From the very beginning of television broadcasting, it has been necessary to 'televise' motion picture films. This apparently simple requirement has led to a wide variety of methods and specialised equipment - the telecine. Even Baird, in his intermediate film process, relied on a mechanical flying spot telecine to transmit the film shot a few minutes earlier.
Standard motion picture film is shot at a speed of 24 frames per second. In Europe the television system runs at 25 fps and it is convenient to reproduce films made for the cinema at this slightly increased speed. Film specifically for television is shot at 25 fps. Because each television frame is composed of two fields, each film frame has to be scanned twice. In North America the television frame rate is 30 fps. Running film shot at 24 fps at this speed is clearly not an option, instead it is run at 24 fps with frames being scanned twice and three times alternately.
This difference between the European and American television systems led to a divergence of telecine design. In this country the twin lens flying spot machine was the subject of intense development but since it could not cope with the American television system, development there concentrated on camera telecines.
Initially the requirement was for 35mm telecines. 16mm film came later, first for news gathering using small lightweight cameras and later, as quality improved, for general production for reasons of economy. Telecines remained as single gauge machines for many years, the 35mm and 16mm mechanisms having so little in common that a true dual gauge telecine was impracticable. Those described as dual gauge used separate mechanisms and shared only the electronics.
Although there are only three basic types of telecine, camera, flying spot and more recently line array, several variations have been proposed and developed. Some remained as ideas for many years until the technology was available to realise them. Only those which have played a significant part in BBC operations will be considered.
CAMERA TUBE TELECINES
This consists essentially of a television camera pointed at conventional film projector.
The flyback time between successive scans of the television system is very much less than the pull down time of a normal projector, but this is not a problem with a camera tube telecine since the tube is able to store the image. Early camera tube telecines used iconoscope and orthicon type tubes. The iconoscope tube did not exhibit any significant frame storage and a conventional intermittent motion film projector would have shown a prominent 'shutter bar', although there is some evidence that the original EMI (1936) method of transmitting films used a special projector that illuminated the Emitron camera tube for a brief period during the frame blanking interval. This would have been quite difficult with the technology of the time.
This is probably why the Mechau optical compensator, which is described later, was used for the early camera telecines; the Mechau, by dissolving from frame to frame, maintained a continuous light output.
In North America the inability to use twin lens telecines on their standard led to further development and eventually to the use of the vidicon tube.
The main advantages of camera telecines are simplicity, the ease of multiplexing (pointing the camera at more than one projector) and the ability to control the intensity of the light source for dealing with dense film.
In the days before colour, camera telecines were used extensively in TV News and in the regions, often in multiplex (several projectors into one camera) or crossfire (two projectors into two cameras with a mirror to effect a changeover) arrangements. Although colour versions were later used by TV News, elsewhere they were superseded by flying spot twin lens machines.
FLYING SPOT TELECINES
In the flying spot system a blank raster on a cathode ray tube is focused on to the film frame and the modulated light emerging is picked up by a photomultiplier tube. For colour operation the CRT phosphor must have an adequate light output over the visible spectrum. The emerging light is split into R, G and B components by dichroic mirrors and filters for the three photomultiplier tubes. Since this splitting occurs after the film is scanned there are no registration problems with the three colour components.
Unlike the camera tube telecine there is no storage of the image and if an intermittent mechanism were used it would have to pull down during the field blanking period. Although such a mechanism is just feasible for 16mm it is impracticable for 35mm since the film could not withstand the mechanical force necessary. Most flying spot telecines therefore use continuous film motion which incidentally decreases the risk of film damage. Several options exist for tracking the 'moving target' presented to the scanning system.
One solution is optical compensation where the film is 'chased' by an image of the raster so that there is no relative movement. There is no need to synchronise the film motion to the television system and such telecines can operate at frame speeds from zero up to the maximum for the particular mechanism. A feature of optically compensated telecines is that they are shutterless, hence each frame appears to dissolve into the next. Unfortunately they also have complicated and therefore rather inefficient optical systems which reduce the signal to noise ratio and the large number of glass to air surfaces causes problems with flare. Although considered here under the flying spot heading, they are also eminently suitable for camera telecines.
The earliest was the Mechau mirror drum which dates from 1912 and was originally used as a continuous motion cinema projector. Eight mirrors, arranged in a circle, are individually pivoted and acted on by cams as the drum rotates to chase the film image. It was used in conjunction with iconoscope type camera tubes in the late 1930s and in other camera tube telecines in the late 1940s. A flying spot version was developed and used at Riverside studios and at Lime Grove where the last machine was withdrawn from service at the end of 1966. As can be imagined, considerable skill was needed to align the mirrors and keep the mechanism in good order.
