US2987571A

US2987571A - Color printing

Info

Publication number: US2987571A
Application number: US718748A
Authority: US
Inventors: Allen Gordon Stanley James; Mawby David Harry
Original assignee: CROSFIELD J F Ltd
Current assignee: CROSFIELD J F Ltd; J F CROSFIELD Ltd
Priority date: 1957-03-11
Filing date: 1958-03-03
Publication date: 1961-06-06
Anticipated expiration: 1978-06-06
Also published as: GB855895A; DE1135295B; FR1202287A

Description

' June 5, 1961 G. 5. J. ALLEN El'AL 2,987,571

' COLOR PRINTING Filed March 3. 1958 3 Sheets-Sheet 1 By 2. M

A Home y June 6, 1961 G. s. J. ALLEN ETAL COLOR PRINTING 3 Sheets-Sheet 2 Filed March 3, 1958 vlll.

June 6, 1961 G. s. J. ALLEN EI'AL 2,987,571

COLOR PRINTING Filed March 3, 1958 3 Sheets-Sheet 5 Inventors Mm I. 77M

MAM. W QM A ftorney United States Patent 2,987,571 COLOR PRINTING Gordon Stanley James Allen, and David Harry Mawby,

London, England, assignors to J. F. Crosfield Limited,

London, England, a company of Great Britain Filed Mar. 3, 1958, Ser. No. 718,748 Claims priority, application Great Britain Mar. 11, 1957 16 Claims. (Cl. 178 5.2)

This invention relates to the reproduction of colored originals.

In color printing, a reproduction of a color original is made by the superposition of differently colored inks, the amount and distribution of each ink being determined by photographing the original through filters of suitable colors. Thus in three'color printing, photographs of the original are made through red, green and blue filters, and the resulting negatives are used to obtain positives from which the printing cylinders or plates are prepared. By means of the three printing cylinders, cyan magenta and yellow inks are laid down, usually on white paper, to absorb red, green and blue light, respectively, where these colors are not required. However, in addition to non-linearities introduced by the photographic system and by any etching system employed, the printing inks themselves suffer from impurities. As an example, cyan ink appears to contain a large amount of magenta, that is to say, the cyan ink absorbs some of the green light, which it should reflect completely.

To overcome this difliculty, a correction which can be called single-stage linear masking, can be introduced photographically. The correction required is the reduction of the density of each point on the magenta printer (taking the example given above) by an amount which varies with the density of the corresponding point on the cyan printer. To achieve this correction, a positive mask with a reduced contrast range is made from the cyan printer negative, and if the mask is then superimposed in register on the magenta printer negative, an exposure through the combination produces a magenta printer positive which is at least partly corrected.

One disadvantage of this system is that, since all the negatives have densities which are substantially equal to each other in areas representing grey tones, the density contrast of the grey tones obtained by the combination of the negative being corrected and the positive correcting mask is less than the contrast of the tones in the negative alone. This reduction must not be too great if the required contrast is to be obtained in the subsequent positive.

The greatest drawback of single-stage masking, however, is its inability to overcome the non-linearity known as additivity failure. When printing inks of different colors are to be superimposed, the sum of their effective densities measured separately is often considerably higher than the density of the superimposed inks. It has been stated above that cyan ink appears to contain a large amount of magenta. If a given weight of cyan ink is deposited on two areas, one containing very little magenta ink and the other containing a large amount of magenta ink, the etfect of the apparent magenta in the cyan ink may be very great in the former area, but quite small in the latter area, which is already almost saturated with magenta. If, as in single-stage linear masking, the densities of both areas on the corresponding black-and-white magenta printer positive are reduced to the same extent, the area containing a large amount of magenta ink will be seriously overcorrected, which is very undesirable in color reproduction.

' It has been proposed to overcome these drawbacks to some extent by a photographic process which may be ice called two-stage masking. In two-stage masking the correcting mask for a given negative is made from a combination of the negative of the correcting color and a positive made from the negative to be corrected. Thus, the positive made from the magenta negative is of such a range that when it is registered with the cyan printer negative the neutral scale of densities is cancelled out, that is to say the scale for which the densities for magenta and cyan are equal. Thus there is no reduction of the contrast of the neutral scale. However, considerable non-linearity must be introduced into the process if good correction is to be obtained in colors other than neutral tones. In an electro-optical scanner of the kind envisaged in the present invention, a large range of modification of the correcting signal would be required accurately to compensate for this non-linearity.

