We have witnessed a revolution in digital cinematography over the past few years. Features once reserved for $100,000+ professional digital cameras have found their way into camcorders priced under $10,000. Most of the tools required for the long sought after “film-look” (24P, HD recording, ciné gamma and extended dynamic range) are now finally in the hands of the low-budget filmmaker.
Unfortunately, as the overall size and price tag of digital camcorders shrink, so does the physical CCD size. A typical HDV camcorder contains three 1/3” CCDs with an imaging area roughly only 6mm diagonal. The side effect of a smaller chip size is an almost uncontrollably deep depth of field (DoF.) This “feature” of very small CCD camcorders generally excludes them from being serious contenders for feature film or dramatic television work.
*Grey regions represent 1.78:1 (16×9) aspect ratio for acquisition. ©Tim Dashwood.
DoF is tied to four main factors: focal length, aperture size, focus distance and the circle of confusion (CoC) of the format. There is much written on the details DoF, so I won’t delve into all aspects in this article. However, there is one concept that any digital cinematographer needs to understand:
The smaller physical size of the CCD requires us to use much shorter focal lengths to achieve the equivalent angle of view of larger film formats (sometimes referred to as the cropping factor,) and hence shorter focal lengths (wider lenses) mean deeper DoF.
For example, let’s say we were shooting a typical medium shot with a 25mm focal length in Super 35mm format, at T2.8, and our subject was 6 feet from the lens. Our angle of view would be around 50°, and our near “in-focus” distance would be about 5′ with a far distance of 7′ 6″, for a DoF total of 2′ 6″.
If we wanted to achieve the same angle of view on a typical HDV camera with a CCD of 1/3″ size, at F2.8, we would need to use a focal length of 5mm to achieve an angle of view of 50°. This now means that our near focus is 1′ 10″ and our far focus is infinity. With everything in the scene “in focus,” it is very difficult to draw the audience’s attention to our subject.
So what can we do to shorten the depth of field?
If the 1/3″ lens is fast enough we could open the aperture a couple stops to F1.4 and control the exposure with some ND filters. This would give us a near focus distance of 2′ 9″, but the far distance would still be infinity. A slight improvement, but the disadvantage of opening the aperture all the way is that we are now susceptible to extreme chromatic aberration or white shading.
A common technique is to move back and use a longer focal length (telephoto.) This appears to shorten our DoF considerably but it doesn’t really change because the act of moving farther from our subject has cancelled out the effects of the longer focal length. Instead we have just “compressed” the magnified background elements together with the foreground for a telephoto look.
The only viable option is to use a ciné lens adapter. From the home-built versions that sell for a few hundred dollars to the very professionally engineered P+S Technik mini35 or Redrock Micro M2, these adapters all utilize the exact same concept. An image is projected by a ciné lens or SLR photo lens onto a focusing screen or ground glass equivalent in size to 35mm or 16mm. The relatively larger size of the projected image allows for the use of longer focal length lenses, and therefore shorter depth of field. The projected image is then “re-photographed” with a relay lens connected to the 1/3″ CCD camera.
These adapters are increasing in popularity among budget minded filmmakers and music video cinematographers.
Lens adapter options: P+S Technik MINI35 (left) & Redrock Micro M2 (right) connected to JVC HD100 series cameras.
The downside of ground glass lens adapters?
There are a few things to consider when using a typical ciné lens adapter. First of all, the ground glass/focus screens utilized in most systems have a light-loss averaging at least 2 stops. This means more light or faster lenses will always be required for interior scenes. Secondly, the grain is usually visible in the ground glass/focus screen and therefore most adapters employ a rotating or oscillating system to scatter the grain pattern, requiring battery power and sometimes creating audible noise.
Also, vignetting can be a problem since the light dispersion in the ground glass/focus screen is not always as even we’d like.