The most well known optical compensator is undoubtedly the polygon. A ray of light passing through a parallel sided block of glass can be made to shift laterally if the block is rotated. If the block is replaced by a polygonal prism, then a pair of opposite facets acts in the same manner. Each facet corresponds to one film frame and, with the motions of the polygon and film coupled, the image of the scanning raster "chases" the film exactly.
In the monochrome days Lime Grove was equipped with four 16mm polygon telecines. TV Centre had two, one of which was a valve colour channel. A significant use of the latter was for transferring topical material to North America because it could operate on the 525/60 standard at a film speed of 24 fps. Both of these machines were subsequently upgraded and converted to variable speed operation (from zero up to about 33 fps). TV Centre also had two 35mm polygon telecines which were later colourised and transferred to Lime Grove where they were often used for running silent films at the correct speed.
An alternative to optical compensation was offered by the Twin Lens telecine. In this type each film frame is scanned by two consecutive fields at two different positions in its travel. The two lenses provide two optical paths, one for each field, with a shutter allowing only one to be in use at a time. The continuous motion of the film contributes about half of the vertical scanning amplitude needed, the remainder coming from the raster which therefore has an aspect ratio of approximately 4:1½ instead of the normal 4:3.
Early 35mm twin lens machines made by Cintel and by EMI were moved from their original site at Alexandra Palace to Lime Grove where they were in regular use until made obsolete by the BBC1 change to colour in 1969. Television Centre was equipped mainly with Rank Cintel twin lens telecines from the outset; by the late 1960s there being twelve 35mm and five 16mm machines. Some were converted for colour operation in the days of valves and most were later upgraded to transistorised colour channels. Many of these survived a number of refurbishments and saw over 30 years service.
In the 35mm machine the film was driven entirely by sprockets with a mechanical filter smoothing out speed variations. The smaller image size and slower speed of 16mm film necessitated a slightly different approach and the 16mm machine was sprocket driven except for the gate where the drive was by continuous motion twin claws. As one claw approached the end of its travel the second claw entered the next perforation and then caught up on the first until it took over the drive. The first claw was then free to retract and repeat the process. The action of the claws was controlled by cams, the profiles of which had to be ground to very close tolerances.
In the late 1960s, the increasing use of colour film revealed the need to improve telecine control positions. The operator of a typical black & white telecine had faced a picture monitor and a waveform monitor built into a row of bays. In order to provide better and controlled viewing conditions for colour correction, some new 16mm telecines were modified by building the control position into a desk at 90 degrees to the bays. This proved successful and the concept was incorporated into an order for 18 Rank-Cintel 16mm telecines to be installed in London and the regions over the following years.
Hopping Patch / Jump Scan
By 1976 the central telecine area at Television Centre was almost fully equipped with machines, so when Rank-Cintel announced a completely new telecine called the Mk3* there was no immediate need for one; it was left to others to buy the first and to deal with any teething troubles. The first Mk3 in the BBC was installed at Television Centre as TK44 in 1977.
*[There had been two earlier generations of Rank-Cintel twin lens telecines although they had not been known as Mk1 and Mk2]
The new machine was significantly different from the familiar twin-lens. It used the hopping patch principle - two rasters and one lens instead of one raster and two lenses. The scanning patch is 'hopped' field by field to follow the continuous motion of the film frames. Mechanical complexity was reduced and replaced by more complex electronics. The machine looked similar to a then current (2" quad) VTR and the film ran from left to right instead of top to bottom as in a film projector. It offered the following advantages: Dual gauge (16/35 mm); Capstan drive (kinder to film than sprockets); High speed shuttle (twin-lens machines could only run at 25 fps); Coherent Still Frame (twin-lens display half the frame stretched out vertically when stopped); Multi-Standard 625/525. However, there were the following disadvantages: New sepmag sound followers were required (see later); Geometry Correction (to register the two rasters properly over the whole film frame); Shading and Burn Correction (to eliminate brightness flicker over the whole film frame.
