The Bolex H16 RX cameras and the more recent Bolex models utilize a beam-splitting prism for reflex viewing and still allow lens interchangeability. This leads to optical complications. The prism being between the taking lens and the film must be regarded as an optical element of that lens. Even though it is a very simple element with no curvatures, it has optical significance. Often this significance is misunderstood and underestimated, as when Bolex users recommend a simple shimming to match a C-mount lens to a Bolex H16 RX camera. Often too the significance is misunderstood and overestimated, as when the manufacturer recommends that for focal lengths 50mm or less H16 RX cameras require RX-mount lenses, while for greater focal lengths C-mount lenses are all right. This note will clarify the matter of the optical significance of the Bolex prism.

Reflex Systems

There are two main methods of achieving reflex viewing in cine cameras: the displaceable mirror and the fixed semi-transparent mirror. The displaceable mirror moves out of the image path for each exposure, so that the reflex viewing flickers. This may or may not be a disadvantage. The semi-transparent mirror remains in place, always deflecting a fraction of the image forming light to the viewer, providing flickerless viewing. But the viewing is dimmer than by the first method, and most important, there is some loss of light to the exposing film. The displaceable mirror is the preferable reflex method, and the fixed semi-transparent mirror is adopted as an economy.

There are two main types of fixed semi-transparent mirrors used, the pellicle and the beamsplitter prism. The pellicle is a thin membrane coated on one side to be semi-transparent/semi-reflective. By making the pellicle exceedingly thin and by applying a sophisticated anti-reflective coating to the other side, ghosting can be held to tolerable levels. The beamsplitter prism is typically a pair of 45-degree prisms sandwiching a semi-reflective coating. Together they make a glass block, which has the advantage of rigidity over the pellicle. Since there is only one surface at 45° there is no problem of ghosting. The main difficulty with the beamsplitter prism method is that the glass block introduces optical aberrations.

Imaging through a Pane

A pane of glass, even if it were of perfect composition and perfectly flat and parallel faced, introduces optical aberrations. Light exiting a lens, which would come to a point of focus in air, will not come to a point of focus when a glass pane is interposed. The Bolex prism is like a pane 9.5mm thick (and refractive index 1.518). Sometimes the dysfocusing effect is great enough to warrant designing special lenses for imaging through the prism. These RX-mount lenses are like C-mount lenses except that in the air they do not produce sharp images. RX-mount lenses exit rays which come to a particular dysfocus which just cancels the dysfocusing effect of the prism, and this results in their excellent sharp images on the film in the Bolex H16 RX camera. Sometimes the prism effect is minimal. It is necessary to consider the optics in some detail to understand why.

The major aberration due to the beamsplitter prism is negative spherical aberration and only this aberration will be considered in this note and only for images on axis. As shown in the first diagram, rays which exit the lens converging toward a focal point R have their convergence slowed by the prism. When they exit the prism they resume their original convergence angles, but now fail to come to a point.

The prism has had two effects. First is the effect of moving the focal point back from R to the vicinity of S. In the Bolex this displacement is 3.24mm. This is a harmless and not at all subtle effect. The Bolex H16 RX camera compensates for the displacement by increasing the lensseat-to-filmplane distance by 3.24mm. From lensseat to filmplane is 20.76mm in the Bolex H16 RX, while it is 17.52mm in a C-mount camera. The second effect is dysfocus. The prism slows the convergence of all the rays, but unequally. The more central rays are slowed most so that they converge the closest. The 3.24mm displacement is measured from to the convergence point of the rays very close to the axis. The more peripheral rays converge to points more than 3.24mm behind R.

The second diagram shows a RX-mount lens with the positive spherical aberration neutralized by the prism. Notice that the RX-mount lens differs from the C-mount lens in its aberrations, not in its backfocus. Both lenses focus the most central rays to the same point (in the air as well as through the prism).

Since the dysfocus due to a pane is the result of disagreement in the focus of central and peripheral rays, if only one sort of ray is involved there is no dysfocus. When a camera lens is stopped-down so that only the central rays may exit, the dysfocus effect of the prism disappears. The third diagram, which is drawn roughly to scale, shows how effective this is. It shows the vicinity S of diagram 1. The rays are labeled according to the f-stop from which they would be peripheral. The rays from tiny apertures focus at P and those at the f/2 periphery focus at Q. Q is approximately .11mm behind P Notice that the focal length of the lens does not figure in the geometry of the diagram at all.

If a film is positioned at P it receives a sharp point of light from the central rays but a big blur, as wide as AB, from the totality of rays. If a film is positioned at Q it receives a sharp point of light from the peripheral rays but a big blur, as wide as CD, from the totality. According to this diagram, which represents a sampling of the rays from an f/2 C-mount lens dysfocused due to the Bolex prism, where should the filmplane be? By one analysis (which is overly simple because it only considers expanse of blur without considering intensity distribution) it should be positioned where the blur is smallest. This is at the "waist" of the surface which envelopes all the rays. The waist is shown at O.