Finally, the majority of the cine lens adapters on the market require the use of the camcorder’s lens in macro mode to relay the image to the CCDs. Some systems, like the mini35, have their own proprietary relay lens. Either way, this usually means very long lens assemblies.
Pure Optical Solution:
The ground glass/focusing screen seems to be the common weak link in these systems. So what would happen if the ground glass was taken out of the optical path?
The engineers at JVC have answered that question with the development and release of the HZ-CA13U PL mount ciné lens adapter, designed to attach directly to the bayonet mount on the JVC ProHD line of HDV camcorders. It is not dissimilar in design to Angeneuix/Zeiss’ CLA35HD, but designed for 1/3” CCDs.
This new ciné lens adapter has no ground glass/focusing screen. Ciné lenses attached to the industry standard PL mount focus an aerial image on a “virtual focal plane” at the standard 52mm flange distance, maintaining the angle of view and CoC/DoF characteristics of 16mm film. Eleven optical elements in the adapter then reduce the size of the aerial image down to 1/3″ CCD, while compensating for the aberrations of the prism.
The lack of a ground glass/focus screen means that almost all of the light captured by the ciné lens is transmitted to the CCDs. (I’m told that the actual loss in transmittance is less than ½ stop.) This is a huge advantage over the GG based competition for shooting in low-light.
There is no added grain to worry about, and also no extra power required.
The JVC HZ-CA13U PL Lens Adapter Prototype.
I first crossed paths with this new lens adapter at NAB ’06. At that time the product was still in development. I honesty must say that I didn’t have very high expectations at that time…mostly because I didn’t understand how the concept could possibly work. Skip ahead to Sundance 2007 where JVC had the prototype finished and on display at HDHouse. I informally tested the prototype on Main Street in Park City. The only PL lens focal length available was 16mm, but the quick results impressed me so much that I volunteered to do a full array of tests on the device.
Frame Grabs from 16mm Zeiss lens (16mm lens, T2.8, approx. 3’ focus).
Notice the fairly short DoF for such a wide lens, and the very nice prime lens flare!
JVC also loaned me a GY-HD250 for the test. The GY-HD200 and GY-HD250 cameras have a built-in “image flip” option that allows upright image orientation recording required by most lens adapters. As soon as I had the camera and adapter in my hands I contacted Clairmont Camera and arranged a test of their lenses with the Samuelson depth of field charts. With the help of my friend and fellow cinematographer Brad Smith, we spent two days testing every prime lens we could get our hands on. Cooke S4s, Zeiss Ultra Primes & ‘Super Speeds,’ even Clairmont Anamorphics, we tested them all.
Dozens of prime lenses in boxes. Capturing HD frames on the MacbookPro.
DPs Tim Dashwood (left) and Brad Smith test DoF of the HZ-CA13U
next to an Arri SR3 (photos: Grant Padley).
The first challenge we faced was properly backfocusing the HZ-CA13U. Backfocusing is very important with lens adapters, but the traditional designs allow for the relay lens to be focused on the grain of the ground glass. Since the HZ-CA13U has no ground glass, we had to develop a better plan. Our first instinct was to pop a 10-100mm Zeiss zoom lens on the adapter and the use traditional backfocusing technique with a Siemens Star.
This worked fine, but who wants to rent and bring a zoom lens on every shoot just to set the backfocus?
We tried a few different techniques, but eventually decided that using a long prime (32mm, 50mm or 85mm,) setting focus on the lens barrel witness mark to an exact distance (say 15 or 30 feet,) and then backfocusing on the Seimens Star at that measured distance seemed to work well. The only caveats are that you must have a HD monitor on set, and you must ensure that your lenses’ flange depth is set correctly in prep. Rental houses like Clairmont maintain their lenses to a very high standard, but beware if you are brushing the dust from some old lenses at your film school.
The camera mount and adapter will both be sensitive to temperature changes, so it will be important to maintain a reliable backfocus regiment when changing locations, and to let the camera warm up before setting backfocus. The backfocus adjustment on the prototype seemed to be very sensitive to small adjustments. JVC tells me that it will be ‘geared down’ in the production model.