There is, however, a limit to the amount of compensating non-linearity which can be introduced into an equipment designed to provide a consistent performance and to be capable of reproduction with the same performance, and moreover it is found that when such extreme nonlinearity is provided any small variations in densities or in the corresponding density signals from one subject to another may cause a considerable change in the resulting print, and small blemishes in the original may be greatly magnified. As an alternative, the non-linearity may be set so that the system will give satisfactory, although not perfect, color reproduction, but so that the results obtained from it will be consistent from one subject to another and from one day to another. If it is accepted that perfect color reproduction cannot be achieved, it is important to appreciate where the compromise should be made. As an example, it is usually more serious for a printer to overcorrect than to undercorrect.

In the method of color reproduction according to the invention an original or uncorrected transparencies or prints derived from the original are scanned to provide electric signals representing color information, and an electric signal representing a correction to be applied for a given color is obtained by deriving, from a signal representing the color to be corrected and a signal or signals representing a correcting color or colors, a resultant signal which represents the result of subtracting the density or a function thereof of one or more correcting colors from the density or a function of the density of the color to be corrected. The resultant signal is passed through a non-linear circuit which is arranged to attenuate the signal when it exceeds a predetermined amount. The output of the non-linear circuit is the correcting signal, and a photographic emulsion serving to provide the printer for the given color is then exposed in acordance with the density of the transparency to be corrected as modified by the said correcting signal.

Although the combining of the signals representing the color to be corrected and the correcting color (if a single correcting color is assumed) must be carried out in such a manner that the color densities are added in opposing sense (for example, in opposite polarities), the signals themselves need not represent densities directly and need not be algebraically added. For example, they can represent the transmittance values of transparencies corresponding to the two negatives. In this case, since the transmittance values vary inversely with the antilogarithms of the density values, one signal must be divided by the other in order to obtain the effect of adding the densities in opposing sense.

In the preferred embodiment of the invention the correcting signal is used to modulate a light source the light from which passes through an uncorrected transparency, representing the color to be corrected on its way to the photographic emulsion which is to be exposed.

cathode raytube, on the face of which a scanning raster is formed, the modulation of the electron beam by the correcting signal causing the image on the face of the cathode ray tube to constitute a correcting light mask.

A number of separate scanning beams from the modulate'd light source may be transmitted through a number of uncorrected transparencies to light-sensitive devices which provide electric signals representing quantities of light which vary in accordance with the modulated source and with the transmittance values of the transparencies, and the emulsion to be exposed may be placed behind the uncorrected transparency for which the correction has been computed, the light passing through the latter and through the said emulsion to the associated light-sensitive device. i 'It is preferably arranged that the signals which are used to form the correcting signal are combined at an early stage in such a manner that signals corresponding to neutral tones in the original are substantially cancelled out. Any desired form of non-linearity can be introduced after this stage, to correct the color values, without affecting the neutral tones. I

Good results can be obtained when the non-linear circuit includes as the non-linear element a component having a sharp cut-01f, for example a diode.

In order that the invention may be better understood two embodiments thereof will now be described with reference to the accompanying drawings, in which:

FIGURE 1 is an explanatory diagram;

FIGURE 2 is a schematic diagram of a system according to the present invention;

FIGURE 3 shows an alternative form of correction computing unit; and

FIGURES 4, 5, 6 and 7 are circuit diagrams giving details of circuits used in the apparatus of FIGURES 2 and 3.

The method according to the invention will now be described with reference to FIGURE 1. For the purpose of this explanation, it will be assumed that a photographic plate having exposed upon it three squares of differing density is used to etch a cylinder to print cyan density distribution is represented by diagram (a) in FIG- URE 1, in which the ordinates represent densities across the plate. This diagram shows that each square is of uniform density but that the densities of the squares increase from left to right. Another plate exposed to the same three density levels but having in each square the step-wedge density distribution indicated in diagram (b) (i.e. three strips of increasing density) is used to etch a cylinder which is to print magenta The concentration of the inks is balanced so that when the cyan, magenta and yellow inks are superimposed on White paper, equal amounts of the inks will produce a grey tone or black. The superimposition of equal amounts of cyan 55 and magenta produces the neutral color, blue.

