GB2352902A

GB2352902A - Film scanner correction.

Info

Publication number: GB2352902A
Application number: GB9918004A
Authority: GB
Inventors: Terence William Mead; Graham Barber
Original assignee: Cintel International Ltd
Current assignee: Cintel International Ltd
Priority date: 1999-07-30
Filing date: 1999-07-30
Publication date: 2001-02-07
Also published as: GB9918004D0

Abstract

An illumination corrector for a film scanner, such as a telecine, has means for deriving a compensation signal as a function of two reference signals received from light detectors 55,57,59 each arranged to detect a different colour. A multiplier 70,72,74 either analogue or digital, may be used to provide the compensation which is mathematically represented as a matrix of incoming signals each representing a different colour of red, green or blue light. Prior to compensation, the signals are amplified in amplifiers 100-300 and undergo afterglow correction.

Description

2352902 FILM SCANNER CORRECTION This invention relates to correction of

various errors which occur in film scanning. In particular, it relates to the correction of light level errors in a film scanner using a Cathode Ray Tube (CRT), caused by defects such as burn.

A modern flying spot telecine consists of a main optical path in which the light from the raster scanned CRT is focused onto the colour film. The emergent light modulated by the film is imaged into a cell box that analyses it into its constituent red, green and blue components. The optical RGB components are then converted into electrical signals using photocells. The cell box consists of dichroic splitting mirrors with one or two trimming filters to obtain the desired ideal colour analysis. The filter elements have tolerances which mean that there are slight variations from telecine to telecine.

Burn is a known problem in the field of CRT scanners and telecines which use a CRT to scan film for conversion to a video signal. Burn causes a resultant unwanted modulation of the signal obtained in addition to the modulation caused by the film scanned. A description of flying spot film scanners may be found in Chapter 39 of "TV & Video

Engineer's Reference Book", Millward J.D., edited by Jackson, K.G., and Townsend, G.B., published by Butterworth and Heinemann, ISBN 0 7506 19538.

The variation in brightness caused by burn can lead to spurious variations in the brightness of light incident upon the film and thus ultimately spurious signals in the generated video signal or data. A modern flying spot telecine, therefore, also has a reference optical path to correct for the ageing, imperfections, and burn of the raster scanned CRT. This reference path looks directly at the CRT and not the film path. In a colour telecine there are three reference paths one for each of the primary colours i.e. red, green and blue. The reference paths have optical colour filters consisting of absorbing and reflecting filters to make a optical bandpass filters for each colour. Each nominally single colour light impinges upon a photocell to produce an equivalent electrical signal corresponding to the energy of the light output energy of the CRT at any instance in that colour. The reference path is used to correct the main path signal for variations in the CRT output.

An example of such a known a system is shown in German Patent Application No DE 2,525,073; in which an apparatus is described for correction of local variations in beam brightness in a CRT due to variations in phosphor, solarization or burning effects. This apparatus uses three additional sensors viewing the CRT face to provide correction signal to take account of brightness variations. The use of three separate colour sensors allows independent correction of the signals in each channel. A block diagram of the main components of a telecine similar to this prior art system is shown in Figure 1. The features of this apparatus will be described later. By having three separate correction channels, one for each colour, it can be seen that the correction signals are independent.

We have appreciated that, notwithstanding the improvements presented by three channel correction, inadequacies remain because the optical path of light to the detector used to produce the signal representative of the image differs from the optical of light to the detector used for correction in each colour channel. We also have appreciated that, in an ideal system, the spectral response of the optical paths for the correction channels and the corresponding main colour channel should be identical. However, we have also appreciated that in a practical system differences in the spectral characteristics will always remain, because of the tolerances of optical filters.

In addition the fact that the reference cell does not look through the film causes a difference in spectral response between the main channel and corresponding reference channel. Film, be it negative, positive or intermediate print stock, affects the colour analysis of the main path through changing the effective colour characteristics of the rastered CRT light. An example of this is that the orange mask of negative film filters out much of the blue light from the rastered CRT light before it gets modulated is by the dye layers containing the image on the film. But all film stocks shape the spectral characteristics of the CRT light.

The reference path light therefore does not undergo the colour spectrum shaping of the film. This in combination with the differences in the colour analysis of the optical components in the two paths means that the reference path is less good at correcting the main path for imperfections in the CRT light than would be the case if the main and reference paths were truly equally matched in colour analysis.