The position as well as the size of the waist O depends on the lens opening. Considering the same diagram, if some of the peripheral rays are omitted then there is a new envelope with a new waist. The further the lens is stopped-down the closer o comes to P and the size of the blur approaches zero. At f/2 the waist O is about .084mm from P and about 1/80mm across. At f/3.2 the waist is only about .034mm from P and only about 1/300mm across. An additional blur of 1/300mm may be regarded as insignificant, even for a very good photographic lens. This suggests a rule-of-thumb for choosing lenses for a Bolex H16 RX camera:

RX RULE: STANDARD C-MOUNT LENSES WILL WORK WELL THROUGH THE BOLEX PRISM PROVIDED THAT THEY ARE STOPPED-DOWN PAST ABOUT F/3.2

This rule contradicts the rule given in the Bolex literature that for focal lengths less than 50mm and only for these, H16 RX cameras require RX-mount lenses. I fact a very fast 100mm C-mount lens will give poor results wide-open on the Bolex H16 RX, whereas a 10mm C-mount lens will give fine results stopped-down on the Bolex H16 RX. There is no truth in the old prescription.

The RX RULE is reversible. RX-mount lenses will work well on C-mount cameras provided that they are stopped-down past f/3.2. There are also implications for macrophotography. For example, at 1:1 magnification the effective f-stop of a lens is twice the indicated f-stop. So at 1:1 an f/1.6 C-mount lens will work about as well on an H16 RX camera as on a C-mount camera, and similarly for an f/1.6 RX-mount lens.

The construction of the Bolex H16 RX places the film at P, rather than at O, where the blur is less. But this is of no consequence for lenses which focus by moving to and from the film. The RX RULE presumes that the plane of best focus will be located in the course of visual focusing. When a C-mount lens is used on an H16 RX camera its distance markings will simply be found too high. There is no reason to shim a C-mount fixed focal length by moving to and from the film, and in order for a zoom lens to maintain focus while changing focal length the distance between its rear elements and the film must be precisely set. If an f/2 C-mount zoom lens hold focus in air at the standard C-mount flange distance then it will hold quasi-focus through the Bolex prism at point O. To effectively place the film at O, rather than at P, the lens may be shimmed away from the camera by the distance PO, about .084mm. This common procedure greatly improves the zoom lens performance at f/2, but it overlooks the prism effect.

The position as well as the size of the waist O depends on the lens opening. With spherical aberration there is, besides dysfocus (measured by the size of O), also a shifting of the quasi-focal point (the location of O) with change of f-stop. This does not defeat the RX RULE, because the dysfocus is already slight at f/3.2 and even if the C-mount lens is focused at f/3.2 and then stopped down for the taking, the dysfocus can become no larger. It does not spoil the procedure of shimming C-mount zoom lenses for use on the H16 RX camera. With the .084mm shim the blur at f/2 is about 1/80mm (the width of the waist O), which is hardly negligible but perhaps tolerable. Without the shim the blur at f/2 is about 1/20mm (the width of AB) which is disastrous. But the .084mm shimming calculated for f/2 worsens the zoom lens' performance at smaller apertures. As can be seen from the third diagram the more central rays make relatively large blur circles at the waist which is determined for more peripheral rays. When the shimmed C-mount zoom lens is stopped down to f/5.6 the prism will still be introducing blur of about 1/90mm, whereas without the shimming the blur at f/5.6 would be an insignificant 1/400mm. Some painful compromising is in order. A shim thickness less than .084mm must be chosen and the blur at f/2 must be worse than 1/80mm in order for the blur at f/5.6 to be better than 1/90mm. It is common practice for zoom lens setups to be made at full aperture. This example shows the danger in this practice for Bolex H16 RX cameras.

The Bolex Reflex System

It has been assumed so far that the image on the film, focused or dysfocused, agrees wit the image on the groundglass, so that the fast RX-mount lenses can be accurately focused and so that the stopped-down C-mount lenses can be wisely compromised. This assumption is unfortunately false. The Bolex prism stands not only between the lens and the film, but also between the lens and the groundglass. In fact, the groundglass is the upper surface of the prism, ground. This simplification neatly eliminates two glass-to-air surfaces and achieves rigidity. However, as shown in the fourth diagram, it also introduces approximately double the thickness glass between lens and groundglass as there is between lens and film. This means that the RX-mount lens, which is well corrected for imaging through 9.5mm of glass onto film, must dysfocus through 19mm of glass onto the groundglass. In the second 9.5mm it acquires as much spherical aberration as a C-mount lens acquires through 9.5mm of glass onto film. There are some mitigating factors, notably the beamsplitting mirror size which tends to favor central rays for reflection, but the result is an unsharp, hard to evaluate groundglass image with fast RX-mount lenses, and shifts of focus with changing f-stop.

The focal shifting makes it strictly impossible to set the position of the Bolex groundglass. For example, it can be set right for focusing at f/1.1, but then it is wrong for focusing at f/2.2, and vice versa. The errors are significant and they undermine careful focusing with fast RX-mount lenses on the Bolex H16 RX camera. The called-for compromises resemble those discussed above in connection with shimming an inappropriate zoom lens to an H16 RX camera.

This note has examined the optical significance of the Bolex prism. The analysis is limited to the major aberration, but test confirm the general conclusions. In itself, the beamsplitter reflex method is workable, because the resultant aberrations can be neutralized by the use of special lenses or ignored when lenses are stopped-down. However, in elongating the prism to bear the groundglass, and in making it too thin for a full-sized 45-degree mirror, Bolex violated the first principle of reflex design: Make the path to the groundglass optically just like the path to the film.

--Dennis Couzin 3/76

The author is indebted to a thirteen page circular dated October 1958 from Paillard S.A., "Pourquoi et comment le prisme de la camera H16 RX modifie-t-il la correction des aberration de certains objectifs?". However, readers familiar with the circular letter will recognize that this note is correcting the fundamentally faulty analysis in that letter.