The two cylinders are used to make a print in which cyan and magenta inks are superimposed. This colored subject is then photographed through red and green filters respectively, and contact positives are made from the two 80 negatives. If the process described so far has been linear and if the inks are free from impurities, then these positives would be the same as the original positives (a) and (b). It is found, however, that whilst the new cyan positive agrees fairly well with the original cyan plate rep-re- 66 to say in the first step of the first square, the second 70 step of the second square, and the third step of the third square. This is because the concentration of the inks has been balanced to provide this agreement for the neutral tones. Because of additivity failure, however, the

" aa mrl A i a creased, the decrease in contrast being greatest in the right-' hand square in which a large amount of cyan has been printed. Diagram (0) also represents the signal in logarithmic form obtained from a photomultiplier during scanning of the corresponding magenta negative with an unmodulated light source. If the light source could be modulated insuch a manner that the magenta signal from the photomultiplier appeared as in (b) instead of as in (c), then this signal could be said to represent a corrected magenta positive. A plate exposed through the magenta negative by means of a light source which is modulated in this manner would form a corrected magenta positive. If this were now used in conjunction with the cyan positive,'a second print would be obtained which, when photographed through red and green filters, would provide negatives such that the corresponding positives would be close copies of those represented at (a) and (b). a

To obtain the signal which is required to modulate the light source, a signal corresponding to the new cyan positive (i.e. corresponding approximately to diagram (a)) is inverted (see diagram (d)) and combined with a signal corresponding to the new magenta-positive (diagram (0) The resultant signal is as shown at (e).

The waveform (e) is passed through a non-linear circuit which is arranged to attenuate this signal when it exceeds a given value (i.e. when the magenta signal (c) exceeds the cyan signal). The e'ifect of a circuit of this kind is seen in the waveform (f). This waveform is the correcting signal and is used to modulate the brightness of a scanning light source which is used to transmit light through the uncorrected negative (corresponding to wave form (0)) to the emulsion to be exposed. The waveform (1) therefore also represents the spot brightness level for the scanning of a single line of the uncorrected magenta negative. The brightness level of the light reaching the emulsion to be exposed (after passing through the uncorrected negative) then takes the form shown at (g), which has been obtained by combining diagram (1) with the un'corrected'magenta signal in diagram (0). It will be see'nthat the magenta contrast (the difference in amplitudebetween the first and third strips of the square) is restored in the second and third squares,'and that diagram '(g) closely resembles diagram (b). The condition has therefore been fulfilled, showing that Waveform (f) is the required correcting waveform.

It is seen that the signals cancel for the densitiesreprcsenting neutral colors and no modulation occurs in these areas.

The apparatus which is used to carry out the method according to the invention will now be described.

In FIGURE 2 the cathode ray tube 10 constitutes the light source for scanning the uncorrected negatives and also for exposing the emulsion which is to form the corrected printer positive forone of the printer colors. The light spot on the face of the cathode ray tube which is modulated in accordance with the required correction for one of the colors, is focused through three lenses 12 onto three uncorrected separation negatives 14 so that the three negatives are scanned simultaneously and in synchronisrn. The light beams pass through the negatives onto photomultipliers 16. Assuming that the middle negative corresponds to the color for which a corrected printer negative is being prepared, the emulsion which is to become the corrected printer positive is placed immediately behind the middle negative 14, as shown at 18. Thus the scanning light spot reaching the emulsion 18 contains variations due to the modulation of the brightness of the cathode ray tube spot and also variations to the uncorrected negative '14. The emulsion 18 may be blue-sensitive so that, if it is provided with the yellow backing 20, the yellow component of the light will pass through to the photomultiplier 16 but the blue component will be prevented from fogging the plate by reflection contrast between the densities in each squarev has .deto the back of the 61111 519 1.-