We have appreciated that one solution would be to use identical optical components in each of the main and reference channels and a piece of unexposed film in the reference channel to produce as close a match as possible.

However, even with such measures, differences in spectral response remain.

We have appreciated that such differences in spectral characteristics result in a failure to correct illumination errors in which the colour profile of light produced by a light source changes in addition to the illumination level overall. Accordingly, in a broad aspect, the invention provides improved correction in a film scanner in which variations in coiour are corrected in addition to variations in illumination level.

In particular, there is provided an illumination corrector for a film scanner, the film scanner comprising: a light source for illuminating film to be scanned with generally broadband light comprising a range of colours; at least one image light detector arranged to detect light modulated by film to produce an electrical image signal representative of at least one colour of an image on film; an additional light detector arranged to detect light from the light source unmodulated by film and arranged to produce at least two reference signals each representative of the illumination level provided by the light source at a respective colour; wherein the illumination corrector comprises: an input for receiving the at least two reference signals each representative of the illumination level provided by the light source at a respective colour; means for deriving a compensation signal as a function of the at least two reference signals; and an output for providing the compensation signal for compensating the electrical image signal.

The illumination corrector according to the invention thus allows correction of an image signal with respect to a correction signal derived from signals representative of the intensity of light produced by the light source at different colours, and not just at one colour as in the known system described above. The correction signal can therefore be arranged to be a better representation of the variation in light intensity at the colour represented by the image signal. This improves the correction for a system in which each additional light detector does not exactly sample the same spectral range of light as the image detector; in a practical film scanner or telecine this is usually the case because of differences in the spectral selection by filter components.

In a preferred embodiment the additional light detector is arranged to provide three reference signals respectively representing red, green and blue light, and the input comprises three separate inputs for receiving a respective one of the three reference signals. The additional light detector preferably comprises three separate sensors with respective filtration for red, green and blue light. The embodiment of the invention is thus particularly suited to 15 three colour telecines. In the preferred embodiment, the image light detector is arranged to provide three electrical image signals respectively representing red, green and blue images on film. The image light detector preferably comprises three 20 separate sensors with respective filtration for red, green and blue light. The embodiment can thus correct for errors in each of three colour image signals in a film scanner. The additional light detector is preferably arranged to 25 detect red, green and blue light to produce respective red, green and blue reference signals, and the means for deriving a compensation signal comprises means for producing a linear function of the red, green and blue reference signals. The image light detector is preferably 30 arranged to detect red, green and blue light to produce respective red, green and blue image signals, and the means for producing a linear function of the red, green and blue reference signals comprises means for producing a different linear function of the red, green and blue reference signals in respect of each of the red, green and blue image signals.

The means for deriving a compensation signal preferably comprises a circuit arranged to produce a compensation signal with a matrix of the following form:

Er a b c Brll E9 ai bi ci Eg:

Eb a2 b2 C2, Bb Where Erl, Eg' and Ebl are respective red, green and blue compensation signals, Er", Egli and Eb" are respective red, green and blue reference signals and a, b, C; a,,. bl, cl; and a2, b21C2 are coefficients.

The illumination corrector embodying the invention can thus correct each colour image signal by a compensation signal which is a function of three colour reference signals with varying coefficients. The coefficients of each linear function preferably sum to unity so that gain is maintained constant overall. Whilst a linear function is preferred, other functions could be appropriate. The invention is applicable to a film scanner or telecine with further preferred features set out in the dependent claims.

The invention also provides a method of configuring an illumination corrector as set out in the claims.

Embodiments of the invention will now be described, by way of example only, in which:

Figure 1 is a block diagram of a first known CRT telecine with a correction system to which the invention may be applied; Figure 2 is a graph showing the spectral response of main channel and reference channel optics; Figure 3 is a graph showing the spectral response of main channel and reference channel optics in the event that the colour of light illumination changes; Figure 4 is a block diagram of a telecine including an illumination corrector embodying the invention; Figure 5 is a diagram of one channel of an analogue illumination corrector embodying the invention; Figure 6 is a diagram of one channel of a digital illumination corrector embodying the invention; and Figure 7 is a diagram of scan patch variation used in configuring an illumination corrector embodying the invention.

