In 1918 William Kelley, subsequently the inventor of Kelleycolor, writing in the Transactions of the Society of Motion Picture Engineers, complained that he had several times been attracted to a cinema where colour films were advertised only to find that the film was coloured black and white. He was not the first to want to distinguish between "real" colour films and tinted, toned or stencilled black and white films, which were far more common at that time.
Kelley proposed that films be called "natural colour" for the first and "coloured" for what he described as "films arbitrarily coloured by dyes.... to suit the individual taste".
The problem still exists today and we often still use Kelley's terms, in English, to distinguish between coloured films and natural colour films.
Natural or photographic colour film systems can be defined as those that have an analysis and a synthesis stage, and in which the reproduction attempts to stimulate the human eye in a manner similar to that of the original scene.
From about 1910 to 1920 "natural colour" films were rare, and many of the devised systems were made in the USA, and never seen elsewhere. The principles of the additive and the subtractive systems were well known before 1900, and the analysis procedure using three cameras and therefore three film strips, or using one camera with one film with three sequential images was well understood, and had been outlined by Du Hauron in the 1880's.
In Europe and the USA, in the first 20 years of the decade, there were numerous demonstrations of natural colour film in cinemas, but most used cumbersome multiple projection, or rotating filters in sequence with the synthesis frames. While it is clear that some of these demonstrations were extremely dramatic, effective and the results were occasionally of the very highest quality, a fact that is not always appreciated today, none of them were practical solutions that could be applied to cinemas, in all their infinite variety, throughout the world.
By 1920 it was recognised, by the most realistic inventors that the essential requirement of a commercial colour system was that the positive should be projected on a conventional unmodified 35mm projector, and this could only be achieved by subtractive prints. Nevertheless many bizarre and complex projection methods still made their debut, only to be almost instantly rejected as too complex, too costly or uncontrollable. A number of systems were devised that were compromises with genuine natural colour and used only two primaries instead of three. Additive mosaic systems [for example "lenticular" Kodacolor and Dufay] made several appearances in the 1920's and 40's, but their low screen brightness, mosaic image and complex printing made them obviously impractical.
The search for a practical colour print system was the basis of the majority of the inventions tried out until 1935 when Technicolor Inc., an American company that had tried out a number of alternative processes since 1917 [including 2-colour and all called Technicolor], finally introduced the 3 colour imbibition or dye transfer print system we think of today as Technicolor. By 1950 Technicolor was dominant in colour print production, and although the principle of the Technicolor print system was largely unchanged for 30 years, the camera films and camera techniques varied quite widely. There were competitors to Technicolor, which was not alone in the market at any time, and although, to the general public, all colour prints were thought of as Technicolor, other systems also existed especially in the 1940’s.
Some of the competition to Technicolor was 2-colour film made by exposing in the camera two separation films [sometimes in the form of bipacks]. The print was made on "duplitized" i.e. double coated, film toned blue-green on one side and orange-red on the other. Eastman, Agfa and DuPont made films for this process, but the laboratories gave them trade names such as Magnacolor and Cinecolor.
Technicolor prints were distributed widely after about 1936, but largely these were American or, to a lesser extent, European feature films. Technicolor imbibition prints were made in Los Angeles from 1936 to 1973, New York from 1936 to 1973, London from 1938 to 1973, and Rome from 1950 to 1973. Many countries making their own films could not afford Technicolor prints, nor was it often practical to travel to a Technicolor laboratory to have one made. Technicolor was a process that was costly for the production of just a few prints and it was often said by the mid 1960's that making less than 25 prints was so expensive that any other method was to be preferred.
The negative-positive colour systems first appeared in 1935 [Agfacolor] but it was not until Eastman produced the first masked negative in the early 1950's that this system became effective. The early colour negatives, especially up to 1960, used dyes that have faded considerably and these films represent some of the most difficult material of all to restore.
In a world view colour in the cinema was very patchy until the mid 1960's. "Natural colour" was a very rare event until 1935, whereas "coloured" films were probably very common everywhere until 1930. From 1930 to 1935 the cinema was almost entirely black and white everywhere in the world, with just a handful of "spectaculars" being made in the USA and Europe. From 1935 to about the mid 1960's colour gradually increased in the developed world, but only from the 1960's onwards did colour extend world wide with the advent of the negative-positive systems especially Eastman. Although the first negative-positive incorporated coupler colour films [Agfacolor] were on the market in 1935 it was only from about 1950 that Eastman began to dominate the colour market, and an important aspect of negative film success is the use of integral masking originating from coloured couplers.
In many major producing countries [such as India, USSR, Pakistan and Turkey] which could not afford to pay for colour film stocks in dollars or sterling, colour only became common with the advent of cheaper colour print materials from East Germany [Orwo] and the USSR [Sov] in the early 1970's. China imported the Technicolor print system in the 1970’s but until that time almost all Chinese film was black and white.
Although all cinema and most television film is negative-positive, reversal colour films have played a major part in the historical changes. Kodachrome in 1936 had a major impact on 16mm documentary and amateur use and in 35mm was, for a while, the camera original film for Technicolor prints. Later reversal films played important roles in documentary production and news film.
