Three companies are challenging therecent wisdom of top-end digitalcamera design: Kinetta, Dalsa andArri. They take the view that themajor camera manufacturers have

a much greater economic interest inthe TV market rather than the filmmarket, and this has led to the creationof high definition cameras for cinemathat, in essence, are upgraded TVcameras. They believe cinematographersare better served by an approach todigital cameras that take its inspirationfrom traditional film cameras.

Here, in their own (sometimescontroversial) words, Jeff Kreines ofKinetta, John Coghill of Dalsa and MarcShipman-Mueller of Arri explain theirreasoning and their approach.

Jeff Kreines, designer of the Kinettadigital camera: At the recent GenevaAuto Show, Volvo showed a newconcept car, designed by women, forwomen. Most of the press treated thiswith bemused condescension – theBBC said, “if the Calendar Girls weremaking cars, this is what they wouldlook like”. But shouldn’t this approachbe dreadfully obvious? Pay attention tothe people you are making cars for!

The same, of course, could be saidof the manufacturers of HD/digitalcinema cameras. Look at the SonyF900 or the Panasonic Varicam andwhat do you see? An overgrown TVnews camera. While the currentgeneration of TV news cameras mightbe good for… well, shooting TV news,this doesn’t necessarily makes themsuitable for making films.

There are so many reasons why. The ergonomics are terrible. Way

too many controls, most of them in the

wrong places. If you want to use amatte box or follow focus, you can’treach any of the controls on the frontof the HDCam, as they’re obscured bythe rods. If your sound recordist needsto see the audio level meters on thecamera, they can’t, because your headis in the way.

There are way too many switchesand connectors. Every inch of thesecameras seems to be covered in dreck.Use all the connectors at once and youend up with a Medusa. I pity the poortechs that devote their lives to buildingcustom snakes for these messy rigs.

And before you can shoot, youhave to tweak all sorts of menus,setting colour matrix, gamma,sharpness, and more – because if youdon’t, you may not have enoughlatitude in post to colour correct. That’sbecause the recorded image is not onlyheavily compressed, it’s filtered downto 1440x1080 – and with 3:1:1 coloursampling in the case of the F900.

This leads to the disaster of trying to

set up a de-facto colour-correctionstation in the field, with accuratemonitoring – because it’s important tocapture an image that is as close to finalas possible.

Philips (or is it Thomson, or GVG,or – they change their name so often)had the right idea with the Viper –output a raw 4:4:4 image with greatdynamic range, and treat it as a ‘digitalnegative’. Colour correct in post. Likefilm. Capture it uncompressed to ahard-disk recorder.

Great idea. But they are dependingon third parties to make fieldrecorders, which, if the Director’sFiend is any example, make for anunwieldy field setup. Not justphysically… these hard-disk recordersare computer-based – meaning thatyou have to boot them up and dealwith operating systems. Just what Iwant to do – entrust my film (er, data)to Microsoft. Blue screen of deathroulette, here we come!

The CineRam does offer a solution

digital movie cameras

If DoPs designed cameras. . .

issue 4 summer 2004 | showreel 69

Geoff Boyle solicits the views of three manufacturers who areapproaching digital cameras from a different angle – they’re

emulating traditional film cameras, rather than TV cameras.

Jeff Kreines shows offthe Kinetta digitalcamera at NAB.Photo: Mark Forman.

to some of these problems, but it’s notdesigned for long takes in the fieldwithout frequent downloading to acomputer. And the S.two field recorderis a big improvement over the clumsierDVS and Director’s Friend devices (seeChoosing a Camera for DI, page 75).

Of course, the Viper looks like avideo camera, and uses the dreaded B4lens mount. It’s not the mount that’sdreaded; it’s the block of glass behind it.Sony pioneered the B4 mount, and indoing so spec’d a special glass for theprism that focuses the red image a fewmicrons behind the blue and greenimages – meaning that all lensesdesigned for this mount must also throwthe red image out by 10 microns.

Big deal, you say. Let the lensmanufacturers deal with this. Theyhave. Zeiss makes the marvellousDigiprimes. Wonderful lenses. But a setof them costs about US$115,000.That’s a lot more than a set of S16Zeiss Superspeeds, which also havepretty decent optics.