The embodiment of the invention described is an illumination corrector for use with a film scanner such as a flying spot telecine. A film scanner embodying the invention is also described.

In a flying spot telecine light from the raster scanned CRT is focussed onto film via a main optical path. The emergent light is modulated by film and imaged onto a cell box which analyses it into constituent red, green and blue components. The optical RGB components comprise dichroic splitting mirrors and trimming filters to obtain the desired colour analysis. The filter elements have tolerances which mean that there are slight variations from telecine to telecine.

A flying spot telecine which may embody the invention is shown in Figure 1, described later in detail, and briefly comprises a CRT (10) which illuminates film (30) via a lens (20) The light is then split into separate colour components using dichroics (40,42) and trimming filters (41, 43, 44) an IR filter (45) and imaged onto three separate detectors. The separate red, green and blue detectors (50, 52, 54) feed separate colour channels (60, 62, 64) to produce three colour signals: red (Er), green (Eg) and blue (Eb). In telecine these signal are then typically converted to a television signal. As previously described, the light level produced by the CRT may change, and for this reason additional sensors (55,57,59) are provided to detect changes in illumination level for each colour and produce correction signals.

The three additional sensors, one each for red (55), green is (57) and blue (59) signals, receive light from the CRT (10) via a respective red (155), green (157) and blue (159) filter. The additional sensors (55,57,59) are positioned to view light from the CRT without impeding the light path to the imaging lens (20). The red, green and blue filters provide light of wavelength in approximate ranges 600-685 nm, 510-590 nm and 440-490 nm respectively, though other wavelength ranges may also be used and are within the scope of the invention.

A colour of light passing through one of the optical filters (155,157,159) is received at one of the additional sensors (55,57,59) which produces an output to one of three amplifiers (100,200, 300) which include afterglow correctors to improve the accuracy of the correction signal. Each amplifier (100,200,300) such as Video amplifiers constructed from BFR 90 and BFR 96 transistors produces an output signal Er', Eg' or Ebl respectively.

Now light direct from the face of the CRT should have a constant illumination level. Any changes in the light level are, therefore, unwanted and are due to the various imperfections described before. As the sensors view the CRT (10) direct, these variations are detected and the resultant signals Er', Eg' and Ebl vary with changes in light level.

In known Cintel manufactured telecines the optical filters (155,157,159) are chosen to be as similar in spectral characteristics as possible to the corresponding colour selection components in the main channels i.e. red filter (5S) is similar to the spectral characteristics of the red trimming filter (41); the blue filter (159) is chosen to be similar to the spectral characteristics of the blue trimming filter (43) and the green filter (157) is similar to the spectral characteristics of the red dichroic (40) and green trimming filter (44) combined. The spectral responses are actually a combination of the responses of the dichroics, filters and light produced by the CRT and are shown in Figure 2.

In figure 2 the spectral response of the colour selecting components in the main channels and reference channels are shown. The spectral response of the optical components in the main channels is shown in solid lines and in the reference channels in dashed lines. The spectral response in the main channel is a combination of the response of the CRT, the dichroics and the trimming filters. The response of the red channel is limited at the long wavelength end by the CRT output, and at the shorter wavelength end by the red trimming filter (41). The response of the green channel is limited at longer wavelength end by the red dichroic (40) and at the shorter wavelength end by the green trimming filter (44). The response of the blue channel is limited at the longer wavelength end by the blue trimming filter (43) and at the shorter wavelength end by the CRT output.

As can be seen from Figure 2, the spectral characteristics of the reference channels differ slightly from the corresponding main channels. In the reference channels the characteristics are set by the filters (155, 157, 159) with approximate ranges red 600-685m, green 510-590nm and blue 440-490nm as previously described.