Colour film systems fall into a number of broad categories each of which directs the restorer to a specific approach, while a smaller number are not easily categorised.
The following are examples of the main categories - the first three used separate films to produce separation records that were synthesised in different ways. Technicolor became solely a printing method by 1955, but began life as a series of complete colour systems.
1. Kinemacolor type: 2 or 3 Colour analysis/additive synthesis/2 or 3 colour additive projection.
In the silent period most colour systems were additive. Exposing three separate frames through three coloured filters made separate records of each colour, red, green and blue. When projected, each positive image (containing the information of a single colour) was projected back through the original coloured filter. On the screen an image reproducing natural colours was obtained as a result of the addition of the three colours.
In some cases, in order to simplify the process, only two colours were used (orange-red and blue-green) which resulted in an imperfect colour reproduction but probably still quite striking in the cinema. The two-colour systems derived from this principle have mostly left us with negatives and print in black and white. These are only recognisable due to the fact that each frame is slightly different in rendering one from the other [for example, a red flag appears to be transparent in the red frame and opaque in the green and blue ones], or from the unusual geometry of some of these inventions [in some the two separations were inversions of each other, since one was exposed through the base, and the other not].
Several different systems based on this principle were developed. Gaumont Chronochrome, developed in France by Gaumont, was a system using three different frames (red, green and blue) taken at the same time with three different lenses placed one under the other. Behind each one of the three lenses was placed a filter of the corresponding colour. The film was projected in the same way, using a projector with three lenses. Chronochrome frames were shorter than those of standard 35 mm films and had only three perforations per side.
The Kinemacolor system, developed commercially by the Englishman, Charles Urban, was designed to shoot two different frames through a rotating filter that alternated between blue-green and orange. The film (standard 35 mm) ran at a speed of 32 frames per second and the rapid sequence of the pictures in the two different colours produced a plausible, even if imperfect, colour effect.
Another English inventor, Friese-Green, had developed a system similar to that of the Kinemacolor, but which was based on shooting of three successive pictures through three filters.
2. Nature Colour/Cinecolor type: 2 colour analysis on 2 separate films/ subtractive print on double coated toned film.
Typically these films were exposed in a special camera designed to handle two films in the same gate, or two films exposed to the same image by a beam splitter. The two films, often sold as "bipack" by several manufacturers, produced separation images corresponding two the different sensitivities of the films or filters used were records of blue or blue green and orange, the 2-colour primaries. The resulting separations were printed onto double coated or "duplitized" [USA term] print film, one separation on each side, and the silver images toned or dyed blue-green one side and orange the other. There were innumerable variations in dyeing, film sensitivities and camera mechanisms, some of which became defined by trade names. The first use was probably about 1917 and the last about 1948! One of the early Technicolor systems fell into this category. The first Technicolor, called Technicolor bipack or bichrome, was introduced before 1920. It was based on two negatives exposed through red and green filters, which were printed on two black and white films, then toned in cyan and magenta and finally glued together. This system was used in several films, starting with The Toll of the Sea in 1922.
3. `Technicolor 1935 type: 2 or 3 colour analysis on 1 to 3 separate films/ print by toning, dyeing or dye transfer
These all commence with separate separation records but transfer the image onto a common support by some form of image or physical dye transfer. In 1928 Technicolor abandoned one method - of two dyed positive films glued together, and transferred attention to the dye-transfer system, still as a two-colour system.
In 1932 the Technicolor "tripack" or "trichrome", commencing at first with animated cartoon shorts, later with a few medium length films and finally a feature, Becky Sharp, by Rouben Mamoulian.
Dye-transfer is the principle of the later Technicolor, from 1935, and is very similar to a wash-off process for paper printing called Dye Transfer. From the three negatives (red, green and blue) three matrices are made, that is, three positives in black and white. The silver image development also tans the gelatin and prevents it washing-off in hot water. The remaining relief image absorbs cyan, magenta and yellow dye, in proportion to the original density. The dye is transferred onto a "blank" film with each image in register.
Sequential exposure animation was a term given to exposures in a normal camera in a sequence through red, green and blue filters to produce a singles strip of negative with sequential separation negatives. Once made a subtractive print could be made by a wide range of techniques - Technicolor, onto duplitized film, Cinecolor, Gasparcolor, or onto a modern tripack colour intermediate. Many animation films were made this way, and the process is only applicable to animation. Some films were issued first in one print system and re-released later in another. [N.B. Disney were always filmed on Technicolor 3-Strip]
4. Dufay type: 3 colour analysis on single film in discrete areas / additive projection via original filter mosaic or optics.
Several additive systems involved the use of a mosaic [termed a reseau in Dufay] of red, green and blue filter patches on the emulsion surface or in the lens system, exposing discrete separation images relating to each patch. In Dufay the patches were lines and squares of filter on the film emulsion, in lenticular Kodacolor the lines were created from the lens system. Reversal processing and projection through the original or similar filter patches yielded a colour image not unlike colour television. Negative positive systems existed as well although printing was complex in order to avoid interference patterns from the filter mosaics.
