Back to the Panoramic KAP Rig
Posted by Tom Benedict on 14/11/2012
I got my pano rig out again over the weekend, flying it at the south end of Hapuna Beach. I flew my Canon T2i and 18-55mm lens set at 18mm. It was… less than satisfying. But I think it’s pointing me in a direction I was already heading, so in a way it was a good experience.
Two problems jumped out. First, using the 18-55mm lens at the wide end makes for fuzzy pictures with horrid chromatic aberration. In the past I’ve flown it set to 24mm. It performs far better at this focal length. I just wish I could use its full wide angle capabilities. Pity. But right now I can’t afford a higher quality wide angle lens – prime or zoom.
The second had to do with the pano rig itself. With the exception of one orbit of the rig, in every other orbit it had gaps in the sequence when the rig was pointing toward the beach. It’s as if the servo sped up, slowed down, and then sped up again at some other point of the rotation. I couldn’t stitch the images because there wasn’t enough overlap between images. In the worst cases there wasn’t ANY overlap.
This is simply unacceptable. What I wanted when I built this was a KAP rig I could put in the air and without fail come back with good solid sets of images I could stitch every single time. I’m not getting that. So obviously something has to change.
My original plan when I built this was to stick a relative shaft encoder on the pan axis. That way I could drive the rig to a real position, trigger the shutter, then rotate to the next real position. I punted on this when I saw how consistent my pan servo was. I probably shouldn’t have. It was good, but obviously it wasn’t good enough.
My plan now is to add a plate on top of the pan axis gear. The plate will have a series of small rare-earth magnets embedded in it. I’ll then hot glue a hall-effect sensor to the gear guide so it’s positioned directly over the ring of magnets in the plate. The Micro Maestro I’m using as a rig controller can be configured to use one of its I/O pins as an analog to digital converter. I’ll wire the hall-effect sensor to the ADC input and use that to sense the position of the magnet plate. That, in turn, will give me an encoder on the pan axis that I can use to position the rig for each exposure. The plate and hall-effect sensor should also stay inside the existing envelope of the rig. No extra things protruding out, and only one extra wire. It should have a minimal impact on the rig itself.
I hope this works. I’ve got a lot of work still to do on rig stability. But the rest of the rig has to work first. Then it’s time for post-sunset panoramic KAP!