US2286730A - Electric circuit for color correction - Google Patents

Description

June 16, 1942, Q HALL 2,286,730

ELECTRIC CIRCUITS FOR COLOR CORRECTION Filed May 14, 1941 5 Sheets-Sheet 1 BLU FIG. I. FILTER I LIGHT VALVE Wu LINEAR NON LINEAR LINEAR AMPLIFIER AMPLIFIER AMPLIFIER I05 1 I5 5 16 i i i i GREEN 1-! Ha LIGHTVALVE FILTER we a l LINEAR AMPLIFIER VINCENT DI'IALL INVENTOR A TTORNE Y June 16, 1942. v. c. HALL 2,286,730

ELECTRIC CIRCUITS FOR COLOR CORRECTION Filed May 14, 1941 5 Sheets-Sheet 2 BLUE F|B.5.

B I CIRCUIT LIGHT vALvE LINEAR Non-LIN A LINEAR [I Q AMPLIFIER- AMPLIFIER AMPLIFIER a A A 155 se GREEN 5 MIL CIRCUIT if; f l 3 LINEAR oR LINEAR Non-LINEAR LINEAR U @AMPLIFIER AMP IFIER AMPLIFIE flaw 11G G I56 i 32 i RED J 31 cIRcuIT LIGHT VALVE LINEAR NON-LINEA LINEAR AMPLIFIER AMPLIFIER AMPLIFIER I l In IsR I(4R R FIG 5 '2 IE2-\/87 82 D I U U o w- 3 8I\ a? a4 i VINCENT E.HALL

INVENTOR I.'o 2.'o 31o mum/fl DENSITY BY A T'TORNE Y June 16, 1942. v. c. HALL ELECTRIC CIRCUITS FOR COLOR CORRECTION 5 Sheets-Sheet 5 Filed May 14, 1941 June 16, c H L ELECTRIC CIRCUITS FOR COLOR CORRECTION Fi l ed May 14, 1941 5 Sheets-Sheet 4 VINCENT l3. HALL INVENTOR BY M2,. M

A TTORNE Y June 16, 1942.

Filed May 14, 1941 5 Sheets-Sheet 5 VINCENT HALL INVENTOR BY W1" r89 8 515% ad r :35 23m :83 EIEE 554 55 2% 52:2 ETE $5 $29 a Q a? n i m n p 16:: E3

A TTORNE Y Patented June 16,

SLATE-S -AT-ENT .aLac'rnrc CIRCUIT'FORCOLOR CORRECTION Vincent 0. Hall, Rochester, N. Y., assignor to Eastman Kodak Company, Rochester, N. corporation oi New Jersey Application May 14, 1941, Serial No. 393,418

19 Claims.

This invention relates to electro-optical systems for the reproduction of multi-colored originals. It relates particularly to a method and means for introducing color correction in such systems.

In a copending application, Serial Number 120,964, filed January 16, 1937, now Patent 2,253,086, A. Murray and R. S. Morse describe the broad idea of introducing the color correction in electro-optical systems of this type. In U. .S. Patent 2,231,669, I have described the application- .of this broad idea to the problem of correcting errors which occur in the duplication ofmonopack color transparencies. 'The present invention may be applied to the specific idea of that patent as well as to the system broadly disclosed by Murray and Morse. It may also be applied to the transmission of colored pictures by wire or televisiom The present application is a continuation in part of my application Serial Number 234,422, filed October 11, 1938, now Patent 2,249,522.

It is an object of the present invention to give improved color correction and better control over the amount and type of color correction than hitherto available.

It is the object of this continuation inpart to provide improvements of these types even to a much higher degree than obtainable with the arrangement described in my parent" application.

The type of electro-optical, systems to which the invention belongs usually include some scanning device and a beam-splitter together with color filters, for example, the three primary colors, for scanning the original in plurality of 1 tive corresponding to the color falling on the photoelectric cell or any type of light controlling devices such as used in reproduction of pictures sent by television. According to the Murray and Morse application above-mentioned, color correction is introduced by modifying the energy in one of the channels in accordance with the corresponding energy in another of the channels. The less preferable systems wherein color separaoriginal color picture (or color separation negation negatives are first made and the color channels are controlled by scanning these negatives simultaneously and synchronously may also benefit by the present invention.

According to the present invention the outputs of the channels are further modified in one or both of two ways; First, the output is made proportional to some power other than unity of the input by including a non-linear amplifier in the channel. Secondly, the modification of the energy in one channel may be in accordance with a non-linear power function of the energy in one of the other channels.

More specifically, one or more of the channels include a non-linear amplifier having an output to input transmission function between the .25 and unit power, i. e; between the fourth root and linear. By modulating the output of the photoelectric cell in another channel in accordance with the output of this non-linear amplifier, a degree of modulation proportional to the transmission function-of the non-linear amplifier is obtained. On the other hand, the energy in a circuit having only linear amplifiers may be used to modify theoutput of this non-linear amplifier and hence the degree of modification in this channel which contains the non-linear amplifier will be proportional to a power between 4 and 1 of the energy in the other channel.

My parent application Serial Number 234,422 referred to. above, related to this broad idea of modifying the energy or signal in one color channel by a non-linear (the specific example being an exponential) function of the signal in another of the channels. The particular example given in that application employed "electrical modification, i. e. modification of the signal while it is ,in the form of electrical energy. In any electrooptical color reproduction system the signals are first setup as light energy variations in the scanning beams reflected or transmitted by the tives) The light energy variations are converted by the. photoelectric cells to electrical energy variations and then are reconverted by the light valves (this term includes glow lamps, ribbon valves operating on light beams, etc.) to light energy variations.

The present continuation-impart relates to an improvement of the species disclosed in the parmathematical theories or found by mathematical analysis of perfectreproductions and secondly to extend the range of reproducibility of this conforming modification.