[Note: JVC sent me a production model for evaluation after this initial review was written. I can confirm that the backfocus ring does offer much more resistance, moves smoother, and is therefore easier to adjust.]
Adjusting the backfocus (photo: Grant Padley).
The HZ-CA13U appears to be very well built. It only has two moving parts – a PL mount lock and the backfocus knob. The prototype didn’t have a focal plane marking, but the production models will have a focal plane line scribed around the circumference of the barrel. It didn’t have a hook for the measuring tape either, but I set a lens rod support at the focal plane to hook onto.
85mm Zeiss Ultra Prime mounted on the JVC HZ-CA13U and HD250.
The length from mount to mount is just under 5 and a half inches, and the adapter weighs less than 2 pounds. However, I highly recommend using a lens support for the adapter to avoid too much stress on the camera’s bayonet mount. The production model will have a ¼” screw hole on the bottom.
We tested angle of view, transmittance, sharpness, chromatic aberration, and depth of field characteristics of primes on the adapter and compared the results to a standard 1/3” video zoom lens and a Super-16 Arri SR3.
Field of View/Angle of View:
We found that the HZ-CA13U has an equivalent angle of view of standard 16mm. The fact that JVC chose a 16mm frame size aerial image will come as a surprise to most because the trend tends to be 35mm for all of the other ciné lens adapters. I was also disappointed when I first read the spec. However, I now realize that this was a calculated move on JVC’s part. There is an abundance of standard PL 16mm lenses available around the world. From film schools to rental houses, it won’t be hard to find a half-decent set of 16mm format primes somewhere in your part of the world. Of course, any PL prime or accessory (35mm or 16mm format) will work with the adapter.
HZ-CA13U frame grab layed over Super-16 video tap image of Arri 1.78:1 ground glass.
I was slightly disappointed when I determined that the maximum usable aperture of the HZ-CA13U seemed to be T2.8 The documentation states an exit aperture of f/1.4, and after a day of research, I realized that my findings of T2.8 on the lens side are a result of a basic optical imaging law known as the Lagrange invariant. This affects all lens systems in the same way. Using the horizontal frame size values of 16mm (9.35mm,) 1/3” CCD (4.8mm,) and the exit aperture of f/1.4, I determined mathematically that the max lens aperture is f/2.73.
The bottom line is that using high-speed lenses will have no benefit to exposure or depth of field once they are opened past T2.8.
However, if you are used to shooting at open apertures in low-light on your video lens, you will be surprised to find that the equivalent T-stops marked on the ciné lenses all seem to be about 1½ stops brighter than the same video zoom F-stop values. This means that T2.8 on a PL prime lens has an equivalent exposure to about f/1.7 on the video zoom lens. This phenomenon seems to be an added benefit of the optical invariant. The easiest way to manage this is to rate the camera 1½ stops more sensitive than usual.
Sharpness & Vignetting:
I tested sharpness on every available focal length from 9.5mm to 85mm, in both 16mm and 35mm format lenses. All of the primes in the 12mm to 85mm range showed exceptional edge to edge sharpness, at least as good as the Fujinon 13×3.5 video zoom I used as the benchmark. The 9.5mm Zeiss showed a slight loss of sharpness in the corners, but nothing out of the ordinary. There was hardly any visible vignetting on any of the lenses we tested.
Chromatic Aberration & White Shading:
Chromatic Aberration (CA) is a big problem with 1/3” 3-CCD HD cameras. Absolute precision is required in the optics to avoid CA. JVC ProHD owners who use the “stock” 16x lens know all too well about the green/magenta fringing common with ‘entry-level’ lens.