Although Geometry, Shading and Burn corrections could be kept under control on 625/50 by judicious tweaking, they were particularly significant for 525/60 operation where hopping patch scanning requires five different field positions to be used - two for one film frame, three for the next. The positions overlap and produce a prominent fixed pattern where the scanning tube phosphor becomes darkened (burnt) with use. Geometry correction is more difficult because of the greater total scanned area. Errors in both geometry and brightness flicker repeat at 12 Hz - a very objectionable frequency. These problems in the 525/60 world led Rank-Cintel to develop the next type of flying spot telecine:
Digiscan avoids geometry and flicker problems by scanning each film frame sequentially and then re-assembling the lines into interlaced sequence in a digital store. To avoid burning the scanning tube phosphor by concentrating the beam power (about 5W) into a limited area, a small amount of frame scan is used; each film frame is scanned in about 30ms compared with the 40ms for a complete television picture. Approximately 3/4 of the vertical scan is provided by the film motion and 1/4 by CRT frame scan. The line frequency is 21kHz and the video processing which precedes the store has an extended bandwidth of 7MHz. Digiscan soon brought the advantages of twitter-free and flicker-free pictures to 625/50 as well as 525/60.
The Mk3 gradually took its place in the BBC in both hopping patch and Digiscan forms. The full still frame and rapid film handling proved important for programmed colour correction which was becoming more extensively used. The more adventurous use of film for dramatic effect, together with the presence of older twin-lens and hopping patch Mk3s alongside, showed up deficiencies in the contrast handling performance of early Digiscan Mk3s. Eventually, BBC engineers came up with a package of improvements for existing machines (dubbed 'Festival') and at about the same time Rank-Cintel introduced a new version called Digiscan-4 which, as well as producing cleaner pictures, provided a standard digital output and incorporated an additional store for a comparison picture (useful for colour grading).
Gradually eight dual-gauge Mk3 telecines replaced the 35mm and 16mm twin-lens machines, some of which had been in service for over thirty years, allowing the second floor of the central telecine area to be released for a new Network Transmission Area. Since the Mk3s had been bought over a period of about twelve years, they varied in detail and in features (such as Super-8, slide scanning, XY Zoom, Dolby Optical Stereo Sound and Cinetrace). They were all transferred to the Stage 5 Post-Production Centre at the end of 1991. The telecine operation in Stage 5 is mainly concerned with transferring film to videotape. An interface developed for the BBC by Digital Solutions allows the telecines to be controlled from an edit controller using the same serial control system as the Betacam and D3 VTRs. This has allowed the telecine to be fully integrated into the post-production process for manipulating and assembling film material into a finished programme, not simply as a means of 'televising' motion picture films.
LINE ARRAY TELECINES
In the 1980s, BBC Research Department carried out much pioneering work in using CCD line arrays for telecine scanning. Line array telecines have features similar to both Digiscan and camera tube types. The film is illuminated by a quartz halogen filament lamp and is imaged via a colour splitting block on to R, G and B line array CCD sensors. Horizontal scanning is provided by clocking the output of the CCDs into a digital store. All of the vertical scan is derived from the film motion. As with Digiscan, the store re-assembles the lines to provide an interlaced output. When the machine is stopped, there is no vertical scan and hence no sensible live output of the still frame. This makes programmed colour correction (see below) difficult unless the machine is designed with this requirement in mind. This lack of a live still frame output, together with the poor dynamic range of the early sensors, made them more suited to hands-off transmission of well graded film.
By 1993 machines had been produced by Rank-Cintel, Marconi, and Bosch-Fernseh (later BTS). Although they all had the advantages of lower running costs and simplicity compared with flying spot telecines, the BBC had not bought one by that time. Telecine operations had moved away from direct on-air transmission, and most film use required some form of programmed manipulation. Although improvements had been made to the dynamic range of the sensors and the manufacturers had introduced palliatives for making colour correction easier, they were still not as versatile as flying spot machines.
Picture Phasing and A-B pulses
Before the advent of video tape, the only way of making recordings of TV programmes was on film (film telerecording). Studio productions would often include filmed inserts from telecine. These inserts would sometimes show additional defects on the film recording at cuts between shots in the original film and at the edges of objects in motion. This was because there was no positive correlation between the TV fields and the film frames of both the telecine and the film recorder. The telecine might turn one film frame into fields 1 and 2 and the telerecorder make one film frame from fields 2 and 3. The recorded film frame would then be a mixture of two original film frames. Picture phasing was also found necessary when programmed cinemascope was introduced; a jump change in pan position must coincide with a picture cut on the film. It must not occur a field late if a hiatus is to be avoided. The discipline of picture phasing was therefore introduced such that telecines, film recorders and vision mixers all changed film frames and camera cuts during the same field blanking interval.