Even with great lenses, prismcameras tend to have problems withchromatic aberrations – ever notice theblue stripe on the right side of anycontrasty image captured with anHDCam? It’s subtle, but once you’veseen it, you’ll never be able to ignore it.

Of course, you won’t see it in themediocre viewfinders that these HDcameras come with. They’re low-res,flickery monochrome CRTs. Of course,there are better aftermarket finders,from Sony and Accuscene, but they’renot cheap – prices start at $10,000.

The housings of these cameras allfeel cheap – lots of plastic and lots ofpressed tin. Not solid, like a filmcamera. Some rental companies, suchas Clairmont Camera, have had toruggedise these cameras so they canstand the rigours of rental use.

I could go on. But I won’t. I’ve beenshooting film since the late 60s, and Ilike the way film cameras look and feeland, most importantly, work. Some ofmy favourite film cameras have onlytwo switches – on/off and frames-per-second – and one connector, forpower. You have to think a bit whenyou use these cameras, but you don’thave to fight the camera’s personality inorder to use it. Kind of like shooting

with a Leica rangefinder camera. Haveyou ever struggled to shoot a simplesnapshot with your little pocket digitalstill camera? In the attempt to designthem to be idiot-proof, manufacturersdesigned them to be unfriendly toserious users.

So, what can you do about thismiserable state of affairs? Well, you canlook to the past for inspiration.

In 1914, Carl Akeley couldn’t find asuitable camera for shooting his Africanexpeditions, so he built his own – andthe Pancake Akeley remained popularfor over 30 years. In the mid-60s,Jean-Pierre Beauviala wanted to make afilm about architecture – a film thatrequired several cameras in differentlocations running in absolute sync. Heasked Éclair, and they said it wasimpossible, because at the time therewere no commercially available crystal-controlled motors. Not to be deterred,Jean-Pierre designed and built one. Andtook it to Éclair, who hired him.Although his film never was made,camera history was. After Éclair died asad death at the hands of HarrySaltzman, Beauviala founded Aaton,and has made many wonderfulcameras – because he paid greatattention to the needs of users.

Inspired by Akeley, Beauviala, andothers, we chose to make a digitalcinema camera that completely ignoredthe past 30 years of TV-news-cameradesign. We instead looked to 100 yearsof film camera design for inspiration.I’m lucky to be working with MartinSnashall, a pioneer in digital videodesign (he designed the Abekas A64disk recorder and the A84 vision mixer)– it’s amazing what he can do with afew Xilinx FPGAs – and Glenn Smith,who collaborates with us onmechanical design.

One great thing about film camerasis that they get a free upgradewhenever Kodak or Fuji introduces anew emulsion. We designed theKinetta HD camera to be sensor-agnostic – the camera’s architecturewill support future CCD and CMOSsensors with resolutions up to 16megapixels (over 4K). Just replace thesensor and its daughterboard – verysimple. No planned obsolescence.

Film cameras are completely

variable speed – they can even behand-cranked. There’s a hand-crankoption for the Kinetta camera as well –and you can change just framerate orframerate and exposure, like a filmcamera. Of course, you can also doconventional speed ramps and timelapse.

We record RAW data – 4:4:4, 10-bit log, uncompressed – to a magazinethat contains 12x40Gb hard drives –the same drives that are used in iPods.Tiny and rugged. Each magazine,recording in a parity-protected RAID-3format, will store 110 minutes of1920x1080 material at 24p. Themagazine can be used on the camera,for shoulder-mounted use, or it can beseparated from the camera head by athin 3mm fibre-optic cable by over akilometre. Combined, the camera,magazine and battery weigh about6.5Kg. You can attach a magazine to anon-linear editing system via single ordual HD-SDI links and digitise, or youcan export data directly as DXF files ifyoud prefer.

Since we record RAW data –everything the sensor puts out, werecord, unaltered – all you have to dois make sure you don’t overexpose anypixels you want detail in – and you cantweak the rest in post. The camera hasmultiple user-defined zebras – you cando the zone system if you’d like – anda histogram. All are available in thefinder when desired.