The effect of variation in the colour as well as level of illumination can be seen in Figure 3, which shows the green main channel and three reference channels in a situation in which the blue content of light from the CRT has reduced and the red content has increased. As can be 10 seen, because the spectrum of the green reference channel differs from the green main channel, the level of the green reference channel has reduced more than the green main channel. As a result, the correction signal provided by the green reference channel will be too low leading to 15 a resulting visible blemish on the resultant image. This would not have occurred in the ideal situation in which the optical paths of the main and reference channels were such that the spectral characteristics were identical. A similar condition exists in the red and blue channels. 20 It was realised that the technique for manipulating the effective colour analysis, or display characteristics, of systems using a linear matrix could be applied to correct for the differences. A linear matrix allows a prime colour to have added to it proportions of the other two 25 prime colours (in a three colour system) with a proportionate reduction in the original prime colour constituent e.g. Green' = a, x Green + b, x Red + cl x Blue; where coefficients a, + b, + cl = 1 The coefficients are chosen to sum to one to keep the 30 overall gain constant. It was realised that by using a linear matrix in the reference path its effective taking analysis could be manipulated to better match that of the main path i.e. we could achieve a trimming of the colour response of the reference path.

As stated above when the main and reference paths are not matched residual errors can under certain circumstances be seen in the main path images. A particular concern is that burn edges remain visible which are not removed by the shading map. The shading map is a fixed mapping of the uniformity of the light output from the CRT and optics prior to the cell box photocells. The shading map is generated on demand by the Operator at the start of the session using the telecine but, once generated it remains fixed until the next user demand for regeneration, and thus it is not dynamic and is unable to compensate for effects which vary with temperature. The shading map is also finite in resolution for practical cost and complexity reasons. The shading map cannot correct for:

1. Fine details and edges of burn 2. Effects that are temporally changing.

These effects have to be corrected by the burn correction process which must therefore be as good as it ca. be made.

Dynamic/ temporally changing effects are largely due to temperature. For example if the patch size on the CRT is changed from one frame to the next then the energy put into the original patch area can make a visible difference in the energy output in the new patch area i.e. a hot patch within a cooler patch.

By implementing a linear matrix in the reference cell paths it has been possible to demonstrate that the theory matches the practice and the small errors brought about by the differences in the taking characteristics of the reference and main paths can be eliminated. The main error has been found to be in the green reference channel and by adding some red and or blue reference channel signal to the green to make a new green reference and superior burn correction has been achieved.

An embodiment of invention is shown in Figure 4 and comprises all the components previously described with reference to Figure 1. The components of figure 1 in common with those of figure 4 are thus also described.

The film scanner comprises a CRT (10) such as a GU 2140 CRT manufactured by Brimar Ltd illuminating film (30) through a lens (20). Two dichroic reflectors (40,42) such as Dichroic relay optics reflect red and blue light respectively onto respective sensors (50,54) such as 10 Hamamatsu Photo Multipliers type R2154 photo detectors, but allow transmission onto a third photo detectors (52) for green light. Light reflected by the red dichroic (40) is filtered through a red trimming filter (41) and an IR filter (45) before arriving at the red sensor (50). Light 15 reflected by the blue dichroic (42) is filtered through a blue trimming filter (43) before arriving at the blue sensor (54), and light passed by both dichroics is filtered by a green trimming filter (44) before arriving at the green sensor (52). The photo detectors (50,52,54) 20 provide image signals. The telecine also has three additional sensors; one each for red (55), green (57) and blue (59) signals and which receive light from the CRT (10) via a respective red (155), green (157) and blue (159) filter. The additional 25 sensors (55,57,59) are positioned to view light from the CRT without impeding the light path to the imaging lens (20). The red, green and blue filters provide light of wavelength in approximate ranges 600-685 nm, 510-590 nm and 440-490 nm. respectively, though other wavelength 30 ranges may also be used. A colour of light passing through one of the optical filters (155,157,159) is received at one of the additional sensors (55,57,59) which produces an output to one of three amplifiers (100,200,300) which include afterglow correctors to improve the accuracy of the correction signal. Each amplifier (100,200,300) such as Video amplifiers constructed from BFR 90 and BFR 96 transistors produces reference signals Er", Egli or Eb" respectively.

Now light direct from the face of the CRT should have a constant illumination level. Any changes in the light level are, therefore, unwanted and are due to the various imperfections described before. As the sensors view the CRT (10) direct, these variations are detected and the resultant signals Er", Egli and Eb" vary with changes in light level.