A mathematical comparison of electro-optical an photographic systems of color reproduction aids in understanding-the reasons for the broad invention and in differentiating the various forms thereof. As an example of electro-optical systems, the case where the original is scanned directly (no uncorrected negatives being made) and where color corrected negatives are exposed by controlled light valves, is analysed in comparison with purely photographic systems of making color separation negatives, the following features being analogous. The reflectivity or "transmission of the original to red, green and blue is the same in both cases. The exposure" of the photographic negative and the exposure of the photoelectric cell are the same and are directly proportional to the original reflectivity or transmission. In the electrical case the signal is proportional to the exposure of the cell and any number of linear amplifiers in the circuit merely changes the proportionality factor. In the photographic case, on the other hand, the negative density is not linearly proportional to the exposure but follows a characteristic curve, part of which involves a logarithmic relationship between density and exposure 1. e. a linear relationship between density and log exposure. The slope of this characteristic curve is the contrast of the negative relative to the original and the value of this slope for the linear portion thereof is called gamma.

Even when there is no color correction, the gamma of each color component must be controlled and balanced against that of the other colors.

Furthermore color correction of the masking type when performed photographically involves the superposition in register (masking) of a high or normal contrast color separation negative and a low contrast positive corresponding to a different color. In fact in every type of color correction of practical importance, it is desirable and necessary to be able to control the gamma (contrast) and/or similar features.

The present invention provides in electrooptical systems a control similar to, but in many ways better than, the control of gamma or contrast in a purely photographic system of color correction.

Mathematically masking consists of the addition of densities one of which is the density of a positive and the other of a negative. equivalent to the subtraction of quantities which are logarithmically or exponentially proportional to exposures (i. e. to original refiectivities or original transmissions). It is also equivalent to This is the division of quantities linearly proportional to the modification is equivalent to the subtraction of quantities exponentially proportional to exposures. It should be understood that simple modulation involving the multiplication of one signal (linearly proportional to exposure) by a quantity negatively linearly proportional to another signal would not simulate masking. Basically the "masking type of correction is the preferred one, but slight improvements thereon (e. g. to correct for the toe and shoulder of the characteristic curve of the original or for the toe and shoulder of the curve of the negative materials being exposed; or for disproportionate losses in signal transmission, especially in television; or for any of the other factors not peculiar to color reproduction but present even in black and white systems and/or for correcting for those lesser errors present even in purely additive processes of color reproduction), may be combined with pure masking to give more accurate and pleasing color reproduction. The present invention permits the practical realization of such combinations which were not possible with simple elect'ro-optical systems or with photographic systems.

The specific embodiment of this continuationin-part gives the subtraction of quantities exponentially proportional to exposures over a range greater than that obtained with the specific arrangement of the parent application and furthermore even the small variations from pure subtraction of quantities exponentially proportional to exposures are such as to aid in the correction of the minor errors listed above. This specific embodiment (the continuation-impart) consists of optical modulation using two light valves in cooperative optical tandem, i. e. two valves operating simultaneously on the same light beam, the valves being controlled by the energy from difierent color channels. The optical tandem arrangement may be two ribbon-type valves in series with respect to the light beam, it may be a glow lamp whose output is modified by a ribbon-type valve or othervariable aperture or it may be a ribbon valve with two or more independently operated ribbons in the same magnetic field as described in application Number 417,540, filed Nov. 11, 1941, by John Streifiert and myself. By way of precise definition, it is pointed out that the three valves making the black printer in Fig. 5 of U. S. 2,183,524, Yule, are not in cooperative optical tandem since they operate alternately not simultaneously on the light beam as is necessary to the present invention.

Furthermore each of the valves of this tandem pair may separately consist of two valves in tandem to give increased range as described in my copending application Serial Number 378,781 filed Feb. 13, 1941.

Light valves in optical tandem are not per se new, but the advantages of the present invention are due to the combination of this feature with the non-linear amplifier invention of my parent application. These specific advantages became apparent to me only after careful investigation of all modification of the broad invention including that specifically disclosed in the parent application. The following mathematical analysis indicates that the present embodiment and that of the parent application are theoretically identical and hence the reason why the present embodiment is so superior could not be predicted and cannot even now be explained mathematically.

The color components of an original are in When . 3 correction better than the best obtainable with purely photographic systems.

One embodiment of my invention combines both electrical and optical tandem valve types of correction must be in the direction which simulates masking.

In purely photographic masking R K 'a b (m)| l l+ 3 8 r a- RIL 0.2.

where all C's are constants, Db and D; are the densities of the blue and green negatives uncorrected, D(pos)g is the density of the positive mask made from the uncorrected green negative, D's is the density and T'h the transmission of the corrected blue negative, Rb and R; are the refiectivities or transmissions of the original to blue and green respectively and K and L are the gammas to which the negative and mask are respectively developed relatively to the original.

In the arrangement taught in my parent application, the basic equation, without regard to the variation due to circuit factors and the form of the light valve response, is:

's b wi EIN) where E'b is the corrected blue signal, Eb and Eg are the uncorrected blue and green signals and M and N are the exponential coefflcients of the non linear amplifiers. The signal E'b passing through a light valve controls the exposure of a corrected blue negative whose density is therefore:

D's P.1og (C5EDMIZC4 END where P is the gamma to which this negative is developed.

Similarly with light valves in optical the basic equation is D,, P.1og (C' E wi E,"]) which is identical to the one for purely electrical correction.

However it has been found practically that either system gives results which simulate masking over a wide range of energies (i. e. greater than 30 to 1 and up to 1000 to 1 and more, which corresponds to a range of densities up to 0 to 3 and even greater.) Obtaining these greater ranges with the system described in 'my parent application is extremely dimcult. The optical tandem system however simulates the masking more closely and over a more extended range with much simpler apparatus. Furthermore at the ends of the range I can now control the departures from true masking both in directionand amount to be useful for the correction of some of the minor errors in color reproduction systems. rendition for the various hues and tints, the optical tandem system has greater stability, greater uniformity and numerous other factors tandem,

all tending toward better color rendition. In

fact, not only can this system accurately simulate masking over a range far greater than is In addition to more accurate color.

correction. In this case both corrections may be of the masking type or one of them may be of a different type to correct for minor errors such as those present in additive processes and hence present to a certain degree in subtractive processes.

The objectsand advantages of the invention and the invention will be more clearly understood from the following description when read in connection with the accompanying drawings in which:

Fig. 1 shows-an adjustable form of the invention.