The light passing through the middle negative 14 is diffused by the emulsion 18 and in order to diffuse the transmitted light in the upper and lower channels in a similar manner, diffusing plates 22 are placed behind the upper and lower negatives 14. The output signals from the photomultipliers 16 are applied to logarithmic circuits 24 which provide signals representing the logarithms of the input signals. The signals from the logarithmic circuits are applied directly to contacts a of switches 26 and also to inverting circuits 28, as a result of which the logarithmic signals are applied in inverted form to contacts 0 of the switches 26; contacts b are not connected. The switch 26 may thus be used to select any or all of the three signals in either polarity. The movable contacts of the switches 26 are connected to a summing amplifier 30 in which, in the example shown, a positive signal from the middle channel is added to a negative signal from the upper channel, the lower channel being disconnected. The output of the summing amplifier 30 is applied to a mixer 32 which receives a driving voltage from a drive circuit 33, this driving voltage determining the reference level of the brightness of the spot on the cathode ray tube. The signal from the mixer 32 is then applied to an antilogarithmic circuit 34 by means of which signals are derived representing the antilogarithms of the signals from the mixer. These antilogarithmic signals are applied to a non-linear circuit 36 by means of which they are attenuated when the difference between the density of the color to be corrected and the density of the correcting color exceeds a predetermined amount (see FIGURE 1(f)). The output of the non-linear circuit 36 constitutes the correcting signal and is applied to the grid of the cathode ray tube.

If the upper channel in FIGURE 2 is the cyan channel and the middle channel the magenta channel, the signals from the upper photomultiplier 16 will be proportional to BXC, where B is the spot brightness and C represents the transmittance of the scanned portion of the cyan negative at any instant. Similarly the signals from the photomultipler 16 in the middle channel will be proportional to BXM where M represents the transmittance of the magenta negative at some point. After passing through the logarithmic circuits 24, the signals represent log B+log C, and log B+log M respectively. Thus the signals which are applied to the summing amplifier 30 in FIGURE 2 are (log B+log M) and (log B+log C). It will be seen that the result of this is log M -log C, the brightness factor having cancelled out. The values log M and log C represent the densities of the cyan and magenta negatives respectively, and it is the density difference signal which is applied to the antilogarithmic circuit 34, the output of which represents the quotient M/ C of the transmittance values of the two negatives. The output signal from the non-linear circuit modulates the brightness of the spot on the face of the cathode ray tube in accordance with the expression and the transmittance of the uncorrected magenta negative, and is given by M f(% K) or in terms of the densities of the colors log M+f(log M-log C+log K) It will usually be required that the neutral tones, that is to say those with equal cyan and magenta contents, shall be reproduced without change, and this is easily .arranged by making the signal log M log C of reference level when cyan and magenta tones of equal densities are being scanned. If desired, however, a tone compensation circuit may be incorporated for the purpose of modifying the reproduction of neutral tones.

If the yellow signal in the lower channel is also used to correct the magenta signal, the lower part of the switch 26 is set to contact 0, so that the output of the adding circuit 30 represents log M log C-log Y, where Y is thetransmittance value of the yellow negative. In this case, however, the reference level of the output signal corresponds to the value obtained when M, C and Y are all of equal value.

In the example described above the factor representing the modulation of the brightness of the spot on the face of the cathode ray tube was removed during the summation in the amplifier 30. If for any reason it is required to remove this factor before the summation of the signals, each signal can be separately demodulated in the following manner. A photomultiplier 38, shown in dotted lines in FIGURE 2, is exposed directly to the face of the cathode ray tube and its output signal is applied to a logarithmic circuit 49 and then through an inverting circuit 42. The output of the inverting circuit is then algebraically added to the output signals from each of the logarithmic circuits 24. It will be seen that the subtraction of the logarithm of the signal from photomultiplier 38 from each of the signals in the colour channels (representing the logarithms of the outputs of photomultipliers 16), is equivalent to dividing the signals from the photomultipliers 16 by a factor representing the brightness of the tube at any instant. Thus the three resultant signals will represent the logarithms of the transmittance values (that is to say the densities) of the three negatives.

With some emulsions, it has been found possible to emit the filter backing 20 and therefore to extend the plate sensitivity into the green or even into the red part of the spectrum.