In addition, the film scanner comprises an illumination corrector (400). The illumination corrector receives signals Er", Eb" and Egli from the additional detectors (55, 57, 59) and produces correction signals Er', Ebl and Egi which are then used to correct the main signals Er, Eb and Eg. The signals Erl, Ebl and Eg' are functions of the signs Er", Eb" and Egli according to a linear matrix of the following form:

Er a b C Er E9 a i bi ci Eg Eb a2 b2 c 2) Eb In a known system, not having the illumination corrector (400), the signals Er' = Er", Eg' = Egli and Ebl = Eb". However, by using a matrix linear functions of the input signals Er", Egli and Eb" can be produced. As an example, for the green reference channel the values could be: a, = 0. 07, b, = 0. 93 and cl = 0. This gives Eg' =0.07Er" + 0.93Eg" i.e. the green reference signal Eg" is boosted by addition of some of the red reference channel Er". Referring briefly to Figure 3 again, it can be seen that this has the desired result of boosting the green reference channel to the correct level. The coefficients of the matrix are chosen for particular spectral characteristics of film and filters. A method of setting up the coefficients is described later.

The appropriate correction signals are produced by processing circuitry comprising, for each colour channel, wide band multipliers (110,210,310) type MC 1496 differential amplifiers (120,220, 320) Elantec type EL 2073, resistors (16A,16B) and scaling controls (140, 240,340) which are variable resistors to produce is independent red, green and blue correction signals Vr, Vg and Vb, which are expressed as:

Kr V Kg Kb Vr = Er / g= Eg / Vb = Eb The factors Kr, Kg and Kb are parameters which may be varied using the scaling controls (140,240,340) to adjust the overall level of correction in each channel.

The outputs Vr, Vg and Vb are then applied to video multipliers (70,72, 74) for example type SG 1496 T video multipliers which multiply the colour signals Er, Eg and Eb with the correction signals Er', Eg' and Eb' to produce corrected colour output signals which are Red= ErKr Green= EgKg Blue =EIKI Erl Eg / Ebl The reason that the matrix function provides a good solution can be seen by considering figure 3 again. The reference detectors do not detect at exactly the same frequencies as the corresponding main image detectors, and so variations in colour would not necessarily be corrected as has been described. However, variations in colour caused by various influences such as burn, temperature or previous scan patches are typically are gradual spectral functions. Thus, by taking at least two samples of light at different colours, the illumination level at a third, slightly different colour can be calculated as discussed above.

An analogue implementation of the matrix (for the green channel) is shown in Figure 5. The signals Er", Eb" and Egli are received from amplifiers (100, 200, 300) (Figure 4) and a negative of each signal produced. Three potentiometers (410, 412, 414) are provided to allow adjustment of the amount of each signal added together. The first potentiometer (410) provides adjustment of the amount of red reference (coefficient a,), the second potentiometer (412) provides adjustment of the amount of green reference (coefficient b,), and the third potentiometer (414) provides adjustment of the amount of blue reference (coefficient Cl). A feedback resistor (418) and balance resistor (416) control overall gain of the summed signal amplified by an amplifier (420).

The analogue matrix has been described with reference to the green channel, similar devices are provided within the corrector (400) for the red and blue channels. The three devices together provide the function of the 3 x 3 matrix.

An alternative digital implementation is shown in Figure 6 for the green channel. In this case, Er", Eb" and Eg" are signals produced from detectors in a digital film scanner and are multiplied by appropriate (positive or negative) coefficient a,, b, and c, in corresponding multipliers (430, 432, 434) and then summed in an adder (436) to produce a digital output Eg'. The same device is also used in the red and blue channels.

The amount of red or blue signal to be added to the green is set up through the following test, with reference to Figure 7. A times two zoomin is applied is to the image i.e. smaller size patch than normal (500), with an instant cut 10 to a half size image zoom-out i.e. larger patch than normal (502). A dynamic is set up on the control desk for these two zooms to be toggled every 10 frames or so (i.e. half second). Then with a piece of clear film stock (of the right type) in the gate there will be a visible 15 difference seen between the small and large patch which is not eliminated by the standard burn correction. The reference cell linear matrix is then adjusted until there is minimal observable difference on a waveform. monitor and or picture monitor between the small and large patch 20 signals. Thus the linear matrix is balanced for minimum difference. As stated practical experience shows that the greatest error can be in the reference green signal which can be compensated for by adding either red or blue reference 25 signal. Ideally, red and blue reference paths also need to be adjusted with a linear matrix to optimise performance. This would lead to a nine coefficient linear matrix. In some cases a satisfactory result can be achieved with a limited version matrix using fixed 30 resistor values rather than potentiometers. The invention could be embodied in a film writer in which a feedback signal is fed from the correction sensor to modulate the drive signal to the CRT such as to cancel the unwanted illumination effects. The signal from the sensor would be compared with the input video signal and the result used to increase or decrease the drive to the CRT so as to make the CRT brightness equal to the input video signal.