5. Agfacolor/Kodachrome/Eastman Colour type: 3 colour analysis on integral tripack film/ subtractive synthesis on integral tripack film
The incorporated coupler colour films using an integral tripack were first put onto the market in Germany in 1935 following the release of the very first tripack, Kodachrome, in 1935. The early motion picture materials of this type were reversal. In reversal/reversal duplication or printing from colour prints, the lack of an integral mask produces colour reproduction problems beyond one stage of duplication. Fourth or fifth generation reversal duplicates are of very poor quality for this reason, whereas fourth generation duplicates using masked original and intermediate materials still return good colour quality.
Although Agfa released the first colour negative in 1935, only in the late 1950’s did a successful negative-positive system emerge, from Eastman Kodak. Much of the success of the Eastman system was due to the integral mask on the negative.
Most other manufacturers followed the Eastman lead and Agfa, Gevaert, Fuji, Sov, and Orwo all produced similar materials.
6. Integral tripack analysis/Technicolor print type: 3 colour analysis on tripack film/ print by subtractive dye transfer
Technicolor was widely used as a print system until the 1970s, and in China until the 1990’s. Even after the advent of Eastman Colour, the three Technicolor plants (Los Angeles, London and Rome) continued to produce positive copies, some produced from 35mm Kodachrome, and later three matrices were made from Eastman Colour negatives.
Although Technicolor has probably been the only print system to use integral tripack film as the starting point, any of the previous printing systems, from duplitized-toned film to Gasparcolor, could do the same
Technicolor was a primarily a cheap production method for printing, but the copies of the later films confuse archivists with credits that say "Print by Technicolor" and also "Made in Eastman Colour" and finally it is better to inspect the film and establish from it's material exactly how it was made.
Almost all manufacturers of colour film code marks can be found on the edges of Technicolor film copies: Kodak, Ferrari, Agfa, Gevaert, DuPont, etc. These indications refer to the manufacturers of the black and white film or blank film that was used for the production of the Technicolor prints and not to the colour system.
7."Dye Destruction" colour printing systems
Gasparcolor was one of several "dye destruction" direct positive printing systems that used a tripack film with three layers sensitive to red, green and blue that already incorporated cyan, magenta and yellow dyes. A positive image was exposed onto the film. The negative silver image produced by development was used to initiate the destruction of the coated dyes, resulting in a subtractive colour positive. Gasparcolor is not unique, there were a number of patents issued for these processes, and Cibachrome is a modern still colour paper print material. Although in theory the process could have been used for a camera original film the dyes coated in the emulsion reduce the photographic speed to very low levels.
Although the dyes used in modern integral tripacks are called Cyan, Magenta and Yellow, they are not perfect, and this leads to some problems. When one film image is duplicated onto another, the imperfections in the dyes compared with the ideal subtractive primary colours, characterised as unwanted absorptions, are duplicated, reducing the colour saturation of the final prints. As multiple generations are produced so the problems of colour saturation are increased. This problem can partially resolved in reversal materials only by complex chemical adjustments to dye formation that takes place in development called inter-image effects. In negative positive systems almost complete correction is possible in the negative stage by the use of integral masking.
The unwanted green and blue absorptions of the cyan dye image print through onto the green and blue sensitive layers of the duplicating stock. This is equivalent to a small pink impurity in the cyan dye image.
The unwanted blue absorptions of the magenta dye image prints through onto the blue sensitive layer of the duplicating stock. This is equivalent to a small yellow impurity in the magenta dye image.
The yellow dye image is satisfactory and needs no change.
These impurities vary in density with the density variations of their own dye images, so producing actual print through images in the duplicating stock.
If these impurities of pink and yellow in the dye images are balanced exactly with amount of pink and yellow in the unexposed parts of the emulsion to produce an overall pink/yellow cast to the film, but no print through images, then the pink/yellow, (or orange) cast can be filtered out in printing, to produce correct colour balanced duplicates, without print through images from the unwanted dye absorptions.
To achieve this the colour couplers in the cyan and magenta dye layers, which are used to form the dye images in exposed and developed image areas, are themselves coloured pink and yellow respectively. Once a coupler becomes part of the dye image, it loses its own original colour.
Thus, in areas with no dye formation, there is a strong orange mask colour present, and in areas where cyan and magenta dyes form, the orange mask disappears to balance out the unwanted dye absorptions now present. The result is known as an Integral mask and is used in all colour negative and intermediate film stocks, to retain good colour reproduction through multiple duplication stages.
Integral masking has made restoration simpler, since once a colour image is recorded on a masked film several integral masked stages of duplication can be undertaken without excessive loss of colour quality. Duplication using unmasked films loses some of the original hues and contrasts in just a single generation.
All dyes used in photography fade with time. Many of the dyes selected for early additive filters are reasonably stable but those dyes developed in some way during processing fade to considerable extents. The dyes used for toning, Prussian blue, uranium Ferro cyanide and the basic dyes used for the two colour processes like Cinecolor and the numerous duplitized print processes, all fade although Prussian blue and perhaps uranium Ferro cyanide tend to darken and desaturate until almost no colour remains but the image is darker than the original.
