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

more intractable. Figure 12 shows that points at 13p will be separated by 6.26 in. on the screen instead of the proper 2.5 in., which would induce serious eyestrain, especially for spectators in the front rows of seats. If the magnification is raised from 218 to 300 (i.e. a screen width of 20 ft 8 in.), the separation on background points would amount to 8.6 in., more than three times the separation of the human eyes. It should be remembered that divergence does not occur at all in natural vision; its physiological effect can be extremely uncomfortable. Distortions With Fixed-fc Systems But this is not the last of the disadvantages of a "human vision" camera arrangement. The use of large B factors entailed by employing a value of tc which is often too big for the size of screen and the depth range to be compassed, leads to serious distortion of the shape of objects, as may be seen by a comparison of the example worked out under "Image Distortion in the Theater," earlier in this paper, with the same scene shot according to the precepts of "human vision." The data will be exactly the same, save that tc = 2.5 in. instead of 1.25 in., and that the dancer (who will be in the plane of sharpest focus) will be moved very slightly back from N\ .09 to N\. When these new values have been substituted, along with the unchanged data, in Eqs. (11) to (13), the results in Table VI are obtained, which for convenience are placed alongside the characteristics of the shot in the film. Table VI. The Black Swan, Slate 15, Stereo Distortions at the Plane of the Dancer's Final Position. M = 218; V = 2.5PF. By "human vision" In the film md 12.4 niu) 3.4 4.9 3.1 M 3.6 1.6 What has happened with "human vision" is that the width magnification has been kept down by making tc equal to t, which is one of the conditions for ensuring that mw — 1 (see Eq. (15)). On the other hand, the depth magnification has increased enormously owing to the squared term in Eq. (11), and this has more than doubled the distortion of shape, as indicated by the figures for the shape ratio. This consequence of a "human vision" approach can be even more clearly demonstrated by the graphical technique already described. Figure 13 is in essence an enlargement of the relevant part of Fig. 12. It shows Beryl Grey's shoulder placed at 48p (10 ft 5 in.) from the camera, with arm outstretched as she pirouettes, so that the fingers are at 60p (8 ft 4 in.). The x-axis has been graduated in N values, so that the value of P (distance from spectator to a point in the image) can be readily calculated from Eq. (1). By taking two points, one at the shoulder and one at the fingertips, the actual stereoscopic length of the arm can be found by simple subtraction, in spite of the fact that the magnification varies nonlinearly between the two points. Figure 13 shows that the arm length, as shot for The Black Swan, is 102 in., whereas by "human vision" principles it would have been 208 in., or more than twice as great. The overall depth magnification works out as 4.1 in the first case, and 8.3 in the second, since the real arm length is 25 in.* As the dancer continues her pirouette, her outstretched arm moves into a plane parallel with that of the camera lenses, where its magnification will be uniform, and is given by mw in Table VI. The two shape ratios are therefore 1.2 for the film and 2.4 for "human vision," which is again twice as much distorted. * The overall magnification is of course lower than the magnification at the shoulder, because md decreases as N increases, and is therefore least at the fingertips. 280 October 1952 Journal of the SMPTE Vol. 59