Fig. 2 shows one form of electric circuit which can be used to provide the embodiment shown in Fig.1.

Fig. 3 shows another embodiment of the invention involving various forms thereof.

Fig. 4 shows the circuit for an embodiment of the invention involving light valves in optical tandem.

Fig. 5 illustrates one form of optical tandem light valves.

Fig. 6 illustrates another form of optical tan- .dem light valves.

Fig. 7 illustrates a third form of optical tandem light valves.

Fig. 8 is a graph of light valve response vs. densities of the original.

Fig. 9 is a detailed circuit drawing of the arrangement shown in Fig. 4.

Fig. 10 shows a preferred embodiment of the invention.

In Fig. 1 the green channel includes a green filter IIJG through which a corresponding scanning beam falls on a photo-electric cell HG hav- 7 ing a linear amplifying circuit IZG, a. non-linear linear amplifier I33, and a light valve I5B. Ac-

cording to Murray and Morse the energy in the blue channel should be modified in accordance with the energy in the green channel. Applied to this Fig. 1, their system would be such that the output of the linear amplifier IZGthrough a suitable modifier such as shown at' IG would modify the output of the linearv amplifier I213. According to the present invention the energy in the blue channel is modified in accordance with some non-linear power function of the energy in the green channel. If, for example, the non-linear amplifier I3G has a square root transmission function whereby its output is proportional to the square root of its input, the output of the linear amplifier I2B will be modified through a modifier l8 in accordance with the square root of the energy in the amplifier IZG.

Non-linear amplifiers are. well known and the type here described are those which employ variable resistances (29 in Fig. 2) commonly called varistors. Such arrangements have an output which is a constant exponential function of the input over a wide range of input intensities. Since color pictures have densities up to and above 3, the refiectivities or transmissions vary over a range of 1000:1 or more. Therefore to effect a change in a signal from one gamma to fact that density is a logarithmic function and that 3 is the log of 1000 and 1.5 is. approximately the log of 30.

By suitable adjustment of the resistances in the modifiers I6 and I8 the modification of the energy in the blue channel may be made proportional to any desired additive combination of the linear function and the square root function of the energy in the green channel.

The modification may be linear subtraction e. g. by merely superimposing the potentials corresponding to the energies in opposite directions across a'resistance. It may simulate division e. g. by adjusting a resistance in the blue channel in accordance with theenergy in the green channel. It may be multiplication either directly or as shown in Fig. 2, where the energy in the blue channel is multiplied by a constant minus the modifying energy which may for example be proportional to the square root of the energy in the green channel. The term modify is used to include all such modifications or embodiments. i

If the non-linear amplifier BB in .the bin channel has a transmission factor for example according to the third root power, a modifier ll .which introduces the output of the linear amplifier I2B into the linear amplifier MB will effectively modify the output of the blue channel in accordance with the third power of the energy in the green channel. A fourth modifier l9 would in the example taken permit modification of the energy in the blue channel in accordance with the three halves power of the energy in the green channel. In all of these cases the modification is uniform over practically the whole range of intensities so that, over the major portion of the total latitude, the linearity of density relationships is maintained at least approximately. Since density is logarithmic this maintenanceof linearity with respect thereto is really a maintenance of the constant of non-linearity wtih respect to intensities themselves, e. g. a constancy of the exponent in an exponential function. As pointed out, specific variations from this constancy may be introduced to correct for minor errors such as the toe and shoulder of the reproduction curves or the errors inherent in additive processes. These latter specific variations are easily obtainable only with the combination of non-linear amplifiers and optical tandem light valves here disclosed.

If it is desired to modify the energy in the blue channel in accordance with some nonlinear power function of the energy in the green v channel, but to have the light valve I5G react in direct proportion to the input from the photoelectric cell HG, the light valve I5G may be operated directly from the linear amplifier I26: and

circuit corresponding to Fig. .1. A portion of the output of the photoelectric cell IIG is taken off through a condenser 20 and a rectifier 22 to modulate the energy in the bluecircuit in the linear amplifier i2B before the non-linear amplifier I328 and/or alternatively in the linear amplifier B after the non-linear amplifier 133. Similarly a portion of the output of the linear amplifier G is taken off through a condenser 2i and a rectifier 22' to modify the energy in the blue channel at either or both of the linear amplifying stages I23 and B. The potentials on the points 23, 24, 25, and 25 depend respectively on the output of the condensers 20 and 2| and the setting of the variable resistances l1, l6, l9, and I8. If, as in the example discussed in connection with Fig. 1, the non-linear amplifiers I3G and I313 have transmisison functions respectively according to the square root and third root powers, this arrangement shown in Fig. 2 is completely adjustable whereby the output of the blue circuit may be modified by the energy of the green circuit in accordance with any additive combination of linear, square root, three-halves and third powers. More correctly the output of the blue channel will be in accordance with the third root power of the input of the blue channel modified by a function of the energy in the green channel which includes additively, a linear function, a square root 'wise almost any type of modifying is possible by suitably adjusting the tube responses. The rectifiers 22 and 22 act merely as rectifiers and have no appreciable effect on the non-linearity which is established in the non-linear amplifier I33 and 13G. The varistor 29 in this unit includes two copper oxide rectifiers but may be any of the other well known types of varistors, such selenium type, opposed diodes type or satu rated core transformers.

Fig. 3 shows a preferred embodiment of the invention wherein through suitable modifiers 30, 3|, 32, and 36 various modification functions are introduced. The .light valve I5R of the red channel varies according to some power function, introduced by the non-linear amplifier I3R, of the output of the photoelectric cell HR. The light valve linearly with the output of the photoelectric cell HG or if a non-linear function thereof is desired, the amplifier 35G by which the light valve I5G is operated may'include a non-linear stage. The output of the photoelectric cell HG,

.ishowever modified linearly by the modifier 30 in accordance with the output of the photoelectric cell HR and non-linearly by the modifier 32 in accordance with the output of the nonlinear amplifier I3R. The light valve I5B in the blue circuit operates non-linearly in accordance with the output of the photoelectric cell HB passing through the non-linearamplifier NB. The energy in this channel is modified according to the inverse power of the non-linear amplifier I3B by a modifier 3| which introduces the linear .function of the output of the photoelectric cell HR to the final stage' of linear amplification MB in the blue channel. This energy is also modified non-linearly in accordance with a nonlinear amplifier 33G which operates on a portion of the output of the linear amplifier I2G in the I5G in the green channel varies aasopso D green channel and introduces its non-linear modification through amplifier MG and modifier 38 into the linear amplifier I23.