I expected to see some CA fringing, but I was surprised at how little actually showed up, considering these lenses were never designed for use on a 3-CCD system. The CookeS4 lenses fared the best with the Zeiss Ultra Primes coming in a close second. The Zeiss Super speeds (35mm and 16mm format) and Zeiss T2.1 lenses performed almost just as well. The wider lenses presented more CA near the edges, but all of the longer focal lengths (16mm – 85mm) performed exceptionally well.
White Shading (green/magenta cast at top/bottom of image) is now very easy to calibrate with the custom white balance menu in the HD200 or HD250. Each lens series did require a slightly different white shading adjustment, but they were consistent. The Cooke S4 lenses consistently had the “truest” colour cast.
Zeiss Standard Speed 16mm – hardly any chromatic aberration on tree branches.
Depth of Field characteristics:
We spent the most time testing depth of field of every focal length and aperture. Depth of Field is a highly subjective concept of what portions front and rear of the focus are considered to be acceptably “sharp.” My primary goal was to determine if I could still use the data in the DoF charts in the ASC Manual with the HZ-CA13U.
32 The Samuelson DoF charts rendered enough data for me to determine that the CoC (0.0006”) and DoF characteristics of the aerial image seem to be equivalent to 16mm format. We also rolled film on the SR3 using the exact same lenses for comparison.
I then spent an afternoon comparing the frame grabs to my DoF charts and confirmed the promised capabilities of the adapter.
Cooke S4 18mm T2.8 (above) compared to Fujinon Zoom @ approximately 8mm(below).
What if I want an even shorter DoF?
As an added bonus, we also tried some Clairmont Anamorphic lenses. DoF is somewhat difficult to describe for anamorphic lenses because the circle of confusion becomes an ‘ellipse of confusion’ with separate values for the axes. Confused yet?
Don’t worry, for all intents and purposes let’s say the depth of field characteristics are the same with any given focal length of a spherical or anamorphic lens. However, since the anamorphics have such a wide horizontal angle of view, longer focal lengths can be used at the same focus distance, and hence shorter depth of field at the same horizontal angle of view.
Since most HD cameras capture 1.78:1 (16×9) frame, an anamorphic yields an ultra-wide aspect ratio. Left/right cropping would be necessary to bring it back down to a 2.40:1 ratio. The ProHD cameras do not have 1.2:1 frame guides, so I used the 4×3 frame guides to approximate the final frame.
Clairmont Anamorphic 50mm squeezed (left) and unsqueezed (right).
Comparison with 1/3″ video zoom:
After my initial tests, I borrowed a set of Standard speed T2.1 Zeiss lenses (35mm format) from Clairmont and spent a few days conducting practical tests. The first was to demonstrate the differences between a typical 1/3” video zoom and a PL prime mounted to the HZ-CA13U.
I used my custom built stereo-3D rig to shoot some side-by-side tests. As demonstrated in the photo, the physical length of the HZ-CA13U with a Zeiss prime is about the same length as the standard Fujinon 16×5.5 zoom lens!
A Fujinon 16×5.5 Zoom mounted next to the HZ-CA13U and Zeiss Prime.
As you can see in the comparison frame grabs below, there is a considerable difference in DoF between the 1/3” lens directly attached to the camera and a 16mm lens attached to the HZ-CA13U. This difference may not translate well onto the printed page here, but it is very evident on a HD monitor or projector.
I tested a 9.5mm, 12mm, 16mm, 24mm, 32mm and 50mm in this configuration.
Zeiss 16mm focal length
Assumed CoC for 16mm=0.0006”
Lens Focus: 6’
Near Distance: 4’ 7”
Far Distance: 8’ 8”
Total DoF: 4’ 1”
Fujinon video zoom lens set to 8mm
CoC for 1/3”=0.0011mm (0.00043”)
Lens Focus: 6’
Near Distance: 3′ 2″
Far Distance: 52′ 3”
Total DoF: 49′ 1”
I also tested some PL mount zoom lenses designed for 16mm. The zooms performed as expected. The big disadvantage with most cine zooms is that the minimum focus distance is usually quite long, sometimes as much as four feet. This limitation rules out extreme close-ups at short focal lengths. Video zoom lenses usually include a macro control for working within the minimum focal distance. The backfocus adjustment on the HZ-CA13U can be used the same way, to add macro capabilities to a cine zoom, or just simply decrease the minimum focus distance. Of course every lens requires experimentation, but I have found this feature to be very useful in a couple of situations.
