In all cases the camera must be of slightly greater size than the size of the images required, so that they may be trimmed to avoid any errors of orientation. When extreme accuracy is not necessary (otherwise the camera must be mounted as a theodolite), the parallelism of the two positions of the optical axis is obtained with sufficient approximation by observing the image of some object at infinity and bringing it to the same point on the ground glass, or on a reference mark on the finder (replaced, if necessary, by a telescope with cross wires, or by a rifle sight).
844. Considering the first case mentioned above, if we are to place the eyes opposite the principal points of the pair—the condition necessary to avoid distortion of the reconstructed object (§ 822)—the separation of the two stations must be chosen so that it bears the same relation to the mean distance of the object as the separa tion of the eyes bears to the principal distance (L. P. Clerc, 1917). In other words, if B is the length of the base (distance between the two stations), D the mean distance of the object from the base, b the separation of the eyes, and d the principal distance of the images of the pair (practically equal to the focal lengths of the lenses in the case of a distant object), the length B b B must be chosen to satisfy the relation = — D d It is for this reason that stereoscopic pairs taken vertically from the air are given a separa tion equal to a quarter of the altitude if the lens used has a focal length of about io in., i.e. four times the separation of the eyes.
845. The rule given above is obviously useless in the case of a group of objects of considerable depth we can no longer talk about the distance of the object, as it varies considerably from one part to another.
If a series of photographs are taken of open country from a high terrace, the camera being moved perpendicular to the direction of vision between each exposure, thus obtaining stereo scopic pairs of increasing separation from, say, 9 in. to Jo the sensation of relief obtained in viewing the various stereograins obtained becomes more and more marked as the base is increased. But, when the stations are more than a certain distance apart, it becomes more and more difficult to make the foregrounds of the images coincide. If the base is increased still farther, fusion of the images becomes impossible even in the distance, the eyes being attracted by the discordance of the foreground, however hard the observer may try to neglect it.
In determining the separation of the stations, both the distance of the foreground and the more remote parts of the subject must be taken into account. It is easily seen that for a certain distance of remote parts the base can be made larger the greater the distance of the foreground, with consequent greater relief in the remote parts.
The relation the foreground bears to the distant parts greatly influences the length of base it is possible to use. For instance, in a photograph taken from a cliff over a valley, the various object planes range in increasing dis tances up to the horizon or to the limit of the remote part without any object in the fore ground masking an object in the distance. Under these conditions, if the result is examined stereoscopically, the eyes can view successively the various planes, and are not influenced by quite a considerable variation of separation of corresponding points. If, on the contrary, the photograph is taken, from the ground level, of a bush or tree at about 20 yd. from the camera, standing out against a hillside which forms a. background, the maximum variation of separa tion of corresponding points appears between adjacent points.
Experience shows that in the case of a stereo scopic pair taken looking down on the subject, the eyes can stand a separation of homologous points about five times as great as in the case of horizontal views.
Most observers can endure a variation of separa tion of homologous points equal to iiotli of the distance at which they are examined in the case of bird's-eye views, and of isoth of this distance in the case of horizontal views.
If a stereogram is viewed under normal condi tions, the eyes occupying the respective view points of the two perspectives, the viewing distance is equal to the focal length F of the lens used in taking the photographs, and if the foreground and distance are respectively at distances D and D' from the base, the maximum separation B of the stations is given by the following equations' Bird's-eyeviews B 10 (D' —D) DD' Horizontal views B = 50 (D —D) I The variation of separation e e' of the homologous points (I 819) is actually equal to (e - e') DD' e - e' F - or D' (1)' - D) If then (e is given the value Filo or 175o it is easily seen (L. P. Clerc, 1917) that