The coefficients derived using the method discussed above are correct to enable correction of errors due to illumination changes on changing scan patches between stop and run modes of a telecine. The change in colour of a CRT due to other effects such as burn, previous scan 10 patches, granularity of the cathode ray tube, changes with time and temperature and prior small scan patches are similar to that detected in the configuration method and so improved correction is achieved. Blemishes on a cathode ray tube and dirt typically cause an absence of 15 light, rather than a change in colour and so are not really affected by the use of the matrix, but are corrected nonetheless. Although the embodiment described is a CRT telecine having three additional colour detectors, the invention could 20 also be embodied in a sequential colour scan telecine. In such a telecine, a single detector could be used with a colour wheel to produce the two ormore colour reference signals. Such a device would be non-real time. The invention could also be embodied in a noise corrector for 25 a film scanner using a lamp as a light source. The embodiment would detect colour changes in the lamp light. A film writer could also embody the invention. In the case of a sequential colour scan film writer it is only necessary to have one correction sensor which would be 30 fitted such as to view the CRT face through the colour filter wheel. In this specification the term film scanner is used to include a telecine or a film writer.

Claims

1. An illumination corrector for a film scanner, the film scanner comprising:

a light source for illuminating film to be scanned with generally broadband light comprising a range of colours; at least one image light detector arranged to detect light modulated by film to produce at least one electrical image signal representative one colour of an image on film; an additional light detector arranged to detect light from the light source unmodulated by film and aranged to produce at least two reference signals each representative of the illumination level provided by the light source at a respective coiour; wherein the illumination corrector comprises:

an input for receiving the at least two reference signals each representative of the illumination level provided by the light source at a respective colour; means for deriving a compensation signal as a function of the at least two reference signals; and an output for providing the compensation signal for compensating the electrical image signal.

2. An illumination corrector according to claim 1, wherein the means for deriving a compensation signal comprises a multiplier for multiplying each of at least two reference signals by a corresponding coefficient.

3. An illumination corrector according to claim 1 or 2, wherein the means for deriving a compensation signal comprises an adder for adding proportions of at least two reference signals.

4. An illumination corrector according to claim 1, 2 or 3, wherein the means for deriving a compensation signal comprises a matrix.

S. An illumination corrector according to any preceding claim, wherein the means for derving a compensation signal is an analogue circuit.

6. An illumination corrector according to claim 5, wherein the input comprises potentiometers for varying proportions of the reference signals provided to the analogue circuit.

7. An illumination corrector according to claim 5 or 6, wherein the analogue circuit comprises a summing point providing an input to an amplifier and the output of the amplifier is a compensaiton signal.

is

8. An illumination corrector according to any of claims 1 to 4, wherein the means for deriving a compensation signal comprises a digital circuit.

9. An illumination corrector according to claim 8, wherein the input comprises digital multipliers for varying proportions of the reference signals provided to the digital circuit.

10. An illumination corrector according to claim 8 or 9, wherein the digital circuit comprises a digital adder.

11. An illumination corrector according to any preceding claim, wherein the additional light detector is arranged to provide three reference signals respectively representing red, green and blue light, and the input comprises three separate inputs for receiving a respective one of the three reference signals.

12. An illumination corrector according to claim 11, wherein the additional light detector comprises three separate sensors with respective filtration for red, green and blue light.

13. An illumination corrector according to any preceding claim, wherein the image light detector is arranged to provide three electrical image signals respectively representing red, green and blue images on film.

14. An illumination corrector according to claim 13, wherein the image light detector comrprises three separate sensors with respective filtration for red, green and blue light.

15. An illumination corrector according to any preceding claim, wherein the additional light detector is arranged to detect red, green and blue light to produce respective red, green and blue reference signals, and the means for deriving a compensation signal comprises means for producing a linear function of the red, green and blue reference signals.

