Stereo pair of a female lone star tick

[Graham46]

Elf-

Rik's explanations of the set up were correct. My camera and lights all stay in the same position while the subject is not only moved toward the lens to complete the stack, but also rotated on a metric rotary stage to get the second image for stereo vision. Using the metric rotary stages allows a complete 360 degree rotation as long as the subject is centered correctly. To address your second question, instead of the camera moving between frames, the stage is simply rotated 5 degrees. Turning the subject towards the camera or away from the camera usually produces similar results. And to answer your third question, the distance between the lens and subject is always the same when using the 65mm MP-E. Its a very short distance, about 2 inches. As you extend the belows on the lens, the subject needs to move back by the same amount to remain in the in-focus area, but a stereo image can be produced at any magnification. Hope this was helpful!

[ChrisR]

Hmm.
Would it "work" if you moved the subject sideways, ie with the same axis parallel to the lens/camera axis?

And suppose the subject and sensor stayed still, and the lens were moved sideways? My old Nikon bellows can do that.
I'd go off and try it but I've broken the USB socket off inside the pc...

[rjlittlefield]

Would it "work" if you moved the subject sideways, ie with the same axis parallel to the lens/camera axis?

In theory, this works fine. In fact a "perfect" stereo pair --- absolutely no vertical disparity at any point, and with all point pairs in equally good or bad focus --- corresponds exactly to this situation.

Rotating the subject is only an approximation, though it's a pretty good one in most cases. The approximation can be improved by applying a "keystone" transformation to one or both images to make the geometry correspond & remove the vertical disparity. StereoPhoto Maker provides this function.

When DOF is very shallow, rotating the subject (or an equivalent motion of the lens+camera) also introduces differences in focus between the two images. This is less of a problem than it might seem at first, because the stereo matching part of your brain is tolerant of fuzzy images. After all, it's designed to work mostly with image areas outside the fovea.

If you're very hardcore, then you can use Scheimpflug to make the left & right images have exactly the same focus too. This happens naturally when you do stereo through a single-objective microscope, for example by using polarizers to split the condenser beam into two oblique sections, one for each eye.

The problem with using shift to shoot stereo at macro/micro scales is that if you do it the obvious way, the required shift moves the field of view so far that you lose a lot of overlap -- maybe all of it! Suppose you're at 1:1, for example, with a 100 mm lens. Then lens-to-subject is 200 mm. To get that 5 degree change in viewpoint, you need to slide the lens sideways by about 200*sin(5 degrees), or around 17.4 mm. Even with a full-frame camera, only half the frame now overlaps (18.6 out of 36 mm). At 2:1, it's worse: only 5 mm overlap out of 18. And at just over 3:1, the overlap disappears entirely because the shift is larger than the field of view.

--Rik