Incorporated coupler tripack dyes
All dyes fade in time but some are far more prone to fading than others. Many Magenta dyes used in integral tripack films are azomethines and are very stable, whereas many Yellows and Cyans fade easily. The degree of fading of integral tripack dyes is directly related to the original quantity present and generally the dyes fade to colourless rather than change colour. The visual effect is therefore to reduce the maximum density of the fading dye and reduce its visual contrast resulting in a change in colour balance and the effect known as crossed curves. Many films go Magenta, as the magenta dye is least or not at all effected.
This proportional fading provides possible mechanisms for correcting the effect.
It is helpful to find out the degree of fading in a film. This can be done with a densitometer providing there is in existence a piece of unfaded film of the same stock and period. Select images of maximum density and compare the R, G, and B densities on the assumption that the faded film originally had similar maxima to the unfaded. This technique is useful and easy to apply to some faded prints of the 1960’s and 70’s since fading of this period is often due to poor stabiliser formaldehyde concentrations and therefore unfaded, well processed film of the same period is usually easy to locate.
In the case of many print films that are pale and strongly visually magenta another opportunity for measuring the degree of fading exists. It can be assumed that the magenta has not faded at all [this is never absolutely true but still useful as a premise].
Select a maximum density area and measure the Status A values. It can be assumed that the original densities were more or less equal or visually neutral.
In above case the percentage fading of a cyan dye can be calculated from the formulae
% Fading of Cyan dye = D green - D red x100
D green
Other dyes
The effect of fading tripack dyes is rather different from the fading of metallic toning dyes of early toned films, and 2-colour duplitized films, that sometimes became denser or less saturated or alter hue with time, often in a patchy, uneven way as fading may be dependant on ultraviolet light from projection. No percentage fading measurements are really useful in these instances. It is often necessary to make an internegative or a separation to work with that does not have a record of the patchy or uneven nature of the dye. In some cases this is not possible, but some films can be recorded onto black and white for subsequent Desmetcolor or recreating tinting and toning by using a filter to eliminate or reduce the unevenness in the duplicate image. The filter should be selected by trial and error but it should be of the same or similar colour to the dye colour, and the film recorded on panchromatic film stock.
Identifying the colour system
There have been a great many colour film systems since 1900 and such is the variety of techniques used, their restoration cannot be carried out by a single coherent duplication technique. Successful restoration almost always requires identification of the original colour process as the first step, but there are, undoubtedly, many pieces of film in archives that have not been identified, and many specialists believe that just as many are incorrectly identified.
A major problem is that the literature on colour film systems is now generally in patents, which requires time, great patience and application to both locate and interpret. Some systems were never completely patented and many systems changed with time, sometimes quite abruptly, without good records. Some systems were used for just a single film production before being lost or altered, and in the case of Technicolor in the early 1930's the technology changed repeatedly from film to film but the name of the process stayed the same, as it was the name of the operating company!
Edge data is sometimes of great value in identifying the system, but although much data can be acquired from the edge of a film it is not easy to refer this back to a system. Many archives have extensive knowledge of systems local to them but find it difficult to identify unfamiliar film. No doubt an eventual sharing of information will overcome this problem.
Restoring archive colour film by photographic duplication is not a single procedure that can be described and defined. The number of different processes and procedures used in old colour systems is considerable and each requires a different restoration process. The element available in the archive may be a black and white intermediate or a colour intermediate consisting of dyes, and each will require a different route.
Reviving original techniques for restoration.
A catalogue of restoration techniques can be treated as just that: a list of techniques, and this is how restoration has been viewed to a large extent in the past. The decision as to which technique or procedure to adopt was considered to be dependant on the element available, on identifying the system, and on the purpose for which the restoration is intended.
This approach is still and will remain, the procedure used for a lot of restoration, although interest is being expressed at reviving some of the original techniques used to produce the images in the first place. There are some tentative plans [1995] to revive the old imbibition process for making Technicolor prints and this has been seen as an opportunity to remake copies of old prints. This might be successful a way of recreating these old images but it should be recognised that new commercial development will almost certainly produce quite different process to the old one in order to compete with modern film stocks. It seems more likely that some small-scale laboratory reconstruction of an old process would be more relevant.
In many cases it is quite simply impractical to consider resurrecting old processes. Such a thing may be possible with some of the early additive processes or some of the early subtractive processes that used simple three-strip analysis and imbibition or stripping films, but it is unthinkable for the majority of systems that involved complex multilayer film coatings or elaborate camera equipment. It would also be inconceivable that Dufay film could be made again in order to print and preserve Dufay negatives.
Copying a print
While identification of the process is desirable in every restoration case, a frequent occurrence is the need to copy a colour projection print, exactly as it comes out of the can, and in these circumstances it may be justifiable to make a colour copy without understanding how the process worked or what the original film looked like. It must be clearly understood that such a copy is of limited value.