16. An illumination corrector according to claim 15, wherein the image light detector is arranged to detect red, green and blue light to produce respective red, green and blue image signals, and the means for producing a linear function of the red, green and blue reference signals comprises means for producing a different linear function of the red, green and blue reference signals in respect of each of the red, green and blue image signals.

17. An illumination corrector according to claim 16, wherein the means for deriving a compensation signal comprises a circuit arranged to produce a compensation signal with a matrix of the following f orm:

Er 0 a b c Er"" Eg ai bi Ci E911 Eb a2 b2 c2AEb", Where Erl, Eg' and Ebl are respective red, green and blue compensation signals, Er", Eg" and EbIl are respective red, green and blue reference signals and a. b. C; al. bl,cl; and a2, b2,C2 are coefficients.

18. A film scanner comprising:

a light source for illuminating film to be scanned with generally broadband light comprising a range of colours; at least one image light detector arranged to detect light modulated by film to produce at least one electrical image signal representative one colour of an image on film; an additional light detector arranged to detect light from the light source unmodulated by film and aranged to produce at least two reference signals each representative of the illumination level provided by the light source at a respective colour; means for deriving a compensation signal as a function of the at least two reference signals; and means for compensating the electrical image signal with reference to the compensation signal.

19. A film scanner according to claim 18, wherein the means for deriving a compensation signal comprises a multiplier for multiplying each of at least two referenfce signals by a corresponding coefficient.

20. A film scanner according to claim 18 or 19, wherein the means for deriving a compensation signal comprises an adder for adding proportions of at least two reference signals.

21. An illumination corrector according to claim 18, 19 or 20, wherein the means for deriving a compensation signal comprises a matrix.

22. An illumination corrector according to any of claims 18 to 21, wherein the means for derving a compensation signal is an analogue circuit.

23. A film scanner according to claim 22, further comprising potentiometers for varying proportions of the reference signals provided to the analogue circuit.

24. A film scanner according to claim 22 or 23, wherein the analogue circuit comprises a summing point providing an input to an amplifier and the output of the amplifier is a compensation signal.

25. A film scanner according to any of claims 18 to 21, wherein the means for deriving a compensation signal comprises a digital circuit.

26. A film scanner according to claim 25 comprising digital multipliers for varying proportions of the reference signals provided to the digital circuit.

27. A film scanner according to claim 25 or 26, wherein the digital circuit comprises a digital adder.

28. A film scanner according to any of claims 18 to 27, wherein the additional light detector is arranged to provide three reference signals respectively representing red, green and blue light, and the input comprises three separate inputs for receiving a respective one of the three reference signals.

29. A film scanner according to claim 28, wherein the additional light detector comprises three separate sensors with respective filtration for red, green and blue light.

30. A film scanner according to any of claims 18 to 29, wherein the image light detector is arranged to provide three electrical image signals respectively representing red, green and blue images on film.

31. A film scanner according to claim 30, wherein the image light detector comprises three separate sensors with respective filtration for red, green and blue light.

32. A film scanner according to claims 18 to 31, wherein the additional light detector is arranged to detect red, green and blue light to produce respective red, green and blue reference signals, and the means for deriving a compensation signal comprises means for producing a linear function of the red, green and blue reference signals.

33. A film scanner according to claim 32, wherein the image light detector is arranged to detect red, green and blue light to produce respective red, green and blue image signals, and the means for producing a linear function of the red, green and blue reference signals comprises means for producing a different linear function of the red, green and blue reference signals in respect of each of the red, green and blue image signals.

34. A film scanner according to claim 33, wherein the means for deriving a compensation signal comprises a circuit arranged to produce a compensation signal with a matrix of the following form:

Er a b c) Er") E_q ai bi ci E_q Eb a2 b2 c2) Eb Where Er', Eg' and Ebl are respective red, green and blue compensation signals, Er", Eg" and Eb" are respective red, green and blue reference signals and a, b, c; a,, bl,c,; and a2, b2,C2 are coefficients.

35. A telecine comprising an illumination corrector according to any of claims 1 to 17, or a film scanner according to any of claims 18 to 34.

