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

not be focused beyond 35 ft 9 in. for M = 218. However, the minimum convergence will alter with the size of screen for which the picture is shot, and the focal length of the lenses in use, which might give rise to awkward mechanical complications. Much more serious than this, however, is the restriction on the depth range imposed by such a method of shooting, and — in the coupled arrangement — the undesirable and highly artificial pushing of things nearer and farther away in space, which will tend to negative the 3-D effect of the film, especially when complicated studio shots are undertaken. This can best be demonstrated by reverting to Slate 15 of The Black Swan, and showing how it would have appeared in space if shot by the "human vision" technique. Figure 12 is a repeat of Fig. 11, save that it is extended to cover a higher range of nearness factors, and is marked with the actual distances found in Slate 15. Curve 1 shows the result of setting the back of the scene (13p) at infinity, i.e. at DO, the method employed in the actual shooting. But now the front set of banners (at 79p) will come out to -/V4 7, i.e. (79 13)/13.97, which is almost J of the distance from the screen to any spectator. This is much closer t'lan the cutting continuity allows, and, in the opinion of many, enters the region of eyestrain. Certainly, the extreme nearness of the banners would be quite out of keeping with the rest of the film. No other fixed setting is possible, since it would produce divergence on the back of the set, even with the extremely modest assumed magnification of 218 (i.e. a 15-ft image in the theater). Thus a variable convergence for this shot is required by the "human vision" system under discussion, and we may conveniently assume that the convergence is coupled to the lens focus system in the way already described. The shot opens with the banners at 79p (6 ft 4 in.), and since at this distance, using a 50-mm lens at an aperture of //2.8, the depth of focus is only about 17 in., it would be necessary to focus with some precision for the distance of the banners themselves. This would cause them to appear in the theater at NI (see Curve 2), and once again the wanted effect would not be achieved, though this time the banners would be too far away, instead of too near. As the first pair was lifted, the plane of focus would move gradually farther away, passing through positions such as Curve 3. Following accepted technique, the camera operator would follow focus in such a way that each pair of banners in turn would occupy the plane of sharpest focus; but this would have a disastrous stereoscopic effect, in that it would bring all the banners into the same image plane (i.e. on the screen), and thus would wipe out the wanted recession in space. When the last banners had been raised, the focus would rest at 27p, and the back of the set would be correctly placed at 13p. But towards the end of the shot, Beryl Grey dances forward to 48p from camera, where she would be out of focus. It is therefore necessary to alter focus again, and according to normal practice, the dancer would be held in the plane of sharpest focus, shown as Curve 4 in Fig. 12. This would not only completely neutralize her forward movement in space, but would create divergence on the background, which would still be sufficiently in focus to be fusible. The only way out of the dilemma of neutralizing depth is to juggle with the depth of focus, placing very near objects at the limit of the zone of acceptable sharpness in order to correct as well as possible their misplacement in space. This procedure would unquestionably be much more inconvenient than having independently adjustable stereo settings for it would contradict accepted camera practice and would give rise to a method of shooting in which the sharpness of focus was always under suspicion. The problem of divergence is even Spottiswoode, Spottiswoode and Smith: 3-D Photography 279