Many colour films exist in archives as projection prints, and a smaller proportion as intermediate elements or camera originals. A third alternative [particularly in the case of Technicolor or Eastman Colour] is where the film exists as some form of "protection master", made to store in the event of damage to the original material or fading of the dyes with time. In all cases it is essential to know how the print, intermediate, or protection master would have been made in order to plan a successful restoration programme.
The purpose of colour film restoration
Restoring black and white film results in the production of a new print as much like the old as possible and the approach requires little decision making other than deciding what contrast or grading is needed to restore the original appearance. The resulting restored image is silver just as the original image was, and although modern materials do differ in response to older materials, in general there need be little difference between old images on old film stocks and new ones on new stocks.
Colour film requires more consideration as the dyes in use today in modern subtractive colour films are, in almost every case, different from those originally used, and in some cases modern subtractive dyes cannot be used to produce a match with some of the dyes used in the past [particularly the primaries of the various additive processes].
Almost all modern restoration techniques for archive colour film use modern film materials in a manner never used originally or ever intended by the film stock manufacturer. A good example is to consider the production today of a colour print from a set of three separation negatives made in 1940 for the production of a Technicolor print. The usual procedure is to print the separations onto black and white stock to produce a set of separation positives, print these in register through a set of tricolour [red, green and blue] filters onto Eastman Colour Intermediate Film and print this masked colour negative onto modern colour print stock. The result can be very good, but may bear little resemblance to a Technicolor print with its low resolution and narrow range of saturations. The print may be visually more like a modern negative-positive print than an original imbibition print. Using modern materials usually results in restorations that are a visual blend of characteristics of the old image and the new film material.
Since there is a choice of routes in many instances, it is necessary to establish what is the purpose of the restoration. The alternatives purposes are usually
1Long term conservation of an image that is deteriorating,
2Making a print that can be screened without risk of damage to the original.
A screening copy could be
a) copy of the present image, of the faded dyes
b) The print as it would or might have appeared when first made. Sometimes a print made as it might have originally appeared is called a "simulation" of the original, and do this successfully usually requires a considerable knowledge of the original process.
If the purpose of restoration is primarily to preserve the image in a reliably permanent manner that allows later reconstruction and/ or display, then the method most in use is to generate red, green, blue monochrome separation negatives or positives on a modern silver image material that are independent of the permanence of any dye images. Sometimes polyester base is used but most archives use modern acetate based film stocks. These separation images record the colour of images as they are today, after any fading that might have occurred.
If the elements available are already monochrome separations that need to be restored then high quality monochrome duplication as described in a previous chapter is required. Other systems have also been proposed to do this, such as the use of digital video signals, on disc or tape, but the permanence of a silver film image on a modern film base is still considered more certain than the unknown life of any video system hardware.
Many colour systems in the past included the use of black and white separations as "protection masters", sometimes called "PM's" or "pro-masters", and these were particularly used by major film producers for storage. However they are not a single consistent format or element and their contrasts depended on the originals from which they were made. PMs made from Technicolor separation camera negatives are positives of a high contrast [almost that of a conventional projection print] whereas PMs made from Eastman Colour Negatives [also positives] were of much lower contrast.
Once a set of PMs existed a printing negative or negatives could be made from them at any time.
In the Technicolor process the PMs were printed onto a black and white duplicating stock to make a colour separation negatives. From these the matrices used for the imbibition process could be produced.
PMs from an Eastman Colour Negative could be printed onto Eastman Colour Intermediate through tricolour red, green and blue filters in register to produce a duplicate colour negative. Once these two principle routes are established in a laboratory and set ups exist so that they can be performed repeatably they constitute the basis of almost all colour separation restoration.
If the purpose is to display the image to an audience in the form it is today, for example a Technicolor print, incorporating it's fading or other changes with time, then two alternatives exist.
a] Photographic Reproduction
The image may be copied using a modern colour film [and, if necessary, a modern format] to produce a "good visual match". Generally this is not difficult and the only difficulty usually involves the control of contrast. Faded dyes usually result in low contrast images that are difficult to grade. The most common materials used for this purpose are Eastman Colour Internegative, Eastman Colour Intermediate, or Eastman Colour Negative [depending on the contrast needed] to make a negative, which is then printed onto a modern colour print film.
Contrast of colour materials is difficult to change by changing the process conditions although it is regularly attempted. Pre-flashing is a technique that allows the inherent contrast of a stock to be reduced. This procedure is dealt with in the chapter on Duplication and mentioned again below.
Throughout the world laboratories have experimented with using various masked negative films for the non-standard purpose of making internegatives from projection prints, either pre-flashed or unaltered. There is no doubt that some films respond to flashing more uniformly than others, and some are capable of heavy flashing to lower contrasts by as much as 15%. In 1995 Fuji Colour Negative 64 was used extensively for this reason but it must be emphasised that the manufacturers do not test or promote their films for this purpose and within a matter of months this sort of advice can become incorrect as the batches of film stock alter and with it their ability to carry out these non-standard practices.