We use a standard PL lens mountand a single sensor – so all of yourfavourite film lenses can be used. Noprisms. Our high-resolution colourviewfinder is OLED-based – it’scompact and consumes very littlepower. It can be positioned anywhereon the camera (or a few feet away),and includes controls for finder zoomand various display options.

Note that Akeley and Éclair andAaton (and even Arriflex) are tinycompanies when compared to, say,Sony. Sony is interested in HD notbecause they really care about sellingcameras to this tiny market, butbecause it might help in their eventualdesire to sell lots of HDCAMs to TVstations for news shooting. Ah, nowwe see why these cameras really dolook and feel like TV news cameras!”

cinematography 2004

70 showreel | issue 4 summer 2004

John Coghill of Dalsa, designers of theOrigin digital camera: In 2001, whenwe embarked on the Origin digitalcamera programme, we had twofundamental questions facing us as adevelopment team:

1 How are we going to translate therequirements of what is essentiallyan artistic community intoengineering terms we can effectivelyimplement, and,

2 How are we going to overcomethe potential handicap of not havinga history of video engineeringexperience that the large consumerelectronics companies were bringingto bear on this market?

The answer to the first questionwas fairly obvious. When you don’tknow what you don’t know, go asksomeone who at least knows whatquestions to ask. So very early in thepaper design stage we involved peoplewith experience in the development offilm cameras and working withfilmmakers and asked them to educateus as to the core needs of theprofessional filmmaker. This processbegan by taking the core engineeringteam to a local (Toronto) camera rentalshop and spending the day pouringover the various camera configurationsand gaining better insights into theimportance of the myriad accessories,their use and how they interact withthe set and crew. We studied bookssuch as Every Frame a Rembrandt byLaszlo Kovacs, the AmericanCinematographer Manual and theStudent Filmmakers Handbook byKodak, and asked lots of questions ofthe people who would eventually beusing the camera as their primary tool:the cinematographers. The list of‘credits’ is too long to reproduce here,but suffice it to say there were manypeople who contributed to the shapingof our final thoughts on the designapproach.

We learned very quickly thatfilmmakers are not only creative in theirmind’s eye, but also with the physicalmedia in terms of modifications toaccessories, cameras, rigs, lighting andthe final processing of the film. Itbecame apparent that there are only

three constants that are important tothe filmmaker and everything elseneeds to be flexible. The three keyelements around which a professionaldigital motion picture camera wouldhave to be designed are: existing highquality 35mm lenses, a highperformance image capture media, anda through-the-lens viewing system witha safe framing area. We also learnedabout two other key tools required onthe set: Velcro and duct tape!

At some point along this process itbecame apparent that our secondconcern was not the handicap we hadimagined. In fact, the requirement tohave an ultimately flexible platform thatprovided a wide range of features andfunctions (some yet to be dreamt upon a set somewhere) required thedevelopment of what is essentially ahigh performance electro-optical imagecapture module with an imageprocessing computer hanging off theback. What was a perceived weaknessactually turned out to be a strength,because the design team was notconstrained in their thinking by pastvideo camera design approaches.

The enabling technology in thecamera platform is the large area, high-speed CCD sensor, and it satisfied oneof the three key requirements, namelythe high performance image capturemedia. From this starting point there

were three options to consider interms of the front-end opticalconfiguration: single, dual and threechip optical assemblies. The three-chipapproach was dismissed immediatelybased on its incompatibility with existing35mm lenses, not to mention theoptical aberrations inherent in prism-based optical assemblies. The singleand dual chip approaches each havetheir advantages and disadvantages, soboth were carried forward along thedesign path, as both are opticallycompatible with existing 35mm lenses.The final decision to pursue the singlechip architecture was based on manyfactors, but the underlyingconsideration was that it is the onlyoption that allowed for the rotatingmirror shutter and reflex viewingsystem to be implemented exactly as inprofessional 35mm motion picture filmcameras. Based on the number ofsingle-chip camera designs announcedsince we went public with our design inOctober 2002, it would appear othermanufacturers have come to the sameconclusions we have.