FIGURE 3 shows a simpler embodiment of the invention in which the output signals from the photomultipliers 16 are applied directly to the contacts a of the switch 26, and through the inverting circuits 28 to the contacts c. The signals selected by means of the movable contacts of the switch 26 are applied to the summing amplifier 30, the resultant passing to a non-linear circuit 36 and thence to the mixer circuit 32 where it is combined with the drive signal from the circuit 34. Although in this case the transmittance values are subtracted directly, without couverting them to density values by taking logarithms, it is found that good results are obtained if the signals are maintained within a suitable range of values. The subtraction cancels to some exent the modulation present in the colour signals due to the modulation of the light spot on the face of the cathode ray tube. Again if complete demodulation is required the output of the photomultipliers can be convertedto logarithmic form and a signal derived from the photomultiplier 38 (FIGURE 2 also converted to logarithmic form, can be subtracted from the colour density signals.

Although it is preferred to efi ect multiplication of the transmittance signals by logarithmic conversion followed by summation, it is also possible to employ multigrid valves in known manner to effect the multiplication directly.

FIGURE 4 shows a circuit by means of which the signal from the photomultiplier 16 may be made to bear a logarithmic relationship to the current flowing to the collector of the photomultiplier, and therefore to variations in the light falling upon the photomultplier. The collector of the photomultiplier is connected to the H.T. positive terminal by way of a load resistance- 44 which takes the form of a germanium or silicon diode. Since the photomultiplier itself has a very high impedance, reasonable variations of the load resistance 44 have only a very small effect on the photomultiplier current. As the resistance of the germanium or silicon diode 46 decreases exponentially as the current which flows through it increases over the range of current required, the voltage across the diode follow a logaiithmic cprve when plotted against variations in the light falling on the photomultiplier.

1 FIGURE 5 shows the basic form of an antilogarithniic circuit employing a germanium or silicon diode 46. The input voltage is applied across the diode 4a in series with a resistor 48 which is of small value compared with the resistance of the diode, and the output is taken across the resistor 48. If an appropriate range of current and voltage'is maintained, the resistance of the diode will decrease exponentially with increase of current, and the voltage across the linear resistor 48 will be proportional to the current flowing through the combination and will therefore represent the antilogarithm of theinput voltage. This is only true, however, if the resistance of the component 48 is very small compared with that of the diode 46, in which case the signal across the resistor 48 is very small. Figure 6 shows a circuit in which the input to the diode is obtained from a cathode follower 50 and is applied across the diode 46, a resistor 52 and a resistor 54 in series. The resistor 54 is the cathode resistor of a triode 56, and a signal derived from the junction of the resistor 52 and the diode 46 is applied through a triode 58 to the grid of the triode 56. The resistance across the

components

52 and 54 appears very small [8.5 a result of the negative feedback which is aplied across the resistor 54in this manner. The amplified output signal is obtained from the anode of the feedback amplifier triode 58.

In FIGURE 7 there is shown an example of a simple non-linear amplifier. "in this circuit the input signal is applied to the grid of a triode 6%, the anode of which is connected through a non-linear component 62 to the junction of two

resistors

64 and 66 which are connected in series across the HI. supply. The output is taken from the junction of these two resistors. The component 62 may be a diode with a sharp cut-oil, so directed that current will pass from left to right in FIGURE 7 but not from right to left. In this way information is passed to a subsequent stage when the signals are such that the anode of the triode 60 is positive with respect to the junction of the

resistors

64 and 66, but when the anode is negative with respect to the potential of this junction, the voltage of the output conductor is maintained at the junction potential.

Instead of having a sharp cut-ofi the non-linear circuit may have, for example, an exponential slope.

In some cases, to simplify the design of amplifying circuits it may be desirable to modulate the light beam at constant amplitude and high frequency (in addition to the modulation by the correction signal) so that the output of the photomultipliers is an alternating voltage resembling a modulated carrier wave.

The inverter circuits may be single triode circuits to the grid of which the input signal is applied, the inverted output signal being obtained from the anode of the triode. The summing amplifiers may be of the known kind in which the signals to be summed are applied through individual resistors to the grid of a summing triode, the voltages across the grid resistor of the triode being kept small by the introduction of negative feedback from the anode to the grid through a resistor of suitable value.