Some additive systems [Dufay is a good example] cannot ever yield a matching copy on any subtractive colour film as subtractive colour reproduction is inherently unable to match the range of saturated colours possible by additive colour reproduction and is particularly unable to reproduce the saturation of the additive primaries.
b] Reproduction on Video tape
The image may be transferred to videotape on a telecine machine to be replayed on a TV monitor, transmitted or video projected. Producing a visual near match is not difficult although the overall visual impression may be very different from a projected image. A major problem with all telecine transfer is that operators are used to adjusting the visual image to a "modern" acceptable result, frequently very different from that of the original material. However this is largely a matter of training and management control. The archivist may have to sit with the telecine operator to define what he or she wants.
The greatest problems arise if the purpose is to produce a restoration that attempts to retrieve the original visual appearance, as seen by the cinema audience when the film was first made.
To achieve this in a reliable and believable manner it is essential to identify the system.
Only by identifying the original dyes, filters or tones is any correct simulation ever possible.
It is an essential part of restoration that the final result will probably not have the same visual appearance as archive print material because the faded dyes will be corrected. It is therefore essential to know what these dyes are and to know what they were like when freshly formed. High saturations and sometimes garish colours existed, which will surprise many archivists. Two colours used extensively in two-colour films of the 1930's and 40's were Prussian Blue [sometimes called Iron-tone Blue] and an orange-red made from a mordanted dye or from Uranium Ferro cyanide. Today these dyes look dim by comparison with the same dyes freshly made, and to simulate the original appearance it is necessary to match to the original hues and saturations rather than the faded colours we see on old film today.
It is inevitable that the methods described in these papers are simply a list of techniques. The technician will need to select from them in order to attempt any particular restoration and make calculations based on the contrast rule, choose the most relevant contrast control technique, ensure the straight-line sections of characteristic curves are used, and generally adhere to all the rules of good duplication practice.
Separation methods from coloured originals - two-colour subtractive films
Subtractive two-colour films [Cinecolor, bi-packs and DuPacs etc.] of the period from 1920 to 1950 pose a special problem. Firstly they are generally metallic colorants or basic mordant dyes, quite unlike incorporated coupler colour films, and they fade by darkening and/or changing colour. This poses real problems if it is not known what the original primary colours were or looked like. Some restorers ignore this aspect and make three RGB separations from which an intermediate colour negative [on Eastman Colour Intermediate] can be made. This copies the faded original, using modern three-colour technology, but allows some variation in saturation to be made by varying the contrast of the final colour print. As the forensic study of early colour films improves the original colours of these two-colour primaries will be known - where they are, they are considerably more saturated than originally believed - and better methods will become available. One technique is to make two separations to record just the two original dyes. These separations are then printed one after the other [like A rolls and B rolls onto modern colour print film each exposed to produce a colour corresponding to the original two colour primaries. This is an extension of the Desmetcolor method used for simulating tinting and toning, and is dealt with in the paper on 2-colour subtractive print processes.
Simulations from additive mosaic systems
The additive processes most difficult to simulate or to copy, and retain the screen character of the original. The Dufay process images reversal originals or negative-positives, using a reseau of red, green and blue areas each providing a tiny separation record of that part of the scene, are very difficult to copy. The dyes appear to be reasonably stable and seem not to have faded in up to sixty years. Eastman Internegative is often used to make a colour internegative and thence a colour print and this is usually disappointing to anyone who has seen an original projected. The originals, even Dufay prints from Dufay negative are dark and of a very low screen brightness but very saturated. Modern subtractive dyes have difficulty in achieving the high saturations of the original reseau, and perhaps all subtractive systems, per se, will have this problem. Certainly a modern telecine unit can translate a Dufay image into a far better video representation of the original than any film.
The best approach for all mosaic or lenticular additive colour systems is to copy them onto reversal print film stock such as 16mm Ektachrome Print Film. The higher that usual contrast compensates for some of the lost saturation and the black areas have a higher density that colour print film.
Many mosaic and lenticular and ruled screen systems were used to make documentary film of relatively important events in the 1920’s and early 1930’s and probably were quite dramatic records of those events in their original format. Some were copied to make more easily distributed versions, onto whatever "colour" system was available, with the result that archives contain some extremely confusing prints. British Realita ruled screen film, and Dufay was printed onto two-colour print film such as DuPont Duplicoat or Eastman Duplitised Film or onto Kodachrome print film, even onto Technicolor, with the result that the original is almost unrecognisable.
In the British Pathe collection newsreels were made from many film origins and important images reused in later issues. Sometimes the edge printing comes through - Raycol can be read on the edge of a two-colour print coated with orange dye on one side and blue-cyan on the other and a bright blue, variable density, sound track [presumably a Cinecolor print or similar], or Dufay can be seen on the edge of a early unmasked Gevacolor negative. More often the origin is simply obscure, and some so unsharp that the reseau or ruled lines are gone.
Simulations of additive projection systems
The most difficult simulations to achieve are the early additive projection systems that are now often restored by printing onto modern colour print film via modern colour negatives or intermediates.
The originals were either three separate projectors, or the sequential projection of three separate images through red green or blue filters, or were two records treated in this manner.
