" ).

Helmut Lenhof says, if I may summarise: "A properly designed relay adaptor optical unit matches cine (prime or zoom) lenses correctly to the 2/3 inch HD format, compensating the aberrations of the colour prism in the camera, and arranging for the image forming beams in the camera to traverse the colour prism correctly, to avoid colour shading.". Exactly right,

He also explains the crucial significance of the demagnification which takes place between the cine image and the HD image (approximately 1/2.5), and how (a) this scales-up the spatial frequencies involved by X2.5 and (b) how this also means that the HD colour prism, which only accepts an aperture of f/1.4, stops down the front lens to a maximum aperture of f/3.5, thereby clarifying the situation with regard to the increased depth of field relative to large aperture lenses used on the 35mm format that is currently inevitable with digital imaging (whether it is based on a custom digital lens or on an adapted lens). He could also have said that such a relay adaptor can be matched to the precisely specified three colour chip positions in the digital camera,

So, to answer Scott MacDonald's question about whether Zeiss can claim to have a perfect adaptor "which is loss-less, no light dispersion, no loss of MTF, Unity?", the answer strictly has to be "No", but this is only because no lens is perfect. I have not actually been able to make a critical assessment of many results from that adapter for myself, but, for all practical purposes the answer to Scott MacDonald's question is that a properly designed relay adapter optical unit in combination with prime lens can in principle be very very nearly as good as the prime lens on its own, and the same applies with a zoom lens.

Put it like this: If what John Galt says was right, ordinary range extenders wouldn't work; but they do work and are widely used. Yes of course they have to be designed and manufactured properly, and yes they do reduce the aperture by the magnification ratio (f/2 becomes f/4 if the magnification is 2x), and this is exactly in the same way that the 35mm to digital adapter works, but in the opposite direction ( the digital increases the aperture at the final image, rather than reducing it).

This also answers the question from Charles Caillouet Jr about whether the Zeiss/Angenieux adaptor can be made to work with any lenses designed for 35mm cine applications (eg Cooke S4s of course, but I would say that wouldn't I !). Yes, as far as we can tell, the Zeiss/ Angenieux adaptor can be used with most non Zeiss/Angenieux lenses. The issue here is only about the exit pupil position of the lenses that are used with the adapter, but there is quite a large tolerance on this, precisely because the colour prism aperture only allows the adapter to see f/3.5 of the front lens aperture.

The point about image contrast is, as Helmut Lenhof correctly points out: you cannot calculate or measure the image contrast of an 8 element lens by calculating or measuring the image contrast of the 8 individual elements and then multipling all these values together.

The situation we are dealing with is more subtle, because the primary image is of a high quality, but the principle is exactly the same: The final image doesn't know anything about any intermediate images that have been formed on the way and only cares about the final image contrast. Next, and this is the key point, the final image contrast is dependent on two things: diffraction and, additionally, aberrations. You can't do anything about the diffraction effect, which is only dependent on the final image f number (f/1.4), which, at 50 cycles per mm at the final image gives a basic diffraction-based image contrast of about 95%: that is, halfway down to the 90% that Helmut Lenhoff claims for the Zeiss UltraPrime lenses at that spatial frequency (equivalent to 20cycles per mm at the primary image), which I have no reason to doubt.

Of course, if cine lenses only had to cope with 20cycles per mm on film the lens designer's job would be very easy. 40 cycles per mm is more like it, which is equivalent to 100 cycles per mm at the relayed image on the 2/3inch digital image chip, and at that spatial frequency the diffraction component is about 90%.

Then, the aberrational effects on image contrast are, yes, added to by the relay optic, but in a complex and non-obvious way, depending on the way the aberrations of the front lens and the adapter optic interact. you certainly cannot just multiply the MTFs of the camera lens and the relay optic together and get the right answer.

My own interest in this topic is not based only on my involvement with Cooke Optics. Having recently come from the Optics Section at Imperial College in London, I have a long involvement with imaging issues and it is heartening to find the cine world equally fired up about these finer points of technical discussion.

Jonathan Maxwell Lens Design, Cooke Optics

In response to Helmut Lenhof's comments to my response to C.R.

It is my understanding that the Zeiss claim of 90% MTF for Zeiss Ultra Primes is for 10Lp/mm. On a 35mm film frame this would represent the equivalent of less than 480 horizontal TV lines. A high definition 24P CCD sensor is 1920 pixels or 960 line pairs, the limiting resolution of the sensor is 100 Lp/mm, not 10 or 25 (for a perfect relay system derived from a Zeiss cine lens).