The last phase of my test was to actually shoot something as I would with Super-16. My friend Nathan Fleet volunteered to be my subject. He’s a musician, so I decided to shoot a simple performance video for him. On my way to the shoot I stopped at my dealer to pick up some tape stock and ran into Richard Comely, the inventor of the Comely DV Crane. He carries a DV crane in his trunk at all times and he quickly volunteered to tag along for a couple hours.
We mounted the camera rig to the DV Crane, and set up the first shot with the 9.5mm lens. I didn’t have a means to power the HD monitor at the location so I attempted to backfocus with just the viewfinder. The method described early seemed to work well for all focal lengths, except for the wide shot with the 9.5mm. It looked sharp in the viewfinder, but was a little soft when I checked the dailies. Next time I’ll be sure to bring along the HD monitor!
JVC GY-HD250 + HZ-CA13U mounted to DVCrane (photos: Sabrina Armani).
I used the DV Crane for about four setups and then switched to handheld. I tried a bunch of different focal lengths, but I’d have to say that my favorite is the 16mm. It has a nice short DoF at 3 or 4 feet, T2.8, but is still wide enough to shoot a nice medium shot.
JVC GY-HD250 + HZ-CA13U mounted in a handheld configuration.
Frame grab from music video test shoot. 16mm Zeiss T2.1 & HZ-CA13U.
We shot for a total of three hours with just a two person crew and an iPod for playback. The system performed well, especially considering it was -17°C (1°F) with the windchill, which is well below the 0°C (32°F) suggested operating temperature of the HD250!
I shot all of the material in 720P60 for a 60fps overcranked effect. Nathan had sped up his song 250% for playback and everything held perfect sync when the footage was played back at 24P. I conducted a lot of rack focusing with the Zeiss lenses and, as expected, there was practically no evidence of lens breathing.
After watching some of the dailies something occurred to me: I had just shot a music video in 3 hours, in HD, with Cine lenses, no film stock, and only one miniDV tape! The 56 minutes of tape I shot at 60P turned into 145 minutes when slowed to 24P. I started crunching the numbers and realized that if I had shot the same amount of footage on Super-16 I would have required over 5000 feet of stock (13 x 400’ rolls). At $190/roll, stock would have cost about $2500, processing would have cost around $1000, and best-light teleciné transfer to HDCAM would have taken at least 4 hours at $500/hour + transfer tape stock.
The suggested HZ-CA13U price is $4399. That’s $1000 less than I would have blown in one afternoon on a music video shot on Super-16. Not bad at all! I’ve already decided to shoot my next music with this setup. Stay tuned…
Excerpts from this article were first printed in the April 2007 issue of Videography.
The following images taken from other HZ-CA13U projects
photographed by Tim Dashwood were added in July 2007.
Above: scenes from Elissa Mielke’s “Sing All Night” music video, photographed & directed by Tim Dashwood. Kinor 10-100 Zoom lens (PL converted) with back element net diffusion & HZ-CA13U. (Color Corrected and Enhanced with Apple’s Color.)
Above: scenes from “Lucky 7” directed by Steven Hayes. Cinematography by Tim Dashwood. Zeiss Standard Speed T2.1 primes (32mm shown) with HZ-CA13U and ½ Black Diffusion FX.
Above: scenes from Arian Gillis’ “Boom Dah” music video, photographed & directed by Tim Dashwood. Kinor 10-100 Zoom lens (PL converted) with 0.75X adapter & HZ-CA13U.