Telecine picture phasing of twin lens machines was done by letting the machine run up and then detecting whether it was in the correct phase; if it was not correct, the stop command would be issued briefly and the run command re-established such that the machine did a frame roll. This was most disconcerting and later a modification to the motor servo provided 'primary picture phasing' such that the machine ran up in phase from the start. The film frame position was originally provided by a magnet and reed switch sensor arrangement, later an optical detector was used. This provided two-phase, one cycle per frame, pulses at TTL levels, termed A-B pulses. The Cintel Mk3 machines generated similar pulses from the gate sprocket for the machine's capstan servo system and these were brought out as A-B pulses. These could be used for frame counting forwards and reverse. Several ancillary systems that needed to be synchronised with the film frames made use of the A-B pulses such as Autoset, Sound Followers (Sepmag), Programmed TARIF Controller (PTC), Cinemascope, Subtitles, A-B roll working, Film Post Production Area (FPPA), and VT edit controllers.
In addition to the telecine machine itself, additional equipment is essential for operational use:
Reason for needing it - most film produced for television has its sound track prepared in a dubbing theatre on fully coated magnetic 16mm or 35mm film. There is no need to go to the expense (or time) of producing a print with a combined optical or magnetic sound track. Therefore each telecine has a sepmag sound follower to replay the dubbed sound track.
The first sepmag sound followers used Selsyn drive which required a 3-phase supply. There was a complex locking arrangement (using relays) to align the rotors of the master and slave before applying power. Since the mechanical power to drive the sepmag follower was supplied from the master (the telecine) the system usually incorporated 'aiding' (running the selsyns partially as motors). Reversing direction required making sure the system had stopped before reversing the 3 phase supply. The film was sprocket driven requiring flywheels & mechanical smoothing.
It was with a sense of relief that we adopted an alternative locking system for the regional 16mm twin lens telecines:
Stepping Motor Drive These machines (Sondor) are virtually self-contained, requiring only a single phase supply and a simple optical sender from the telecine. They have mechanical smoothing, but the flywheels are spun up by an auxiliary motor when in standby to give rapid stabilisation on starting. They cannot follow at high speed very well and, being single gauge, were not really a suitable companion for a Mk3 telecine.
When the Mk3 telecine was launched there was no sound follower available which matched it in all respects. Most were single gauge, some could not follow at high speed and used a counter to store 'dropped frames until they could catch up, only one used capstan drive. Fortunately several manufacturers had suitable machines under development and after many engineering and operational trials the Perfectone 'Capermag' was chosen to accompany most of the BBC's Mk3 telecines.
It was realised in the early days of colour television that some means of correcting the colour balance would be necessary, hence TARIF. There is some dispute as to what this acronym stands for but ‘Technical Apparatus for the Rectification of Indifferent Film’ is a common version. In addition to the master lift and gain which affect the red, green and blue channels equally, there were controls for colour gain (picture whites) and colour gamma (mid and dark tones) with 24 position switches to select the colour axes and rotary controls for the amount of correction. It was expected that an average correction could be found for a reel of film on rehearsal and no further adjustment would be necessary. This soon proved to be unduly optimistic and an alternative means of control was produced in the form of a pair of joysticks. The left hand joystick controlled master lift by rotation and colour gamma by direction and amount of movement from the centre position; the right hand one adjusted master gain and colour gain. This was found to be the most satisfactory operational arrangement and persists in the BBC to this day. Later commercial systems often had three joysticks, one each for the lift, gamma and gain functions but these have never found favour in the BBC, possibly due to the lack of three handed operators.
Although joystick control was much more versatile there was still a problem when large changes in colour correction were needed between consecutive shots. This was particularly the case with colour negative which enjoyed a period of popularity in the early 1970s. A means of storing the corrections found for each shot on rehearsal and applying them on transmission was required, hence the Designs Department Programmed Tarif Controller (PTC) came into being. This used punched paper tape for storage - what else in 1971? The punch was preceded by several buffer stores, the information being advanced to the next store at each shot change, until finally committed to holes in the paper tape. If the telecine was reversed, the information was moved back through the buffer stores. This allowed the operator to stop the film during rehearsal, run back a few shots and have a second try without spoiling the tape - running the punch backwards doesn't fill in the holes! On replay, the corrections that were arrived at during each shot were applied at the beginning of the shot thus making the operation appear seamless. If the required corrections were difficult to find ‘on the fly’, the machine could be stopped and the correction applied to a still frame. Twin lens machines gave a vertically stretched picture when stationary but Mk3 machines gave a full frame.
The need for this facility was such that the prototype PTC was immediately in service for transmissions of Panorama, which regularly used colour negative film at that time.
Punched paper tape storage was of course superseded by full semiconductor storage with floppy disk back up. Originally, shot changes had to identified manually before grading the film. This boring and time consuming task was replaced by an automatic shot change detector invented by Research Department and subsequently re-engineered by Designs Department. The shot change detector made it possible for programmed colour correction to be applied to any film when necessary rather than being a special operation.