By dividing the output from linear amplifier I2G it is possible to have the light valve G operate linearly therewith or according to some non-linear function which is different from the non-linear function used in the modification of energy in the green channel, gives satisfactory results. This, of course, reduces the system to a relatively simple form.

For the convenience of those accustomed to I think in terms of the masking method of color correction, the following analogy is given. If the above modification is a direct division, this means that the green intensity is divided by the .45 power of the red intensity and that the blue intensity is divided by the .40 power of the green intensity. If the signal intensities were then used directly to control the exposures, this would be the same for example as masking the blue negative by a positive of the green negative which positive is developed to a gamma of .40. How

ever all of these signal intensities are also reduced to the .40 power before controlling their respective light valves (which are assumed here to be linear as is generally the case, but which may have any desired response function). Therefore the total contrast of the printing light is reduced by this power which compensates for the high contrast due to the processing of certain types of monopack films.

For example the blue printing light is propor-' tional to the .40 power of the original blue scanning light minus (.40X.40) power (i. e. the .16 power) of the original green scanning light. The gamma to which the yellow layer is developed or in photomechanical processes, the gamma towhich the yellow printer is developed and prints, is about 2.5. This has been found to give correct tone reproduction in the final reproduction.

The more complicated cases where the above modification is not a direct division and/or the .valve response is not linear are almost infinite in number and are increasingly more difiicult to explain mathematically. However, it is obvious that something similar or analogous to this reduction in contrast and to this masking with a correcting mask of lower contrast takes place in various types of modification.

For example with multiplication (as in Fig. 2) the masking instead of being by a .40 contrast mask which is a direct positive of the green negative, is by a .40 contrast mask which is a slightly complicated positive function of the green negative. Thus in every case there is something analogous to the reduction in contrast in ordinary masking.

justable resistance 28L By using a resistance of 100 ohms for the resistance 21 and impedance equal to 2000 ohms in the output of this nonlinear amplifier, i. e. the primary coil of the transformer between the non-linear amplifier ISG and thelinear amplifier MG, and by using varistor 29 whose resistance .varies from 6 of an ohm to 100 ohms depending on the currentfiowin therethrough, which in turn depends upon the setting of the adjustable resistance 28 and the intensity of the input, it is possible to provide an output which varies proportionally to a power function of the input, which power is between .25 and 1.

In Fig. 4 the blue and green channels are each shown as containing linear amplifiers I213 and IZG, band-pass filters B and 40G, non-linear amplifiers 13B and I3G, another stage of linear amplification shown at B and HG, a linear, rectifier shown at B and MG and a low-pass filter at 423 and "G. The light modulating devices are shown at 433 and 43G respectively. For the modulation of the blue signal by the green signal, the green circuit includes a branch circuit connected to the output of the linear amplifier I26 and including a non-linear amplifier G, a linear rectifier 45G, a low-pass filter 46G and another stage of amplification,

in this case a D. C. amplifier G, which operates a glow lamp G. The output of the rectifier 45G is a pulsating direct current. The high frequency componentsv of these pulsations are stopped by the filter 46G and the amplifier 41G is'designed 'to' amplify uniformly D. C. or low frequencies passed by 46G (the range of frequencies may be for example 0 to 1000 cycles per second). The glow lamp 48G and the light valve 433 are in optical tandem, the. light from the glow lamp 48G being focused by an optical sys-.

tem 5| on the light valve 433 and then refocused by an optical system 52 onto a sensitive film 53 mounted on a drum for scanning in the usual way. Alternatively, the blue signal may be fed into a glow lamp and the modulator signal fed into a ribbon type valve so that the two valves are interchanged relative to theposition shown in this Fig. 4. However, I find it preferable to impose the correction current, i. e. the green modulator signal, on a glow lamp and to have the main signal on a ribbon type valve as shown at 633.

With the optical tandem valve type of color correction shown, the band-pass filter 40B is-unnecessary. It is quite useful to have such electrical filters present, however, if one wishes to combine the optical tandem type of correction stituting a varistor, and in series with the ad- 76 shown in this Fig. 4 with the purely electrical correction shown in Figs. 1 to 3. To do this, the output of the modulator circuit may also be connected by leads shown by broken line 50 into the linear amplifier I2B in exactly the same way as shown in Fig. 3. 01 course, added stages of amplification may be included at any suitable point in the circuit such as in the leads 50. The lowpass filters shown in this figure may conveniently be arranged to pass frequencies less than 1000 cycles. In this case, the band-pass filter 40B is arranged for example to transmit frequencies between 1400 and 3400 cycles. Obviously, the band-pass filter "B may be in the circuit anywhere between the linear amplifier I23 and the rectifier MB. The arrangement shown, happens to be the one which has been found to give satisfactory results. Even when the correction is introduced entirely by having the light valves in optical tandem, there is no harm in including a band-pass filter 40B and there is some advantage because this filter 40B cuts down noise and thus extends the range of clear signals.

When the combination of optical tandem light valve correction and purely electrical correction is employed, both corrections may be of the masking type, i. e. the subtraction of densities or alternatively may have various forms. I have found that the optical tandem light valve system simu- 10 lates masking very closely over a very extended range. If the leads 50 are connected to the linear amplifier I2B as shown in Fig. 2, this type of correction is available over an even a more extended range. output) of the light valve 433 is modified both by the electrical correction through leads 50 and by the optical correction through the glow lamp "G.

If the leads 50 are connected into the linear 2o amplifier I2B so as to give a linear subtraction of the signals, the minor errors which are present even in additive processes may be corrected simultaneously with a masking by the glow lamp 48G.