In the circuits described above the correction signal has been used to form a light mask on the face of the cathode ray tube which exposes an emulsion through the negative to be corrected, and the same cathode ray tube has been used to scan the uncorrected'negatives to provide information from which the correcting signal is derived. It is not, however, essential for the same light source to be used both to obtain the information and to expose the printer emulsion. Furthermore instead of using the correcting signal to form a light mask on the face of the tube, the correcting signal can be multiplied by signals representing the transmittance of the uncorrected negative for the printer in question (or the logarithms of these signals may be added) to provide a resultant corrected signal which can be used to modulate 8 alight source which exposes the printer emulsion d1 rectly,'

Satisfactory reproduction of the grey scale is a very important part of the overall reproduction process, and

is e ss ential that a system for producing photographic records from which suitable prints can be mad'e'should not only'provide adequate colour correction, but also be.

capable of a reproduction of the grey scale which is either substantially perfect or is modified in a controlled manner. In the methoddescribed above, freedom from distortion is achieved by setting up the scanner in such a manner that the grey scale cancels out in the mask," so that the grey scale appearing in the corrected positive is a direct reproduction of that in the negative. This is done by arranging that, for grey tones, the amplitudes of the two signals used to form the mask cancel out in the subtraction stage. If, however, it is required to produce a positive in which the grey scale is difierent from that in the negative, a tone compensating circuit can be incorporated.

The scanner which has been described may be used to carry out the electronic equivalent of a photographic technique which may be called successive two-stage masking. Considering the masking of the yellow printer by the magenta printer, in the photographic process a positive made from the uncorrected yellow negative is registered with the uncorrected magenta negative, both being made to a gamma of unity so that the grey tones cancel out. :From this combination a mask is made, and this mask is used with the uncorrected yellow negative to. expose the first corrected yellow positive. This is usually the end product of two-stage masking. If, however, this first corrected yellow positive is registered with the uncorrected magenta negative and another mask made from the combination, this second mask can be registered with the uncorrected yellow negative to mask a second corrected yellow positive. This process can go on indefinitely, the amount of correction carried out becoming progressively less at each stage. It will be seen that the process soon approaches a point where the mask is made 'from the combination of a corrected yellow value and an uncorrected magenta value. This kind of result can be easily achieved in the scanner described above by using its feedback properties, and the resulting method is an important form of the present invention.

In one form of this method, the signals from the yellow and magenta channel photomultipliers are passed through logarithmic circuits and the signal in the magenta channel is demodulated to remove the elfect of variations in spot brightness. The resulting signal in the magenta channel, representing density values of the uncorrected magenta negative, and the signal in the yellow channel, representing density values of the corrected yellow negative, are fed in phase opposition to a summing amplifier, the output of which is fed through an antilogarithmic and a limiting circuit to the grid of the cathode ray tube.

The use of a scanner has, of course, considerable advantages in the saving in photographic materials and reduction in processing time. The use of a single light source for both analysis and reproduction, as in the system described above, has further advantages in that the problems of accurate registration are largely overcome, as are other problems associated with possible differences arising from separate analysing and reproducing light sources.

We claim:

1. Color reproduction apparatus comprising an electro-optical scanner, a correction path including a circuit in which at least one electric signal from said scanner representing a correcting color is subtracted from an electric signal representing a color to be corrected, a non-linear circuit for receiving the output signal from said subtraction circuit and having an input-output characteristic in which the gain is reduced as the input increases thereby providing an uncompensated non-linearity in said correction path, and a light source the brightness of which varies with the output of said non-linear circuit and which is arranged to expose a photographic emulsion which will provide the printer for the given color.

2. Apparatus according to claim 1, in which said light source is arranged to expose said photographic emulsion through an uncorrected transparency representing the color to be corrected.

3. Apparatus according to claim 1, including a circuit in which the signal from said non-linear circuit is combined with an uncorrected electric signal representing the given color to be corrected, to provide a resulting corrected signal for the given color which is used to modulate said light source, whereby the brightness of said light source varies both with the uncorrected signal and with the corrected signal from the non-linear circuit.

