SUBJECTIVE SHARPNESS OF IMAGES 



5V9 



the return trace, by the observed area of one figure of confusion of the 

 equally sharp motion picture. The number of cycles per frame period 

 was the nominal band width of the video circuit, in cycles per second, 

 divided by the number of frames per second, or 24. 



Case A 



Case B 



Case C 



Figures of confusion per frame 



Cycles per frame period 



Ratio 



22,400 



33,600 



1.50 



18,900 



26,200 



1.38 . 



14,800 



18,800 



1.27 



The ratio in Case A was suspected to be too large because of un- 

 accounted-for small defects in the film scanner which degraded the 

 image sharpness more than was indicated by the aperture efifect alone. 

 The difference in ratio between Cases B and C was no larger than the 

 measured probable error of each set of observations. Making allow- 

 ance for these things, we concluded that the ratio between the number 

 of cycles per frame period and the number of figures of confusion per 

 frame area had been found to be 1.3. 



This factor 1.3 gave us a basis for calculating the television aperture 

 loss in the direction normal to the scanning lines, and enabled us to 

 compute the nominal video frequency band required to yield an image 

 having equal horizontal and vertical resolutions. 



The stepped nature of the brightness variation across the scanning 

 lines of a television image, in contrast to its continuous nature along 

 the lines, gives rise to the requirement that for equal resolution in the 

 two directions the scanning apertures must be longer in the scanning 

 direction than they are across it. The extent of this departure from 

 squareness has been estimated (see references 3 to 7) at from 1.2 to 1.9 

 mostly on theoretical grounds. Our comparison of a television image 

 of known characteristics with a controlled out-of-focus motion picture 

 furnished a subjective measurement of the effect which yielded the 

 value 1.4 for the ratio of width to height of the scanning apertures for 

 equal resolution. We take width to mean the dimension along the 

 scanning lines, and height to mean the dimension normal to them. 



We found that the nominal video band width of a television signal, 

 in cycles per frame period, was 1.3 times the number of figures of 

 confusion per frame area in the equally sharp motion picture. This 

 meant that the area of each figure of confusion was 1.3 times as great 

 as the area of one scanning line over a (scanned) length of one cycle. 

 By the adopted definition of nominal video band width, the length of 

 one cycle was just twice the length of each one of the pair of rectangular 

 scanning apertures which were considered equivalent to the actual 



