I think there's a relationship between focal length, baseline between pinholes, and distance to the subject. Not sure how to express this mathematically, but you can get good pairs closer up with wide angles, but wider baselines are hard to integrate back into stereo, especially when you're close up. With longer pinhole to film distances, you can take advantage of longer baselines, but you can't be too close to your subject. Not very clear.
The images embedded in this blog are set up for the crossed eye method. If you want to play along and you've got a lorgnette, Brian May's Owl, Google Cardboard or an Oculus rift, here's a link to a PDF with them set up for stereo viewers, in the same order they're in the blog.
If you want to learn cross eyed stereo viewing, check this PDF for instruction.
I had lots of 4x5 paper left over from my August workshop, and the paper developer I bought wasn't going to last forever.
I'm not going to get really organized and quantitative. I just tried a bunch of different things to see if it made a viewable stereo pair, not whether one was best.
About the most outrageous arrangement I could think of was 14oz vegetable cans, which of course would have curved film planes and extemely wide angles. Oddly enough, I tried this before circa Y2K, and I vaguely remember that it worked.
My neighbors ripped out the hedge and part of a fence while we were out of town last weekend, and left it looking really ugly (They plan to put lawn there).
This pair is a little tricky to view, but it works. (Work on getting the pipes at the top of the fences to align, By the way, this one doesn't work at all with a stereo viewer - see below)
With crossed-eye, the left image goes on the right and vice versa. Looking at this pair, it seems like there's lot of wall in the right image, and a lot of trees and drive way in the left image. If you look at the angle of the fence coming out from the house, you can see the left image is on the right and vice versa. I think the cans rotated a little bit without me noticing when I put rubber bands around them to hold to the tripod, so they're actually pointing in a little bit different directions. But where they do overlap, you get a stereo image. By the way, this doesn't work at all with a stereo viewer because you can't vary the placement of the objects in the image the way you can by crossing your eyes.
Tried it again with a little less angular subject matter, and was really careful that they were correctly aligned.
That goes together very easily. I think it helps that I'm a little farther away as well. This is the first picture I've done that I could probably pass off as a Victorian stereogram. It's also kind of interesting that without any straight lines in the middle, you could hardly tell this was a curved film plane.
Next up were the one pound Oaks Candy boxes, with the pinhole 2 inches from the paper, pretty darn wide angle. I started with the vertical format with the boxes right next to each other so the baseline was 4 inches. Started fairly close-up about four feet from the baker's rack.
I went on to try them horizontal. The boxes are wider than the paper so the baseline is now 8 inches.
I can still get it integrated. I thought the longer baseline would give me a more prominent stereo effect, but I really don't see much difference and it does make you cross you're eyes harder to get it into 3D.
So now to the 5 inch workshop camera. First with the cameras side by side.
This is an example of an extremely messy compostion with very few cues to depth, but in 3D the pine branch at the top is noticeably closer with the herb garden, the plants along the path, the arbor and hedge at the back of the yard separated in their respective planes.
I'm doing this with a 15 inch piece of board attached to the tripod, and next I put the cameras at the opposite ends, so a baseline of 11 inches.
I think this one doesn't really work. I can get from the herb garden and back into 3D, but can't get the pine branch at the top to come together. When you increase the baseline, you definitely can't get as close as you can when the cameras are closer together.
I've always thought that stereo views of really grand vistas often suffer because distant objects don't have as much difference as those close-up. Well, telephoto is the photographic solution to getting a closer look at something, and I thought would allow a wider baseline, and end up with better stereo separation.
I got out my 10 inch foamcore camera. I only had one, so I had to use the cha-cha method, successive exposures with the camera moved between exposures. The house across the street isn't going anywhere, and on an overcast day, exposure and lighting would remain consisent.
First, with a 4 inch baseline as though I had two cameras side by side.
Still works in stereo. Now out to the 11 inch baseline.
I think I can detect a little bit of an enhancement to the stereo effect with the longer baseline, especially for more distant objects. I think I can see a little better separation between the house and the trees in the background. One thing you might not have noticed is that giant rut in the lawn the neighbors made by driving a truck up to the hedge, and I think this closer object is more prominently 3D in the pair with the shorter baseline.
I'm afraid I don't have any succinct summary conclusions after all this. You can use a lot of variations of focal length, and you can use longer baselines with longer cameras, but it makes a difference how far away what you're trying to photograph is.
If anyone knows of mathematical rules that determines all this, or just a better explanation, I'd love to see the link.