Here masking is used in the sense of "correction tensity. The two valves may be at any angle to by the subtraction of densities. Even for linear subtraction of signals, the control of contrast by non-linear amplifiers is useful but is not as important as in masking. Similarly it may be possible to have both the optical tandem light valves, and the electrical correction through the leads 50 arranged generally for masking, but to deviate slightly therefrom to correct for minor errors such as the toe and shoulder of the reproduction curves, etc. For example, to increase the contrast at low densities only and still maintain the same percentage of masking, the resistances (such as 28 of Fig. 2) in series with the varistors of the non-linear amplifiers G and HE, are slightly increased from-the normal valve (0 to 2 ohms) up to say 8 to 12 ohms.

The linear amplifier "G is of course not absolutely necessary, but it is more convenient to. operate the glow lamp lBB-from the output of such an amplifier. The optical system 52 is arranged so as to provide a scanning spot of variable intensity rather than variable area. A variable area system would be preferable if a halftone record were being made. Other forms of optical tandem light valves are shown in Figs. 5 to 7 and 60 a detailed arrangement of the electrical parts in this Fig. 4 is shown in Fig. 9. All of these figures relate specifically to the correction of the blue signal in accordance with the intensity of the green signal since this is the commonest cor- 55 rection of the masking type. Obviously, the same principle is applicable to the correction of any of the signals, When it is desirable to modify the blue signal by both the green and the red signals, three valves are used in optical series. That is, the blue channel light valve 433 would be placed in optical tandem with both a green modulator light valve and a red modulator light valve.

Int-he exposing of a negative 53, the light valve 433 is arranged to increase in transmission with 5 increasing blue signal. The glow lamp 486 is arranged to decrease in intensity with increasing green signal. The forms of this response, are described in connection with thegraph shown in Fig- 8.

In Fig. 5 a light source 60 focuses a spot by 7 means of an optical system 6| on the aperture an: a ribbon type light valve 63 which is connected to the output of the blue channel 62. The separation of the ribbons of the light valve 63 de- That is, the response (eifectivelli termines the amount of light transmitted through this valve. For convenience the magnet providing the magnetic field for these ribbons is not shown; but may be of any of the usual types. The light from this valve 63 is refocused by a lens 65 on another ribbon type valve 61 having an aperture 68 and connected to the green modulator circuit 66. So that the total transmission of these two ribbon type light valves 63 and 61, in optical tandem as shown, will be the product of the individual responses of these valves, theyare arranged at right angles to one another. They may thus be made to simulate masking, even with a linear response of the valves. Obviously, if a variable width image of the valve 63 were focused on the valve 61, with its ribbons parallel to the valve 63, the output of the two'valves would not be the product of the separate ones. This output is focused by a lens 69 onto a sensitive film II mounted for scanning on a cylinder or drum I0.

Fig. 6 shows an alternative arrangement wherein the light valves 63 and 61 have their ribbons parallel, but the light striking the valve 61 is not an variable width image of the valve 63. The light incident on the valve 61 has variable inone another and are shown parallel for comparison with Fig. 5. A lens I2, acting as a field lens is placed adjacent to the light valve 61 to focus an image of the valve 63 in a relay lens I3 which is positioned to focus the valve 61 on the sensitive film II. This arrangement produces on the film -'II a spot whose intensity is controlled as the product of the two responses of the light valves 63 and 61 or of the responses of all three valves if 18 is present. A change of azimuth of the valve 18 from this right angle position to a parallel one would result in a scanning spot whose intensity is the product of the response of valve 61 and valve 63 or 18 whichever is smaller, but intermediate azimuthal positions of valve I8 give more useful functions.

' Still another arrangement of light valves in optical tandem is shown in Fig. 7 wherein both valves are placed in the same magnetic field. Although the ribbons are quite close to one another, they are still in optical tandem. A single ribbon I4 is connected to the output of the blue channel 62 and a single ribbon 15 at right angles thereto is connected to the output of the green modulator circuit 66. The aperture of the valve is indicated by the square 16, but as before, the magnet providing the magnetic field is not shown. This embodiment is described in the above mentioned copending application by Streifiert and myself. With the ribbons at right angles to one another, the joint response is the product of the separate responses, as required in the masking type of color correction.

In Fig. 8, the logarithm of the current operating the final light valve is plotted against the density of the original picture being scanned.

This discussion is given in terms of the purely to. The abscissa of the graph may be the I density to blue or the density to green.

Considering first the blue valve only, the

logarithm of the current input thereto will be linearly negatively proportional to the blue den- 3 sity of the original. This is shown by the curve (actually a straight line) in this Figure 8. If the original has zero blue density, i. e. reflects blue light completely, the logarithm of the current into the blue light valve has a value slightly greater than 3.0 as shown. These values are selected by way of example from actual operating conditions. Increasing blue -density reduces this log current as shown by the straight line 80. Since this line 80 does not take into account any effect of the green modulating circuit, it is also the curve for the case where there is no green signal, i. e. the case where the original has an infinite density to green (or for praclical purposes a density greater than 3.5 Say.)

On the other hand, if the original has zero a scale having equal densities to blue and green,

the response should be that shown by the line 83. It will be noted that for zero blue density and zero green density this line 83 coincides with the line 8| and for maximum blue density and maximum green density, it coincides with the line 80.

On the other hand, the horizontal straight line 82 shows the corresponding response for a gray scale when the non-linear amplifiers of the present invention are omitted. In other words, the line 83 is what one should getfand does get by the present invention, whereas in the absence rf the present invention, the horizontal line 82 is the result. The reason that the line 82 is horizontal, is apparent from the fact that any decrease in current due to increased blue density is offset by an increase in sensitivity of the amplifier due to the equally increased density-of the correcting color green. Of course the opticaltandem system of the present invention, does not involve any change in sensitivity of amplifiers, but the analogous situation holds, since if the valves were not operated by non-linear amplifiers,'the net responses of the blueand green valves would be equal and opposite. That is, as one went to higher densities, one of thevalves would close and the other would open by an equal factor. All of this discussion of lines 82' and 83 is in connection with the gray scale, where the blue and green densities are equal.