Numerous commercial controllers followed with varying degrees of success. The one which has now been in use for some years is a version of Digigrade which was designed to meet BBC requirements Proqrammed colour correction has now become the norm and, instead of a few PTCs on trollies being shared around, each Post-Production telecine has one built in to the installation.
In the Cinemascope system the image on the film is compressed horizontally by the use of an anamorphic camera lens so that more of the scene width is recorded. A projector lens with the inverse effect decompresses the image. Systems of this type are generally referred to as Cinemascope although there are, in fact, a number of them with aspect ratios from 2:1 to 2.55:1. Transmission of such films on a television system with an aspect ratio of 4:3 clearly presents a problem!
If the full width of the film image is transmitted there will be blank bands at the top and bottom of the picture amounting to almost half the screen area and much of the impact of the programme is lost. This is not acceptable to the average viewer except perhaps for short clips or title sequences.
The alternative is to fill the television screen height but, because the Cinemascope frame is too wide, almost half of the information on it is lost. A manual pan control can be used to select the area scanned but, its practical use is limited since the operator is unable to see the whole picture and cannot therefore decide on the appropriate area for transmission. It was occasionally used for short clips, often with amusing results.
Clearly a pre-programmable system was needed where the film could be viewed in full on an editing bench adapted to show the possible pan positions and an instruction list could then be prepared. Such a system was being developed by the late 1960s.
For a short time before the system was automated, control of the telecine was by means of a box which had a row of 9 buttons corresponding to pan positions and a rotary switch to determine the time the pan should take, zero being a cut. These controls were duplicated, one set being available for setting up while the other was active. The changeover between controls was initiated by the detection of a metal cue dot fixed to the edge of the film. Setting the controls needed considerable concentration especially when cues occurred in quick succession.
Fortunately the instruction list was very soon transferred to punched paper tape which held the information for a whole reel of film. Over the subsequent years many enhancements were made; the number of available pan positions was increased to 15 and the pan profile was given an S characteristic which removed the jerks at the start and end.
By the mid 1980s a new microprocessor based system called Cinetrace was being developed. Many improvements were made especially in the preparation equipment where the editor can now simulate the required effect on a monitor with an electronic cursor and store the instructions at the touch of a button. Additional features brought about by a combination of Cinetrace and the Mk3 telecines include an increase in the number of pan positions, to the extent that the choice appears to be continuous, and a zoom facility. Zoom can be used for a continuous transition from full letterbox presentation (for titles) to the normal full height picture, or simply for trimming so that extraneous items can be excluded from shot.
The addition of subtitles to foreign films was, for some years, achieved by making up a separate film containing the titles and running it in synchronism with the main film. The two signals were combined, usually by means of a video switch which also inserted a black edge around the characters to improve legibility. This was an expensive exercise as, apart from the title film cost, it doubled the number of staff and equipment needed for transmission.
By about 1975 alternatives were being investigated and a decision was made to move to a fully electronic system. Minicomputers and 8" floppy disc drives were available almost off the shelf, the character generators were specially made for the BBC, the interface hardware and computer program were dealt with by Designs Department and the installation was carried out by SCPD, as it was then called.
For that time the character generators were very advanced, much attention having been paid to profiling the rise and fall times of character strokes to eliminate 'stepping' on diagonals. A great deal of work also went into digitising the chosen font so as to reproduce it accurately on the television screen.
The minicomputer memory consisted of 8k of core store (tiny ferrite rings with fine wire laced through the holes). Although semiconductor RAM was in production it was in very short supply and in any case the core store had the advantage of being non-volatile.
The text for each title, together with on and off times and position on the screen, was stored on the floppy disc. When replaying, the computer read the information from the disc and passed the text of the next title to the character generator. It also compared the running time received from the telecine with the on and off times and sent the appropriate command to the character generator when coincidence was found.
Two sets of equipment - minicomputer, dual disc drive and character generator - were installed as main and reserve systems. For transmissions both systems would be active so that in the event of a failure the reserve could be immediately switched to line. Fortunately this proved to be a very rare occurrence.
This facility rapidly gained popularity and was soon being used for a variety of programmes. Not all were on film and the VT area had to be given access to the system.
With minor modifications the system was also able to originate subtitles for the deaf for Ceefax.
This equipment remained in service for about ten years. With the advance in digital technology the next generation was much more compact and offered two fonts in two sizes and in colour. It is, however, a tribute to the work of ten years earlier that the original font could not be bettered and was retained