From this base, the designproceeded with two guiding principles.Keep the design modular and flexibleand, when faced with a choice, alwaysgo with the option that preservesimage quality or enhances featureperformance in some way. The result

digital movie cameras

issue 4 summer 2004 | showreel 71

The Dalsa Origincamera, a single-chipmodel designed toemulate the way filmcameras work.

of this design philosophy brought us toanother decision point that required usto seek out the opinions and input ofcinematographers once again. Thevolume of electronics in the cameraand the requirement to locate certaincircuits in a particular physicalrelationship to others to maintainperformance dictated one of twopossible physical profiles: a long lowprofile (nicknamed ‘the bazooka’) or ataller profile. The opinions and reasonsfor preferring one over the other weresplit almost 50:50 among the peoplewe spoke with, and we clearly wentwith the taller profile that you now seein the Origin camera. This was notsimply a toss of the coin, however. Wehad 15 different industrial design artists’conception drawings made and it wasagreed that all technical argumentsaside, there was an appeal to a formthat was reminiscent of a 35mm filmcamera with a magazine mounted ontop. Based on the reactions so far, theopinions are still split 50:50.

Going forward from here, thecurrent implementation of the Origincamera will exist for a long time as thefull featured ‘studio-cam’ model, butwe continually get input from acrossthe industry that is being folded intoour product roadmap. As we introducenew models in the future I hopepeople will recognise the influence theyhave had on the various designs.However, regardless of the shape andfeatures of new models down theroad, the fundamentals will remain thesame – use the highest quality optical

input to a system that enhances thecreative control and options available tothe cinematographer.

Marc Shipman-Mueller of Arri,designers of the prototype D-20digital camera: I am currently in NewYork conducting D-20 focus groups, aspart of a larger international tour togather more information on how theD-20 can better serve the productioncommunity.

The film industry is a unique market,with tremendous demands on theequipment. Not only must it be robustenough to work in the Arctic and theAmazon alike, but it must also be soeasy to use that a completely exhaustedassistant after a 16 hour day in thepouring rain can still operate it withoutmaking mistakes. To make mattersworse, cameras are complex assembliesof precision mechanics, optics andsoftware, and to design and manufacturethem for the above requirements is notrivial task. Since in the grand scheme ofthings the film industry is a relativelysmall niche market, few companies findit worth their while building productsspecifically for this market.

Some companies have specialised inproviding cameras to this marketthough, including Aaton, Panavision andARRI, and those companies maintainclose ties with the cinematographersand assistants using their cameras in thefield, and have leveraged theseconnections in new product design.

The D-20 concept is a goodexample of this. It was shaped by

countless conversations withcinematographers, producers,directors and their crews. In theseconversations we were told that thecurrent HD cameras are still designedlike electronic news gathering (ENG)cameras, and that those are not thebest tools for productions thattraditionally have been shot on film.Film cameras, on the other hand, havebeen designed and refined for almost acentury specifically for single cameranarrative productions. As a result ofthe input from the filmmakingcommunity, the D-20 is a digitalcamera specifically developed to fitseamlessly into the process used byfilm productions. A crew that is usedto working with 16 or 35mm filmshould be able to use the samemethods with the D-20.Cinematographers who have honedtheir skills over decades should be ableto apply them similarly to the D-20,thus leveraging their experience, whilegaining the immediacy of outputting ahigh definition digital signal.

The D-20 concept came frominformal conversations with filmprofessionals. Then the functionaldemonstrator (a fancy name for a labmodel) was shown at the IBC 2003show, sparking an avalanche offeedback and many lively discussions onand offline. As part of the process tofurther refine the D-20 specifications,we are now showing the D-20functional demonstrator in variousfocus groups internationally, leading tothe creation of prototypes which willthen be used in pilot productions. Thefeedback from those productions willlead to the creation of the product.Developing a product for us is areiterative process with continuousinvolvement by working professionals;a continuous process of presentingideas, concepts, models and prototypesto customers and continually refiningthe specifications based on this input,until we have a product that works inthe production process just like ourother film cameras.

Geoff Boyle: I’m really looking forwardto working with all three of thesecameras; finally, digital imaging has beenfreed from the tyranny of TV.

72 showreel | issue 4 summer 2004

Arri’s D-20 prototypedigital camera, firstshown at IBC 2003.

cinematography 2004