Saturday 7 December 2013

The “Filmic Look” and why DSLRs are not the best choice for shooting video.


It is amazing how easy it is for people to succumb to the power of the internet, blindly following word-of-mouth advice and information passed on by people who happen to be regurgitating the latest fad with no qualifying proof or evidence to support their argument.


Many years ago a few, mainly American, film-makers were anxious to save money by shooting on video rather than film but were disappointed by the overall look that video had at that time. If they could shoot video as progressive 24 frames per second and transfer the edited piece to film for distribution they could save a fortune in production costs and, at the same time, get a little closer to their Holy Grail, “the filmic look” fro a video camera.


Television, and thus video, in the U.S. is broadcast at 30 frames per second, with each frame made up of two interlaced fields. Film, on the other hand is shot at 24 complete frames per second and projected at 48 frames per second at the cinema. Each frame has to be projected twice to reduce the very annoying flicker inherent in a 24 frames per second film.


The camera manufacturers saw a gap in the market and duly obliged by producing video cameras that could shoot 24 progressive frames per second. Unfortunately, 24 fps shooting began to be adopted by less informed people who were not even transferring their video to film. They were under the misguided belief that they needed to do this to get a filmic look but all they succeeded in doing was to increase their post production costs with an expensive and unnecessary standards conversion. In most parts of the world television is broadcast at 25 frames per second, not 30 frames as in America and Japan. This is only one frame per second more than film, so the difference is insignificant. The difference, however, between progressive frames and interlaced fields is significant.


Shooting at 25 frames per second would be more in keeping with most national video and television standards and prevents the expensive conversion costs associated with shooting in 24P. However, do you really like jittery pictures when you pan the camera? Frame jitter is an artefact of film production that viewers find irritating and cinematographers would very much like to see the back of. Because television uses two interlaced fields to make a complete picture frame, such motion jitter is reduced substantially to the point of insignificance, but at the expense of clarity in the image.


One solution is to take progressive images off the camera's imaging device and record these in an interlaced form – 25P at 50i. This helps towards achieving resolution with less flicker and is compatible with most national television standards around the world but is still not perfect. An improvement on this would be to adopt a system of 50 progressive frames per second but, for the time being, broadcasting in the UK at least, is strictly 50 interlaced fields per second. Standard definition is PAL 720 x 576 anamorphic at 50i and HD is 1920 x 1080 at 50i. If you shoot or edit your production in any other format it will have to go through a conversion process that could suffer some quality loss issues, and will certainly add to the post-production budget.


The filmic look has several components, some good and some bad. Frame jitter on pans is one of the unwelcome artefacts due to shooting 24 progressive frames per second and should be avoided.


Dynamic range is one of the major desirable aspects that film has had over video and is one area where the technology has improved over recent years in a bid to match the contrast range offered by film.


Kodak Vision 3 film has a measured 14.5 stops of dynamic range although Kodak only claim it is 13 stops.
Your typical DSLR is rated at less than 9 stops which is close to CCD television cameras with their dynamic range of 8 or 9 f-stops. Many digital cinematography cameras using CMOS imagers are a little better at about 11 stops but some such as Arri and Red claim to be in the range of 13 to 18.


The main thing that differentiates film from video productions has been the restricted depth of field due to the 35mm (and 70mm) film formats used for feature films. The 16mm film format used for documentary work, however, has a similar depth of field as a 2/3 inch CCD broadcast television camera. Unfortunately, most peoples' perception of video cameras comes from the use of prosumer models with small imaging devices. Your typical domestic camcorder only has an 1/8 inch chip, whilst your average journalist camcorder has only 1/3 inch chips. These small chips give extended rather than shallow depth of field. Many years ago, the manufacturers used to make 2/3 inch cameras for broadcast, ½ inch cameras for industrial, (corporate), or medical use and 1/3 inch chips for consumer cameras. The television news channels chose to adopt “prosumer” camcorders as they were a fraction of the price of 2/3 inch shoulder mount cameras, were more portable, and could be used by journalists with little training. Educational establishments followed suite and ever since then our perception of the video image has been coloured by this extended depth of field inherent in small chip camcorders. Now, all the new kids on the block seem to think that you can only get shallow depth of field from a DSLR or digital cinema camera.


DSLRs will give you shallower depth of field than a 2/3 inch shoulder mount, but this can actually work against you. Many cinematographers have cried out for extended depth of field denied them by their 35mm format. In order to achieve extended depth of field they have had to stop the lens right down which meant having to use much more light and thus more power and the need for generators on location. The desire for shallow depth of field should not be an end in itself. I have seen many productions ruined by shallow depth of field. Corporate productions featuring “talking head” interviews are often marred because the shallow depth of field causes the subject to have one eye in focus and the other out of focus......or the tip of the nose is soft......or they constantly go in and out of focus as they move about on their seat. Throwing the background out of focus makes them appear anonymous whereas an in-focus background reveals more about them and their environment. In a corporate video, the message should be more important than the look. Only the cameraman and director appreciate the filmic look, the client and their customers are more interested in the content, including any information that can be conveyed in the background of the shot. I have also seen several indie film shoots marred by an inappropriate use of shallow depth-of-field and poor use of focus. All of these have been shot on DSLRs.


A common myth being propagated on the web is that all television programmes are now being shot on DSLRs. This is total rubbish. Yes, DSLRs have been experimented with on certain TV programmes, but the producers have always gone back to using proper broadcast cameras afterwards. There appears to be a virus spreading amongst film-makers that causes them to think that DSLRs are better than video cameras. Well here are few reasons why DSLRs are not the best tool in the drawer for film-making:


1 Maximum aperture is often much smaller than film and video lenses especially at long focal lengths. Most TV lenses are f2 throughout the focal length range.