It will be noticed that for zero density in both the green and blue the log current is reduced from about 3.5 to 1.75. which results when the control voltage for modulating the blue signal is the square root of the green signal. Thus for any particular density, the log current on the line 83 is one-half of the log current on the line 80. By way of distinguishing between the blue and green signals it is pointed out for square root control, the amount of, reduction (in termsof density) is one half of the green signal. For zero green density, the green signal is 3.5; therefore the correction is one-half of this and equals 1.75. This correction applied to any blue signal reduces it to the line 8|. For zero blue density, the blue signal is This is the situation I the actual result of the unbalance in the control circuit which shows up at densities greater than 2.0, since it is impomible in practice to hold the two modulating tubes balanced over a range of over decibels electrically i. e. an intensity range of 100 to 1 which corresponds to a density range of 2.0. This deviation shown by the curve 84 is not completely intolerable for the gray scale and in fact the corresponding trouble for other colors is not objectionable in most cases. The gamut of colors which are accurately reproduced, is limited, since if the gr en density were zero, the curve for various sh dos and densities of blue is that shown by the brdken line as. This is due to the fact that in practice when the green density is zero, the output of the controlledamplifier of the blue channel can never fall below that corresponding to the control color (green) density of 2.0; In other wordsgthe curve 85 never falls below the log current value given by the curve 83 at the density of 2.0, Such difllculties appear .only when the original has a low green density and are objectionable only when italso has a high blue density; thus the purer greens are" not accurately and fully corrected.

Even without going over to the optical tande valve type of control, some of these difliculties and errors are reduced by the expedient which is common in radio practice when one wishes tov distinguish between different signals, namely, the use of a band-pass filter as shown in Fig. 4. Referring back to this Fig. 4, the addition of the low-pass filter G in the control circuit before the electrical modification through the leads 50 is. introduced into the linear amplifier 12B, is

possible because the 'rate of change of amplification in this linear amplifier I2B need never be faster than that corresponding to the finest detail it is desired to reproduce. Therefore, since it is possible to use such a low-pass filter, a bandpass filter MJB may be introduced in the blue channel so that the signal being controlled is amplified at some carrier frequency, e. g. 2400 cycles plus or minus 1000 cycles, and thus it is possible to discriminate between the two channels by this combination of a low-pass filter; going into the circuit'and the band-pass filter' in the output. Obviously, a high-pass filter allowing all frequencies over 1200 cycles to be amplified would be just as satisfactory as a band-pass filter, as far as discrimination between the two signals is concerned, but the cutting out of all frequencies above the carrier plus its side'band (3400 cycles in the above example) reduces noise. The word noise is here used in the electrical sense and corresponds to the factor which causes noise in sound reproduction circuits.

However, even with the electrical filters to dis- 'criminate between the signals, the gamut of colors which can be satisfactorily corrected, is more limited than that in the embodiment of the in- The example discussed above in connection with Fig. 4 wherein, at low densities, the contrasts of both the blue signal and the green modifying signal are increased but the degreeof masking is maintained constant, results in the line, deviating as shown by thebranch curve 86 and the line 83 deviating as shown by the branch 81. Such a modification is sometimes useful in certain photomechanical processes.

Fig. 9 shows the circuit details corresponding to the diagrams given in Fig. 4. Of course each of the stages of amplification may be much more complicated than that shown and it is customary to purchase the various units shown as boxes in Fig. 4 as complete units.

In Fig. 10, the figures correspond to those given in Fig. 4 with the addition of a red channel and' scanning drums 53B, 53G and 53B for all of the channels. In this skeleton circuit drawing, the blue signal is corrected by the green and the green signal is corrected by the red. Since the red signal is in general not corrected, a uniform light source 49 is used instead of a glow lamp to illuminate the light valve 43R.

In each branch of the circuit the box representing a non-linear amplifier contains a legend EB" etc. This is intended to indicate that the blue energy (blue signal) is non-linearly amplified to an exponential function wherein the exponent is v. The circuits are actually arranged (i. e. resistances such as 28 in Fig. 2 are adjusted) so that v, a: and z are substantially constant, but w and y vary slightly, decreasing as E increases.

One particular arrangement which I use will now be described so as to illustrate how masking is simulated. Having arranged the circuits to give the degree of masking required by some particular printing inks -blue to be masked by a 50% green and green to be masked byja 50% red mask, I tested each branch separately. Since only one valve (43R) controls the light spot scanning the red negative, the gray scale gamma of this negative is a measure of the exponent z.

Of course the development of the negative also affects its gamma and must therefore be taken into account. The development of all three negatives is preferably done simultaneously and hence the negatives should all have the same gray scale contrast of exposure.

Setting the valve 43G wide open, the effect of the non-linear amplifier R can be separately recorded on the negative scanned at 53G. Similarly constant light from the glow lamp 48R provides means for separately testing and measuring the effect of the non-linear amplifier I3G.

In the example just discussed, my measurements show that when linearly responsive light valves are used at 433, 43G and 43R, the nonlinear amplifiers I3B, 13G, and HR should be purely exponential amplifiers with the exponents v and a: respectively equal to about .6 to 1 and 2 about .3 to .7. The absolute values of these ex ponents are not critical since the processing of the final negatives can multiply each of them by a factor. The relative values are critical if the negatives are to be processed simultaneously which is 'of course the most convenient way of getting reproduceable results. The diflerent absolute values are important however when the range of the apparatus is considered.

If these circuits involved the actual division of signals E B e. g. vw as required theoretically by masking, several relationships between the exponents v, w, :r, y and 2 could be easily written down. Letting V,'W, X, Y and Z (upper case) be the theoretical exponents used in the theoretical case of pure division of signals, it is noted that V and X'could be unity, W and Y could have the value required for masking e. g. .5 for 50% masking and then Z would have to equal V-W and X'-Y in order that. the contrast of the grayscale in the three negativeswould be exactly the same. This is all well known in ordinary photographic masking processes. The upper case exponent is intended to refer to the exponent of the signal which actually exposes the finalfilm and this is sometimes referred to as the exposure gamma of signal.

Of course since absolute contrast depends on the processing, the values of V and X can vary over areasonable range say .6 to 1 with linear light valves.

