Dual Camera rig

[Robin Parmar]

I am looking for an inexpensive option for rigging up two small cameras (Panasonic GH1) so that the axis of the lenses are as congruent as possible. I am thinking that this means one camera on top of the other. (I'll flip the video from one unit in post.)

Currently I have a breadboard on order, plus an articulated arm. I am not sure that will do the trick to get a stable enough configuration.

I will be using adapted lenses with a larger image circle than the Micro Four Thirds sensor. Then I can use a tilt adapter to compensate for the lenses being physically separated and hence off-axis. Am I thinking this through correctly?

[Alastair Traill]

What is the aim of this exercise? What is the subject matter?

[Robin Parmar]

It's an experimental dance film. I want to shoot the same movement sequence at two different depths of field OR with two different fields of view, but from the same perspective. Mounting cameras side by side (as if for 3D) would create parallax I am striving to avoid.

So, certainly not a standard application!

[Alastair Traill]

Are you simply trying to extend your depth of field or do you have something else in mind?

[Robin Parmar]

[I've asked the same question on another forum, so I'll just post what I wrote there, by way of further explanation.]

The idea is to explore the interaction between the optics and the dance (which will be choreographed with this in mind, once we have a chance to rehearse and see what works and what doesn't).

Imagine that two identical cameras with two identical lenses are used from the same perspective. Set a wide aperture so that DOF is thin. Setting each for a different focal plane will then reveal different aspects of the movement depending on where the dancer's movements take her in terms of distance to the camera. You can't do this in post starting with a single reel, unless you impose some blur "effect".

This is just one possible scenario -- another might involve different lenses with different focal lengths, to enable superimposition or sequencing of footage with different FOV.

Why not shoot multiple takes? A well-rehearsed dancer replicating movements will appear to exactly repeat motion when viewed from a distance typical of an audience in a hall. But in fact there are many small differences in each run-through. A tightly-focused camera -- especially using a lens at f/1.2, for example -- would easily distinguish between these.

Besides, this being an art project we want to do something different. The idea is to make he optics an implicit part of the dance, so the resulting video is not mere "documentation" of something that could have occurred without filming.

[Alastair Traill]

My guess is that the effect would be rather confusing to the viewer but that is another story.

No matter how you position two separate lenses you are obviously going to have parallax issues. Your problem seems to be how to produce two images from the one optical axis.

If your problem was my problem I would be investigating using a partially reflecting mirror placed at 45 degree to the optical axis of the lens of one of your cameras with the second camera positioned so that its optical axis was at 45 deg from the same mirror but at 90 deg from that of the first camera. This would mean that the first camera receives light that passes through the mirror whereas the second camera receives light that is reflected from the mirror. Lining everything up would be testing to say the least and require the ability to reverse or invert the image of the second camera.

Partially reflecting mirrors are very easy to damage and also produce an image from both surfaces with the brighter image coming from the mirrored surface. The double image problem is reduced by making the mirror very thin in which case it is called a pellicle.

Partially reflecting prisms are another possibility and are much more robust and do not produce a secondary reflection.

[Robin Parmar]

Thanks. Mirrors seem to be the answer if I need perfect superimposition. I doubt I'll have the time to arrange that at this time, since practical shooting starts next week. Though maybe later in the development, since we may have the luxury of a "round two".

For now I'll see if a tilt lens mount can help align the images "enough", whatever that may be. I am on the lookout for a C clamp large enough for two cameras vertically. (Flipping one take in post is no problem.)

The final effect will permit some transitions not normally encountered, which will be interesting to experiment with if nothing else. For instance, where traditionally one might zoom through a range of focal planes, here I'll be able to fade directly from one fixed plane to another with nothing in between. Further, I will be able to make decisions about these transitions in post, with the dancer viewing the movements and an active part of the process.

I trust that if the video is properly composed, viewer confusion will be minimal... or at least deliberate. :-)