=46inally, theory is great (even if we can prove that bumble bees can't fly), but the proof of the pudding is in the eating. For those of you fortunate (or unfortunate) enough to attend the NAB Convention in Las Vegas later this month I believe that the Zeiss/Angenieux CLA 35 HD will be exhibited at the Sony Demo Center in the same area that Panavision will be demonstrating it's (non-relay) Primo Digital=81 lenses.

Since I still have some word count left, I would like to make a comment on circle of confusion. Resisting my usual description, which is myself and some close friends standing in a circle discussing this subject......

A 1920 x 1080 pixel 2/3" CCD has pixels which are 5 x 5 microns in size. If we are not going to upset Mr. Nyquist (create aliased images or moire patterns on fine detail) then the minimum circle of confusion is 10 microns. In practice, because optical low pass filters have slope, just like electronic filters, the real circle of confusion is more like 2.3 pixels or 12 microns. Please remember depth of field is a convenient construct, a rule of thumb. When you focus a lens only a small point in space is truly in focus. Depth of field implies that a certain amount of defocus will be acceptable. Whether this is true or not depends on many factors. Do your own tests for your particular application.

John Galt,

Senior Vice President,

Advanced Digital Imaging Panavision

6219 De Soto Avenue Woodland Hills CA 91367-2602

Once again I apologize for my tardy response! In response to Helmut Lenhof's comments to my response to C.R.

Galt- In conclusion, even an excellent cine lens, combined with a theoretically perfect relay lens, will produce a relatively poor image on a 2/3" CCD camera when compared to even an average quality video lens. Lenhof- This is simply not true. John Galt's reply combines some true and false statements and completely overlooks the one decisive aspect of the whole concept. Read on.......

I suppose I deserve this. I entered the trap of simplicity and now I must pay the price! However, I stand by my previous statements. I also do not wish to enter into a partisan argument about one particular manufacturer's product versus another's, even if Mr. Lenhof's comments makes this difficult. The difference between theory and practice is execution. I believe the thinking at Panavision back in the early 1980's when they designed and built optical relays for video cameras (including the second generation HDTV camera) was similar to some of the optical simplifications espoused by Mr. Lenhof. Unfortunately, in practice, a relay lens only works well when designed for a particular prime lens.

John Galt,

I am not sure if Mr. Lenhof was being ironic when he asked if cascading MTF's from individual elements in a lens would result in the final MTF of the lens. Conceptually, Modulation Transfer Functions represent the spatial frequency response characteristics of imaging devices or systems. For instance, a lens, a film emulsion, a camera movement, an optical printer etc. can all be characterized with individual MTF's which, when cascaded together (multiplied), will result in a final system MTF. Parenthetically, this is reason that final projected image quality of motion picture film is so inferior to the image recorded on the original negative. Although MTF is a good general method of characterizing optical performance, it is not everything. I am usually most concerned about the performance of wide angle lenses through relays and prisms. ignoring MTF, the lateral chromatic aberrations (color fringing) make most of these systems useless for blue screen and green screen applications. In addition, the veiling glare characteristics of these systems also make them problematic in practical photographic applications.

John Galt,

Hi CML:

Thanx

Les Zellan

Part III (continued from Part II)

Jonathan Maxwell Lens Design, Cooke Optics

Part II

Helmut Lenhof of Zeiss has responded with a neat summary of the technical situation, which says exactly the opposite.

Charles Caillouet Jr at Vision Unlimited has also come back and asked whether Zeiss's adapter optic works with lenses other than Zeiss.

He could also have said that such a relay adaptor can be matched to the precisely specified three colour chip positions in the digital camera,

<x-flowed> Thanks to Helmut Lenhof, John Galt and Jonathan Maxwell for some well thought-out positions on the lens issues.

I guess we are essentially at what John Galt called "the proof of the pudding" stage. I hope that we can ingest some samples of the results at NAB next week.

cheers, crc

Vision Unlimited/LA ...searching for the right tool for the job...

John:

Where and when will Panavision be screening footage of material shot with the Cine Alta transferred to 35mm film. I'd love to attend. Please let me know.

Bob Zahn President, BVR Ltd. Contributing Editor, Digital Cinema Magazine

Former VP of PERA (Production Equipment Rental Association)