However'one case of great importance is that using two light valves in optical tandem and electrical series in place of each of the single valves; either a single or double valve is indicated by the single box 433 for example. Such double valves as that shown in Fig. 7 but connected in series and to one channel only or as those shown in U. S. 2,137,267 Cowley and U. S. 1,746,729 Ives have the effect of squaring the exposure i. e. of doubling the exposure gamma. When these valves are used the values of V, X, etc., can go up to 2. i In practice, a range up to 1.4 or 1.6 has been found to be useful. Therefore since the effect of the light valve can be considered to be part of the circuit as far as these theoretical exponents are concerned, V and X may have a range of .6 to 1.6. I preferably use about.1.2. A .6 amplifier and a double valve system for 43B and 43G would give this value exactly if true division were used-the values of v and :1: actually used to simulate this are approximately constant and equal to V and X when linear valves are used or equal to V/2 and X/2 when squaring valves are used.

For useful masking the theoretical exponents W and Y are usually approximately the same and range between .3 and .7 depending on the how much masking is required for the particular inks being used. (I prefer to use .6 for each of these to give a 50% mask with V and X equal to 1.2.)

Now since V-W and XY must be equal and must equal Z, Z must range between .6-.7 .6 and 1.6-.3X1. 6. That is, between .18 and 1.1. Since the preferred types of non-linear amplifiers do not give a value below .25, this range is between .25 and 1.1 which means that when V and X are as small as .6 one cannot use a 70% mask since anything greater than a 58% mask reduces the gray scale from .6 to below .25.

.If W and Y are slightly different, (to give different percentages of masking) then V and X will have to differ by the same amount to keep V-W equal to XY (for balancing the gray given previously must be made to simulate masking which is represented by R K D =log (I aii Since a rigorously accurate mathematical description of the effect would be so involved as to be confusing and useless, the practical and preferred system only is described here. In practice I arrange the non-linear amplifiers MG and MR so that the exponential functions Eg and E1 are not simple ones but involve exponents which vary slightly with input, decreasing for higher energies. In practice this merely involves adjustment of the resistances and the potentials on the tubes asis well known to those familiarwith non-linear amplifiers.

The simplest case isthe red channelwhichinvolvesno correction. Therefore 3 equals Z-orif 1' F -7 in series to thebluechannclionly ori. This means z may utilize the whole of its available range .25 to 1.0. From tests I have found,

Z involves the squaring valves, 2 must equal .Z/2.

the. preferred'range to be .25--.6v when using the same type (linear or squaring) "valves in all three.

channels.- If alinear valve isused inthe red channel and squaring valves in the other chan nels.-this range for amight well go upto -Inthe preferred example, a" 50%-mask is used both for blue by-green and green by red,-and

0f the figures is replaced, by two such. valves in optical tandem and electrical series, a squaring valve results. For example .the valve 433 could be made by connecting valves 63 andvlil of Fig.

7 by connecting ribbons T5 and 16 of Fig. 7- in series. Similarly the glow lamp 48G could be replaced by a glow lamp and ribbon valve in optical tandem; and electrical series.

l0 Havingthus described my invention, I wish to.

point outthat it is not limited to the specific I .structure shown but is of the scope-of the-appended claims.

. What lclaim and desire-to secure by Letters.

squaring valves are used in all three channels 115 Patent-ofthe UnitedStates is:

with glow lamps (which are essentially linear valves) -in the correction branches'of the channels. In this example using a 50% mask with =X=1.2, u must equal a: which equals .6 and VW'and XY must equal .6 and z=Z/2 =.3. For a 40% mask'W andY equal .4x1.2 equals .48, V-W,' XY, and Z equal:.72, and 2:36.

By similar tests I find the preferred range for v and a: to be .4 to .8. If the red channel'. only" has a linear valve, the presence of squaring valves in the other channels allows this preferred range to go down to .25. Y e

. In the above. mentioned preferred. example v=:r=.6. For the 40% mask system v=:c'-=.6 .as before since this isthe valve which is selected to give the most useful printing contrast. Without masking this contrast (exposure gamma) equals 1.2, due to the squaring valves. For the gray scale it is reduced to Z and equals .6 for a 50% mask system and .72 for a 40% mask system.

Masking systems from 30% to 70% cover practically all useful systems and all commercially available printing inks, dyes etc.

It is noted that v and a: are equal to V/2 and X/2 even though a purely divisional electric system is not used. It is of course possible to have 1: and a: vary slightly with energy input but this has not proved to be as useful as the case where they are constant.

And again testing the efiect of the glow lamps 18G and 48R alone when the amplifiers MG and 64B are set as used in the preferred embodiment I find that w and y are preferably between .3 and .8 and that they actually vary within this range depending on the input. 7

In the preferred example, a 50% masking sy tem requires the average value of w and y to be a little less than .6, which is the value the constant theoretical exponents W and Y must have to reduce V and X to half their values. Similarly a 40% masking system requires the average values of w and y to be about .4 whereas W and Y theoretically should be .48 (i. e. 40% of 1.2)

If an optical tandem system such as shown in Fig. 5 is used with each of the valve systems of the squaring type instead of having the linear glow lamp in the correction circuits, to and. 0 would be lower. Since the varistors described do not conveniently give values below .25 a range of .15 to, .4 is noteasily obtained. Hence it is preferable to raise this range to .25 to .66 and to raise v and :1: to 1. Since such an embodiment could use linear amplifiers at BB and I3G it is a useful one, but I have found the other system somewhat preferable.

I Squaring type light valves can be made oftwo linear valves in optical tandem and connected electrically in series. Thus if each valve in any 1. An electro-optical system for the reproduction of a multicolored original comprising means for: establishing in separate electric channels,

electric energies corresponding to the color com- 20 ponents of eachpoint of the original in scanning succession, each channel correspondingto one color, at. least one of the channels including a non-linear amplifier having means therein for varying the energy output thereof in accordance with an exponential function of the input there-' of substantially uniformly over a wide range of energies, two light modulating devices in cooperative optical tandem, one device being connected to and operated by the output of; the non-linear amplifier and the other of the two devices-being. connected to and operated by the output'of another of the channels, and means for establishmg a light beam modulated simultaneously by said two devices in optical tande 7 2. An electro-optical system for the reproduction of a multicolored original comprising means for establishing in separate electric channels, electric energies corresponding to the color components of each point of the original in scan-v 40'ning succession, each channel corresponding to put thereof substantially uniformly over a wide range of energies, means for supporting a sensitive layer, means for scanning the layer with a light beam and two light modulating devices in -cooperative optical tandem for simultaneously modulating said light beam, one of the devices being connected to and operated by the output of the non-linear amplifier and the other device being connected to and operated by the output of one'of the channels other than that including said non-linear amplifier.

