Otter View Park

It didn’t seem windy enough, so I plucked dried heads of orchard grass and tossed them into the air.  The live feed at WeatherUnderground.com had been reporting 10-12 mph NW winds at the Ralph Myhre Golf Course a mile away, but the grass was falling almost straight down here.  On a knoll in the center of the field I felt a breeze, so I took off my pack and assembled the nine foot Levitation Delta kite (with wind less than 15 mph, the Sutton Flow Form 16 would stay in the pack). The delta flew out of my hand and climbed easily to 100 feet.  It was flying to the south, not to the east over the cattail marsh I wanted to photograph.  Otter View Park in Middlebury is the focus of Galen’s community service project, and he has been describing the plant communities in that marsh.  I would not be able to photograph much of it today unless the wind shifted.

I put my foot through the kite line winder and started to pull things out of the pack.  The kite was flying strongly, so I assembled the full automatic rig.  I had just reconfigured this KAP rig (kite aerial photography rig) with a CAMremote controller which I had yet to use.  It was programmed to systematically point and shoot the camera to capture 50 photos for complete coverage of a half-spherical panorama of the scene beneath it.   I was attaching my best KAP camera, a Canon S95, when the line went limp and the kite began to wander.  It was easy to work it back into the air, but the lull convinced me to abandon the heavy autoKAP rig for a minimal frame (no motors, no batteries) and to use a cheaper camera.  The Canon SD1100 would have to point in a single direction, and I chose straight down.  Vertical kite photos could be joined together to make orthophoto mosaics with 10 or 30 times the resolution of Google Earth.   I started the Ultra Intervalometer script to snap a photo every 10 seconds, but the camera was not shooting.  Instead of troubleshooting, I just replaced it with a Canon A590. That was heavier, but the kite seemed to be flying strongly now, and the image quality of the A590 was better.  There were still a few gadgets in my pack that I had not rifled through, but most of them were now spread in a circle around me in the grass and weeds. I attached the picavet suspension to the line and let the kite carry away the clicking rig.

It was 4:18 PM.  WeatherSpark.com says it was 45° with a 10 mph wind.  The sun was bright and casting long shadows across the field and marsh – poor conditions for capturing map images, but excellent for dramatic effect.  The kite had no problem lofting the rig and holding it steady as I walked around the field.  I eventually let out almost 1000 feet of line and could barely see the rig way over the neighborhood along Weybridge Street.  When I reeled it in after 5:00 PM, the field was mostly in shadow, it was 43°, and the camera was still shooting every 10 seconds.  There were 368 fresh photos on the SD card.  The camera had been set on ISO 100 and shutter priority at 1/800 of a second.  That did not allow quite enough light for the second half of the flight, and those images were a little dark.  But they were in focus with minimal motion blur.  This was a really successful flight except for the fact that few of the photos were of the subject of interest.

Adobe Lightroom allowed me to quickly correct for the failing light toward the end of the flight.  I had to eke more out of the jpegs than they had in them (1.5 stops for some), but noisy photos are better than dark ones.  It took about an hour to sift through the images and select for export 50 that covered the scene with plenty of overlap but no duplication.  Jeffrey Warren had just released a new version of MapKnitter, which is an online tool for aligning vertical photos into orthophoto mosaics.  I had used a previous version, but this was the first appropriate set of images I have had for the tool, so I gave it another try.  It did not take long to manually rubbersheet 19 overlapping images so they aligned with the base image (which is like a Google Maps satellite view). The swaying camera was rarely if ever shooting vertically, so distortion has to be corrected in each image.  Since tall objects like houses and trees are shot from different angles in adjacent photos, they never match well and the final mosaic has lots of obvious seams.  The final mosaic can be viewed here. The gigapan above is a detail of the area included at MapKnitter. Unlike the MapKnitter mosaic, this is a stitched image with blended seams between adjacent photos. I thought the raking light on the last leaves of fall was worth preserving carefully. There is more information about that image at gigapan.org.

In retrospect, I think the wind was steady enough to have supported the full autoKAP rig. A successful half-spherical panorama over this location would have been a valued result. The sun was so low that it would have burned out the western part of the scene, but maybe that would have been a spectacular effect. Like the dozens of decisions required to complete a KAP flight (e.g., time, place, kite, line, rig, camera, focal length, focus, ISO, shutter speed, image stabilization, shutter method, shutter interval, height, duration) this one was made deliberately and then mostly forgotten. It seems that only the obviously bad decisions stay with me, but I guess that allows me to learn from them. Maybe someday I will get that CAMremote controller in the air and learn from that.