36. A telecine according to claim 35, wherein the light source is a cathode ray tube.

37. A method of configuring a film scanner, the film scanner comprising, a cathode ray tube for illuminating film to be scanned with generally broadband light comprising a range of colours, at least one image light detector arranged to detect light modulated by film to produce an electrical image signal representative of at least one colour of an image on film, an additional light detector arranged to detect light from the light source unmodulated by film and arranged to produce at least two reference signals each representative of the illumination level provided by the cathode ray tube at a respective colour, and means for deriving a compensation signal as a function of the at least two reference signals and for correcting the electrical image signal; the method comprising:

controlling the cathode ray tube to alternately scan at least two different scan patches; analysing the level of the electrical image signal at points representative of each of the two scan patches; and is varying the function of the at least two reference signals until there is a minimum difference in the electrical image signal in respect of each of the two scan patches.

38. A method according to claim 37, the step of analysing comprising providing the electrical image signal to a display and viewing the display, the step of varying the function comprising altering the observable difference on the display.

39. A method according to claim 37, the step of analysing comprising comparing the levels of electrical image signal at points representative of each of the two scan patches, the step of varying comprising altering the function until there is a minimum difference in the levels.

40. A method according to any of claims 37, 38 or 39, wherein the function is a linear function in the form of E.," = aYEY + a,E, where E,11 is the compensation signal, EY and E, are the reference signal, and ay and a, are coefficients of the reference signals, the step of varying the function comprising altering the coefficients.

41. A method according to any of claims 37 to 40, wherein the at least one image light detector is arranged to produce three electrical image signals each representative respectively of red, green and blue images on film, and the additional light detector is arranged to produce three reference signals each representative of the illumination level provided by the cathode ray tube for red, green and blue light, and wherein the liner function is a matrix of the form Er a b c Er E-q ai bi ci E-qr" Eb'),a2 b2 C2 Eb") wherein Er", Egli or Eb" are reference signals respectively for red, green and blue light, Er', Ebl and Eg' are compensation signals used to correct the respective image signals for red, green and blue images and a, b, c, and a,, b,, c,, and a2, b2, C2 are coefficients.

42. A method according to claim 42 or a film scanner according to claim 48, wherein the coefficients of S each linear function sum to unity.

42. A method according to claim 41, wherein the step of varying the function comprises altering the coefficients.

43. A film scanner comprising, a cathode ray tube for illuminating film to be scanned with generally broadband light comprising a range of colours; at least one image light detector arranged to detect light modulated by film to produce an electrical image signal representative of at least one colour of an image on film; an additional light detector arranged to detect light from the light source unmodulated by film and arranged to produce at least two reference signals each representative of the illumination level provided by the cathode ray tube at a respective colour; means for deriving a compensation signal as a function of the at least two reference signals and for correcting the electrical image signal; means for controlling the cathode ray tube to alternately scan at least two different scan patches; means for analysing the level of the electrical image signal at points representative of each of the two scan patches; and means for varying the function of the at least two reference signals until there is a minimum difference in the electrical image signal in respect of each of the two scan patches.

44. A film scanner according to claim 43, the means for analysing comprising means for providing the electrical image signal to a display.

45. A film scanner according to claim 43, the means for analysing comprising means for comparing the levels of electrical image signal at points representative of each of the two scan patches, the means for varying comprising means for altering the function until there is a minimum difference in the levels.

46. A film scanner according to any of claims 43, 44 or 45, wherein the function is a linear function in the form of E," = aYEY + aE, where E." is the compensation signal, EY and E, are the reference signal, and ay and a, are coefficients of the reference signals, the means for varying the function comprising means for altering the coefficients.

47.A film scanner according to any of claims 43 to 46, wherein the at least one image light detector is arranged to produce three electrical image signals each representative respectively of red, green and blue images on film, and the additional light detector is arranged to produce three reference signals each representative of the illumination level provided by the cathode ray tube for red, green and blue light, and wherein the linear function is a matrix of the form Er a b c ( Er -E9 ai bi c i Eq, a2 b2 C 2) Eb wherein Er", Eg" or Eb" are reference signals respectively for red, green and blue light, Er', EbI and Eg' are compensation signals used to correct the respective image signals for red, green and blue images and a, b, c, and a,, bl, cl, and a2, b2, C2 are coefficients.

48. A film scanner according to claim 47, wherein the means for varying the function comprises means for altering the coefficients.

49. An illumination corrector according to any of claims to 17, or a film scanner according to any of claims 32 to 34, wherein the linear function of the red, green and blue reference signals has coefficients which sum to unity.