2 Lens build quality is generally poorer than film and TV lenses


3 Aperture is more likely to ramp during zooming


4 Auto focus is too erratic compared to video cameras auto focus systems


5 Aperture changes in click stops and is not continually variable


6 Lens breathing. Focal length changes during focussing causing image size to change.


7 Zoom lenses are often not true zooms but are more accurately called vari-focals as the focus changes through the range. TV lenses have a back-focus or rear-flange adjustment which makes them maintain focus throughout the zoom range.


8 Most stock DSLR lenses are of a cheaper plastic build to enable fast auto focus servos. You have to spend a lot more money on decent lenses.


9 Many stock DSLR lens elements often jerk into place causing a jump in the image.


10 Lack of substantial large focus ring makes accurate and stable focusing harder.


11 Lack of tv zoom servo precludes nice slow zooms


12 Lack of film-style smooth damping mechanism on zoom rings makes manual zooming flaky


13 Physical lens sizes are enormous when compared to 16mm film, 2/3” broadcast, 1/2” professional or 1/3” prosumer cameras.


14 Lack of decent wide angle zoom drama lenses (a 2/3” 3x WA drama lens costs about £12k whereas there are no stills lenses to compete with this quality)


15 Lack of long telephoto zoom lenses for sports and wildlife only incredibly bulky fixed focal lengths


16 No built in 2x extender. TV lenses have this built in. No SLR lens has this.


17 No built in macro switch. Most TV lenses have a physically switchable macro function to enable focusing down to a few centimetres. With pro-sumer video cameras, this switch is electronic and enabled in the menu. DSLRs require extension tubes to be fitted for this functionality with greater light loss.


18 No built in ND filters. Need to buy separate filter sets for each lens or an expensive matte box on rails.


19 No auto-knee or DCC function to compress highlights forcing the need to use graduated filters in front of the lens.


20 Lack of decent tripod mount to prevent twisting


21 Range of tripods are more limited and lack the engineering of high end video tripods that enable smooth movements and accurate balancing cams.


22 Lack of mass and inertia make steady hand-held virtually impossible


23 Lack of decent focus aids such as continually variable peaking on CRT viewfinders


24 Lack of decent viewfinder makes focussing, framing and exposure difficult in live view mode, especially in adverse lighting conditions. You need to buy an expensive add on viewfinder loupe.


25 Lack of adjustable viewfinder on most DSLRs makes low angle and high angle shots difficult


26 Inability to ride focus easily when following a moving subject


27 Inability to adjust exposure easily whilst following a moving subject in variable light


26 No ability to switch smoothly from outdoor to indoor white balance on the fly


28 No decent audio recording forcing the reliance on dual system shooting methods


29 No decent on board audio pre-amps for audio recording


30 No pro-audio balanced XLR connectors to interface with quality pro mics


31 No phantom power capability for P48 mics


32 No decent headphone amplifier for audio monitoring


33 No decent audio level meters


34 AGC causes audio to breath up and down


35 No time code read or write for synching to other cameras or recorders


36 Incorrect aspect ratio of most DSLRs for tv requires cropping and re-sizing in post


37 Large CMOS imager causes “jelly vision” on fast pans


38 Bayer prism single chip imager causes moire patterning in fine detail. Pro video cameras have traditionally used 3 chips, one for each colour, R, G and B.


39 Cameras overheat on long takes so not suitable for shooting corporate events, live concerts etc


40 Restrictions that shallow depth of field imposes (good point - bad point).
Citizen Kane is regarded by many as the best film ever made, and one of the stipulations that Orson Welles made to cinematographer Gregg Toland was to make sure that everything was in focus within each scene. Welles hated shallow depth of field and thought that the viewer should be able to make up their own mind as to what part of the shot they should be looking at......much like in the theatre. They would only be looking in the wrong place if the direction was bad.


41 Dual system recording (separate sound and picture) slows down the edit process and increases production and post production costs.


42 The codecs used in DSLRs are not generally approved for broadcast HD television delivery.

43 DSLRs use 420 colour space not the 422 colour space required for broadcast. 422 colour space takes 2 red minus luminance (R-Y) and 2 blue minus luminance (B-Y) samples for every 4 luminance samples. 422 colour space gives superior results in chroma-keying, vfx and colour grading although high end HD will use 444 sampling. 420 is not allowed for HD broadcast as it isn't good enough quality.


People who promote using DSLRs seem to be totally mis-guided, obsessed with “the filmic look” and “shallow depth-of-field” at the expense of everything else, and continually spread dis-information such as..........”television programmes and feature films are all being made on DSLRs”, which is completely untrue. DSLRs are sometimes used for TV and film but they are certainly not replacing traditional equipment nor methods of working because their numerous disadvantages will continue to outweigh the advantages. If you want shallow depth of field, stick a DOF adapter on your video camera, or use a proper digital cinematography camera and real film lenses.

Where DSLRs are being used as viable alternative to professional video cameras, they are often accessorised with a lot of additional expensive add-ons in order to make them function like professional cameras.........shoulder mounts, better lenses, follow focus rigs, add on focus aids and viewfinders, microphones and mounts, audio amps with bodged agc defeat signals and third party software or firmware.  None of these, however, make them qualify for professional HDTV use and all tend to negate the cost savings made with the initial camera purchase.

If I'm shooting an indie film, I'd rather use a digital cinematography camera. If I'm shooting a documentary, sports, wildlife or corporate video, I'd rather use a 2/3 shoulder mount or a journalists camcorder an I would use a DSLR for some time-lapse, some music videos, some commercials, some underwater filming and some news journalism.