The filters were either over the camera lens or were dyes applied to the film, by conventional tinting. Very few technicians or archivists have ever seen original material projected this way, although, in principle, it is not difficult to do. The early colour filters were at least as good as those produced today - the modern Kodak Wratten tricolour filters used for original photography as well as for the projection, Wratten 25, 58 and 47B are the same today as those made in 1910.
A modern additive system is relatively easy to produce along similar lines to Kinemacolor or
The sequential frame systems, but a still additive image is very easy. Three still negatives are made of a scene, on panchromatic film, through the three tricolour filters, and prints are made. These are projected in register on a single screen and the three filters put over the lens. The relative projector brightnesses may need adjusting using neutral density filters, but providing the contrasts of the three separations are similar a very high quality colour record is produced. This suggests that the primary problem of early additive colour was not the quality of the image but the complexity of the equipment needed to project synchronously and in register the three images. In the case of sequential images this had to be done at three times the normal projection speed to reduce the flicker.
Unfortunately many modern simulations are lacking in the colour quality that must have been present in the early additive systems, probably due to several different effects.
1The sequential frame systems were often tinted frames and these dyes fade - copying on modern colour internegative film simply copies the faded colours.
2Red green and blue filter primaries were very saturated and even ideal subtractive dyes cannot encompass these saturations.
3The subtractive dyes used in modern films have notably poor cyans, which are darker and less saturated than they should be and magentas that are darker and bluer than they should be.
4Above all, our expectation of the quality of these early films is probably too low, and we fail to recognise the degree of correction we need to apply.
In early 1997 when this publication was finished almost all film was restored by creating a new film image of the original. The established principle of preserving both the film image and the film format has been a guiding principle for many years and has been successful in many archives. This has occasionally lead to some unfortunate situations where some archives transferred nitrate film to safety film and then destroyed the original film regardless of it’s actual condition, and regardless of whether the restoration was the best that could be done.
Few archives transferred everything they could and are now left with just the really difficult but invaluable material that could not reasonably be restored by known photographic methods. Those in that happy position now have to wait for technology to provide a new approach to their problems, nearly all of which are either cases of serious physical damage, or [and mostly] colour films so faded that they cannot be restored photographically.
The rest of the archive world are in much deeper trouble; they simply cannot copy their existing collection quickly enough, and even if money were no object some film archives are so large that there are no laboratories capable of transferring the film, nitrate or acetate, onto modern film stocks before nitrate decay or acetate vinegar syndrome overtakes it. And, of course, money is the major problem, forcing archives to consider these other options.
1Separating the collections into material where the image is the prime consideration, and which could be kept as video quality tape for access, - and other film, where the image and the format are relevant, and where photographic image resolution is important.
2Considering non-photographic image storage at various quality levels, from video up to 35mm projection resolution for the entire collection. This has centred the discussion on "digital masters", but the exact nature of this master is problematic - conventional video standards, analogue or digital have a poor record for archival permanence.
Departing from film, if only for some of an archive’s collection, means an increasing dependence on equipment, as video records and digital masters need equipment to translate them to images. In television the life of a video record is only as long as the life of the playing equipment, and considerable less than any film system.
On a practical level, a digital master at video resolution [720 pixels/line by 625 lines] is a small box with three hours of programme, but at 35mm resolution [4000 pixels/line by 3000 lines] is a full storeroom of many black boxes costing £0.5M and holding only 40mins!
Transfer of film images to video
The traditional transfer of film to broadcast or lower quality tape is a relatively straightforward process, once the film is cleaned and repaired, just as it needs to be for film restoration. Modern telecine machinery is continuous motion capstan friction drive and can handle more severely shrunken film than any unmodified printer. In practice, telecine operators are generally unused to correcting archive film and need retraining, otherwise they use whatever technology on their desk to create an image as close to a modern view of image technology as they can. This is not serious in black and white film but is ethically unsound, and certainly not wanted for early colour and coloured film. Telecine operations need new set-up procedures for these films and especially for tinted and toned material. A telecine operator meeting a combined tinted and toned print for the first time is quite unable to proceed without strict guidance and training.
The very flexibility of the telecine system, with it’s extreme range of contrast, masking, programmable scratch and defect elimination, and colour balance controls, does enable broadcast quality restorations to be made of film that cannot be restored by film methods, to film quality, without huge effort, skill and cost. This is particularly true of faded colour prints and early unmasked colour negatives. It is unfortunate that many films remain in archives unrest red, even unseen, because restoration to film is too costly or not practical, when a broadcast quality tape could be made with all the general quality restored except the resolution.
Digital restoration
From digital television technology comes what is being called Digital Restoration. This is the process of scanning film to produce a digital record of the images, stored on frame store, disc or tape. At this stage the image data is in the form of a digital master, although not yet in a practical medium that can be stored conveniently.
Using a computer workstation the image data is manipulated, much as a telecine operator might correct for fading, contrast, dust images and scratches, although at a far higher resolution. Several existing softwares [Cineon, Domino, Matador] exist for this already, although none are directed at archive film restoration, but at special effects for the cinema instead.