3. An electro-optical system for the reproduction of a multicolored original, having a plurality of channels respectively carrying energies corresponding to color components of the original and each including a linear amplifying circuit connected ot the output of a photoelectric cell, at least one of the channels having a non-linear amplifier connected to the output 0t its linear amplifying circuit, said non-linear amplifier ineluding means for producing an energy variation in accordance with an exponential function which is substantially uniform over a wide range of energies, means for supporting a sensitive layer, means for scanning the layer with a light beam and two light modulating devices in coopconnected to and operated by the output of one of the channels other than that including said non-linear amplifier.

4. An electro-optical system for the reproduction of a multi-colored original comprising means for establishing in separate electric channels, electric energies corresponding ot the color components of each point of the original in scanning succession, each channel' corresponding to one color, at least one of the channels including a non-linear amplifier having means for varying the energy therein in accordance with an exponential function substantially uniformly over a wide range of energies, light valves for controlling light intensity connected to the outputo f each of the channels, and a light valve connected to and operated by the output of said non-linear amplifier in cooperative optical tandem with the light valve in another of the channels for modifying the response of the latter valve in accordance with the energy in the one of the channels including the non-linear amplifier.

7 5. An electro-optical system for the reproduction of a multi-colored original comprising means for scanning the original in a plurality of colors, photo-electric devices adapted to receive the separate colors from the original for establishing electric energies corresponding thereto, electric amplifying circuits connected to the outputs of the photoelectric devices, at least ;one of these circuits including a non-linear amin another of the channels for modifying the response of the latter valve in accordance with the energy in the one of the channels including the nonlinear amplifier.

6. An electro-optical system for the reproduction of a multi-colored original comprising means for establishing in an electric channel a signal corresponding to primary blue light from each point of the original in scanning succession, means for establishing in a. second electric channel a signal similarly corresponding to one of the other primary colors from each point of the original, a non-linear amplifier connected to one of the two channels for amplifying the signal thereinsubstantially exponentially over a wide range'of signal energies, two light modulating devices in cooperative optical tandem, one device being connected to and operated by the output of the non-linear amplifier and the other device being connected to and operated by the out put of the other channel and means for establishing a light beam simultaneously modulated by said two devices in optical tandem.

'7. An electro-optical system forthe reproduction of a multi-colored original comprising means for establishing in an electric channel a signal corresponding to primary blue light from each point of the original in scanning succession, means for establishing in a second electric channel a signal similarly corresponding to one of the other primary colors from each'point of the original, a non-linear amplifier connected to one of the two channels for amplifying the signal therein substantially exponentially over a wide range of signal energies, means for supporting a sensitive layer, means for scanning the layer with a light beam and two light modulating devices in cooperative optical tandem for simultaneously modulating said light beam, one of the devices being connected to and operated by the output of the non-linear amplifier and the other device being connected to and operated by the output of the other channel.

8. An electro-optical system according to claim 7 in which both channels include non-linear amplifiers.

9. An electro-optical system according to claim '7' in which the non-linear amplifier is in said second channel.

10. An electro-optical system according to claim 7 in which the non-linear amplifier is in said blue channel.

11. An electro-optical system according to claim '7 in which the light modulating device connected to the blue channel is a ribbon type valve and the other light modulating device is a glow lamp in optical tandem with and for illuminating the ribbon type valve.

12. An electro-optical system according to claim 2 in which the light modulating device connected to the non-linear amplifier is a glow lamp and the other light modulating device is a ribbon type valve positioned to modulate the light from said glow lamp.

13. An electro-optical system for the'reproduction of a multi-colored original comprising means for establishing in separate electric channels, electric energies corresponding to the color components of each point of the original in scanning succession, each channel corresponding to one color, at least one of the channels including a non-linear amplifier having means for varying the energy therein in accordance with an exponential function substantially uniformly over a wide range of energies, light valves for controlling light intensity connected to the output of each of the channels, and modifying means connected to and-operated by the output of said non-linear amplifier for modifying the response of the light valve in at least one other of the channels in accordance with the energy in the one of the channelsincluding the non-linear amplifier, said modifying means including both electrical. means for modifying the electric signal in said other of the channels and a light valve in cooperative optical tandem with the light valve of said other of the channels.

14. An electro-optical system according to claim I 2- in which electrical means are connected to and operated by the output of the non-linear amplifier and are also connected to said other one of the channels for modulating the signal in the latter channel in accordance with the signal in the non-linear amplifier. 15. An electro-optical system for the reproduction of a multi-colored original comprising'means for establishing in separate electric channels, electric energies corresponding to the color components of each point of the original in scanning succession, each channel corresponding to one color, a light modulating device connected to the output of each channel, another light modulating device in cooperative optical tandem with the light modulating device connected to one of the channels, means including a non-linear amplifier for operating said another light modulating de- 16. An electro-optical system according to claim 15 in which one of the two devices in optical tandem modulates the widthof said point of light and the other modulates the intensity thereof.

17. An electro-optical system according to claim 15 in which the two devices in optical tandem consist of a glow lamp and a ribbon type valve.

18. An electro-optical system according to claim 15 in which the two devices in optical tandem consist of two ribbon type light valves and the means for forming a spot of light includes lens means for focusing an image of one valve to constitute the spot of light, a lens near the latter valve to focus an image of the other valve at said lens means and means for illuminating said other valve.

19. An electro-optical system according to claim '15 in which a third light modulating device is connected to and operated by a channel different from the channels connected to said devices in cooperative optical tandem, said third device being in cooperative optical tandem with each of the other two devices forming a cooperative optical series of three light devices modulating said spot of light.

VINCENT C. HALL.

Images