Re: VR-movie of object using random camera positions?
- Wow Scott, that has been a major project! I'm surely not at all in
position to mimick something even remotely similar. The best I can
hope for is either a kite or an RC plane/helicopter. I don't see an
option to apply the GPS trick you explain in those circumstances, so
I'll probably have to follow your original procedure, with perhaps
some use of PTInterpolate...
Thanks for the comprehensive explanation!
--- In PanoToolsNG@yahoogroups.com, Scott Highton <scott@...> wrote:
> For the Golden Gate bridge object movies, I did not have the GPS
> calculator available yet. But I knew the rough dimensions of the
> bridge structure, including its maximum height and the space
> the towers, and was able to figure out what sized circles we could
> reasonably fly in a fixed wing aircraft around one of the towers.
> From there, I chose the lens that I'd use based on the field of
> needed. There was also a flight ceiling that we couldn't go above
> due to airspace restrictions, as well as a minimal safe turning
> radius for the airplane.
> I planned the flights calculating what the aircraft altitude
> to be for each level, or row, and then figured out from maps,
> and aerial photos what good landmarks on the ground could be used
> help define the circle diameters for the pilot. Given wind
> I knew it was impossible to fly perfectly round circles, so instead
> told the pilot to do the best he could, but to make sure that he
> finished each circle as close to the starting point (via ground
> reference object) as possible. For example, in the lowest orbit,
> think we used the exact middle of the bridge span as our start and
> end point reference.
> The first time we tried this, I took an assistant along. We
> the back seat of the plane, as well as the back door, and I sat on
> the floor (in a safety harness) shooting out the door. Of course,
> the pilot's job was to fly the plane. I had my assistant then
> the gyro compass in the cockpit, and once we determined our ground
> references for each circle, had the assistant call out a mark for
> every 10-degrees of turn we made -- at which time I'd shoot
> shot. Of course we were shooting on film, and since there were
> 37 exposures on each roll, every time we made a mistake, we'd
> away another roll of film.
> Unfortunately, our pilot was not quite as good at flying circles
> he needed to be (there were high winds hampering him), and the
> assistant would skip 10° increments at times while being
> out the window. Add the fact that we were all trying something
> and we had not much success the first time up.
> Several weeks later, I tried again with a different pilot and
> different airplane, but my assistant was unavailable. That meant
> that I had to estimate the 10° increments myself as I was
> based on what things looked like through my viewfinder (the 90°
> reference angles of the structure itself made this easier). This
> actually made the shoot more effective, because there was one less
> link in the communication chain. I still threw away rolls of film
> when I'd not kept the bridge tower centered in my viewfinder
> effectively, or the pilot didn't finish the circle at the same
> he started it.
> We just repeated the same process at each altitude, flying tighter
> circles each time. There was more guesswork and adjustment via
> visual estimation going on as we were shooting than I would have
> liked, but we were able to make it work.
> Today, you can simply enter 36 GPS latitude and longitude
> as a route into an aircraft's autopilot system, and the plane or
> helicopter will fly its own (almost) perfect circles, even in
> crosswind conditions. I do use a gyro stabilizer attached to my
> camera for aerial work, and this helps me keep the images more
> and consistent between shots. These gyros can be rented from a
> number of pro photography rental houses. You can also purchase
> directly from Kenyon Laboratories in Connecticut: http://www.ken-
> Scott Highton
> Author, Virtual Reality Photography
> Web: http://www.vrphotography.com