Journal of the Society of Motion Picture Engineers (1930-1949)

points. Direct translation of this value into the number of lines from top to bottom of the picture gives 1,323 X 0.75 = 992 lines. The bandwidth required to transmit this detail, given by the familiar l?RP/2 formula, is therefore (9922 X 4 X 24)/(3 X 2) = 15.75 me. It will be argued that it would not give a balanced picture, i.e. one in which vertical and horizontal definitions are equal, because of the line scanning factor K. Various values have been assigned to K, but taking it at 0.75, the number of lines is increased to 992/0.75 = 1,320. To sum up, therefore, definition along the line corresponding to one-thousandth of the picture height is required, but to take account of diversity in the discontinuous vertical direction, the number of scanning lines may have to be increased to 1,300 with a 25% increase in bandwidth to cater for the increased scanning speed. However, because of the probably greater incidence of vertical than horizontal lines in a natural scene, it may not prove necessary to go much above 1,000 lines, and, since there is a tremendous advantage in keeping the writing speed as low as possible, this figure has been taken as a basis for first experiments. It must be emphasized that the whole of the foregoing is advanced with extreme reserve and is, moreover, the subject of experiments currently being made, as much of it is based on pure supposition and on theories which have always been the subject of fierce controversy. Doubtless, calculations on other bases would yield widely divergent results, but the authors feel that it is essential to make some attempt to determine numerical values, as a starting point for practical investigation. Quite apart from the foregoing, there remains the possibility of introducing novel means of picture dissection — which may prove more adaptable than scanning of the orthodox variety — to the production of motion-picture film by television methods. It is too early, however, to make more than a passing reference to such possibilities, and for the purpose of the paper the authors have confined their consideration to scanning of the conventional type. (4) Interlaced and Sequential Scanning For the purpose in hand the choice between interlaced and sequential scanning involves several important considerations. Interlaced scanning is universally used for broadcast television and, in this connection, is an extremely useful expedient. By interlacing, the apparent flicker frequency of the reproduced picture is doubled, without, however, any increase in the bandwidth required to transmit it. The principle of interlacing therefore possesses outstanding advantages for broadcasting in that its use transforms a television picture of comparatively low repetition frequency, which would exhibit considerable flicker if sequentially scanned, into one which within acceptable limits of brightness is effectively flicker free. On the other hand, the introduction of interlacing is generally held to reduce the apparent definition of the picture as viewed by the eye. A number of effects are involved, of which three may be cited. First, slight inaccuracies of registration of the interlace raster result in "pairing" of the scanning lines, or, in extreme cases, superimposition of the lace and interlace lines. This is bound to reduce the definition progressively as the pairing effect becomes worse, until complete superimposition occurs, when the definition is theoretically halved. It is only fair to record that advances in design of scanning circuits have greatly reduced this defect in the last year or so. Secondly, the movement of the viewer's eye when following vertically moving objects strobes the line structure and momentarily breaks the picture as seen into half the number of lines, giving the impression of a coarse linestructure. A similar effect occurs in Collins and Macnamara: Electronic Camera 453