Once corrected the data is then re-recorded back onto film in one of several devices, at varying resolutions. The highest resolutions possible are approximately equivalent to that of modern Eastman 35mm film, although many archive films have considerably lower resolutions.
Cost is critical, and at present, because the scanning rate, and especially the re-record rate, is so slow, no archive can justify the expense for the restoration of complete film lengths. An affordable re-recorder [the Solitaire III costing about £150,000] takes 40 secs to re-record one frame of 35mm colour, and the fastest re-recorder [Cineon Lightning] takes 10 secs per frame. In London in January 1997 the basic cost of scanning a frame of 35mm film into framestore, and re-recording it back onto film again is £10. This precludes the restoration of faded films where the problem is found throughout the entire length, and only short damaged sections currently justify this type of restoration.
Digital restoration is undoubtedly here to stay, and as the process becomes more productive, it is only a matter of time before the film archive’s budgets and the costs of the service and equipment meet.
Ethics and future restoration methods
In the meantime the archives and restoration technologists must consider ethical problems in preparation for these new technologies. As a consequence of the increased flexibility and range of final achievable results, a major effort is needed for technical film historians to understand and research what the early cinema images really looked like, in order to brief the restorer.
Until now restoration has been a process of using whatever modern film stock the film stock manufacturer currently supplied for modern production purposes, in order to copy an old image. Most archives expect a restored image to look somewhat like the original. When transferred to a new safety film base almost all restorations look "better" than the original simply because the new film base is clearer, less opaque or less "cloudy", than the original.
Lack of precision on the part of archives in specifying what is an acceptable restoration does not matter a great deal when the range of possible results is limited by the narrow range of photographic routes and therefore the final results.
Relatively few colour restorations of faded images have restored the image to it’s original appearance [although sometimes the restorer believes that this is achieved], and there is also some small but effective resistance in some areas of the academic world that objects to the "cleaning up" of old cinema film images. Most archivists are aware of the objection felt by some traditionalists to the removal of all, even some, scratches, which are sometimes seen as a "patina" produced by time, much as it is seen on a fine piece of antique furniture. This syndrome even has an emotive subtitle - "the romantic scratch"! Modern film restoration and wet gate printing can remove most scratches if required, a modern telecine can remove almost every scratch on a broadcast quality transfer, and digital restoration can remove absolutely every scratch on a film resolution restoration. Should we be doing so?
While there seems little doubt that in the early cinema scratches were a regular feature of the projected image, we are only just beginning to understand the true nature of many of the colours seen in the early cinema. For a generation or two some archives have ignored the colours of tinted and toned films, on the grounds that coloured film was not an intrinsic part of early cinema, when actually, it seems, up to 80% of all cinema film was coloured until 1930. Many coloured films were copied in archives onto black and white stock without recording the original colours.
The problem seems to have been a lack of interest in the forensic aspect of cinema film, in distinct contrast to the restorer’s and conservationist’s attitude to the fine arts; considerably more is known about ceramic colorants, oil based paints or medieval wall paints, than about quite recent cinema dyes.
One of the best examples is the metallic colour Prussian Blue, used extensively as a toning colour on toned films until 1930, and as a primary colour on both 2-colour and 3-colour film, until 1950. Almost all old Prussian blue images today are darker, almost black or neutral, a dense "navy blue", while a few seem to have been lost entirely in the centre of the frame, presumably due to the irradiation from projector lamps. Yet Prussian blue produced today, in place of the silver image, using a recipe from 1910, is a clear brilliant saturated cyan-blue, startling and vivid and quite unlike archive images. We should not be surprised at this. Unlike restoring a painting most restorations of film are carried by producing a new image rather than restoring the original artefact, and while this is might be unfortunate, it is necessary, but it could allow the restorer to aim for a "simulation" of the original cinema experience, rather than a copy of the tired image left today.
In practice only a few photographic restoration techniques allow this possibility. Some of these are:
- colour restorations using separations to correct for fading, tedious, costly and slow, and requiring an experienced technician
- the flashed dupe mask methods, now impractical for 35mm film because there are no 35mm colour reversal processes any more
- the Desmet colour method for simulating tinting and toning. Once the technician realises the colour required, for example the saturated colour of unfaded Prussian blue that can be produced, rather than a copy of the faded film colour. This technique also has the potential for restoring other types of original.
There are few photographic restoration techniques that can achieve Desmetcolor precision or control, but digital restoration has the flexibility to cope with all the possibilities.
Digital restoration will bring to film restoration technology a flexibility so great that it will be essential to set guidelines to avoid the restoration becoming an irrelevant image with little relationship to the original. At present few conservationists, archivists or historians have enough knowledge of the original photographic systems to be able to set these limits or prepare a technical specification for digital restoration of any pre-incorporated coupler colour film, or any tinted or toned film.
The problem becomes really serious when the restoration method is capable of such great control and variation, and both convention telecine transfer, and especially digital restoration, is capable of this. The most extreme example, quite realistic even today, at a price, would be to make a restoration of both a 1920 two-colour print and a 1935 Technicolor print look just like modern Eastman Colour. Leave a good Cineon operator alone with no specification to work to, and that is what he would produce!