4. Apparatus according to claim 1, in which the nonlinear circuit contains as its non-linear element a diode.

5. Apparatus according to claim 1, in which the light source is a cathode ray tube.

6. Apparatus according to claim 1, in which said scanner includes signal-generating means which supply to the subtracting circuit signals representing the density of the color to be corrected and the density of at least one correcting color.

7. Apparatus according to claim 1, in which the density signals from said signal-generating means each include signal variations representing a correction to be applied to a given color, whereby said signal variations cancel out in the subtraction process.

8. Apparatus according to claim 1, in which said scanner includes signal-generating means which supply to said subtracting circuit signals representing the transmittance of the color to be corrected and the transmittance of at least one correcting color.

9. Apparatus according to claim 8, in which the transmittance signals from said signal-generating means each include signal variations representing a correction to be applied to a given color whereby said signal variations cancel out in the subtraction process.

10. Apparatus according to claim 1, in which said non-linear circuit provides an output which varies with its input when the signal representing the correcting color exceeds the signal representing the color to be corrected but is of constant level when the signal representing the color to be corrected exceeds the signal representing the correcting color.

11. Apparatus according to claim 1, in which said nonlinear circuit provides an output which varies with its input when the sum of the signals representing the correcting colors exceeds the signal representing the color to be corrected but is of constant level when the signal representing the color to be corrected exceeds the sum of the signals representing the correcting colors.

12. Color reproduction apparatus comprising an electro-optical scanner, a correction path including a circuit in which an electric signal from the scanner representing the transmittance of a transparency corresponding to a color to be corrected is divided by at least one electric signal from the scanner representing the transmittance of a transparency corresponding to a correcting color, the apparatus further comprising a non-linear circuit which receives the output signal from said dividing circuit and having an input-output characteristic in which the gain is reduced as the input increases thereby pro- 10 viding an uncompensated non-linearity in said correction path, and a light source the brightness of which varies with the output of said non-linear circuit and which is arranged to expose a photographic emulsion which will provide the printer for the given color.

13. Color-correction apparatus comprising an electrooptical scanner for scanning an original or color separations therefrom, a correction path including a circuit for combining an electric signal from said scanner representing a color to be corrected with at least one electric signal from said scanner representing a correcting color so that a function of the density of the correcting color is subtracted from a function of the density of the color to be corrected, a non-linear circuit for receiving the output signal from said combining circuit and having an input-output characteristic in which the gain is reduced as the input increases thereby providing an uncompensated non-linearity in said correction path, a light source modulated with the output of said non-linear circuit to vary its brightness, said light source, which forms part of the electro-optical scanner, scanning said original or color separations and also scanning, through a color separation transparency to be corrected, a photographic emulsion which will provide a printer for the color which is being corrected.

14. Apparatus according to claim 13, in which the color-representing signals which are combined are so adjusted that the resultant signal represents zero correction when the tone represented by the uncorrected signals is a neutral tone.

15. Apparatus according to claim 13 including beamsplitting means whereby separate scanning beams from the light source are transmitted through a number of uncorrected transparencies, a number of light-sensitive cells arranged to receive the light beams passing through the transparencies and providing corresponding electric signals, and means for accommodating the photographic emulsion to be exposed behind and substantially in contact with the uncorrected transparency for which the correction has been computed so that light passing through the latter to the associated light-sensitive device also passes through said emulsion to be exposed.

16. Color reproduction apparatus comprising an electro-optical scanner, a correction path including a circuit in which an electric signal from said scanner representing the transmittance of a transparency corresponding to a color to be corrected is divided by at least one electric signal from said scanner representing the transmittance of a transparency corresponding to a correcting color, a non-linear circuit for receiving the output signal from said dividing circuit and having an input-output characteristic in which the gain is reduced as the input increases thereby providing an uncompensated nonlinearity in said correction path, and a light source the brightness of which varies with the output of said nonlinear circuit, for exposing a photographic emulsion which will provide the printer for the color which is being corrected.

References Cited in the file of this patent UNITED STATES PATENTS 2,710,889 Tobias June 14, 1955 2,721,892 Yule Oct. 25, 1955 2,757,571 Loughren Aug. 7, 1956