Can’t seem to get something to synth right? Curious about what lens is best? Ask your fellow Photosynthers here.
Ok, so I have flown about quite a bit with my UAV and am pretty happy with the results but its not the best.
I am thinking that flying lines in a star form, there and back, then onto the next will make images that include a centre point that is the same and that will give photosynth a chance to make sense of what it sees.
We are also considering mounting two cameras, so that each image overlaps, one camera would see the horizon and the other down a bit.
The idea there is to again give photosynth a chance.
I will plan a star type flight now for tommorrow morning and see how it goes.
I can't say that I completely understand what you're describing with the star pattern, so it's difficult to critique or praise.
I like the idea of two cameras pointed at different vertical orientations with overlap between the two. I think that'll set you up for great success.
I honestly don't know how good it will be for Photosynth, per se, but you might be interested to know what the planes that capture imagery for Bing Maps' Bird's Eye View photos do, if you aren't already aware.
There's a great video on Channel 9 that details their flight pattern, etc. here: http://channel9.msdn.com/posts/Mark+Brown/Behind-The-Maps-Flying-the-UltraCam/
Again, whether its appropriate for Photosynth or not, I don't know, as I'm sure that the Bird's Eye View imagery is placed in Virtual Earth|Bing Maps based on professionally recorded GPS and orientation data, rather than relying on photogrammetry, but their overlapping corridors of photos might work for Photosynth as well.
Hmm. The forum busted my link. Try this, if you don't feel like piecing the above link back together.
Ah, I just found your post on ning. ( here for those who are following along: http://photosynth.ning.com/forum/topics/kapaerial-synths-whats-the )
The four path star might work, but I'd be more comfortable with a six path star. Once you fly both ways the star you have pictured would result in 8 passes and what I'm envisioning would result in at least 12 passes.
My only reason for suggesting this is that Photosynth (according to the Photography guide) has a hard time making jumps of more than 30 degrees of difference. Your 4 path star has 45 degrees of difference between two adjoining spokes, whereas a six path star would have 30 degrees of difference between the spokes.
Again, that is only theory. If you test a four path star and it all binds together, then all the better for efficiency.
Ah I just spotted what you said about 30 degrees. I did'nt know that. The star method did'nt really work, maybe thats why!
hi Gary and Nathanael, good to see a discussion of the ever-interesting aerial synths.
I'll offer my take on the issue by starting with a question: What's the goal when creating these aerial synths? My sense is that an aerial synth wants to show a specific landscape from above and at the same time allow the viewer to feel the freedom of flight.
A camera attached to a kite can capture the path of motion that the kite follows. By flying in overlapping paths or patterns of paths (stars or grids) with a single camera (or even a couple of cameras) one creates a series of paths that a viewer may follow. With each path, the possibilites of motion by the synth viewer becomes more dense. A single path will feel pretty limited, whereas many paths, especially over the same area, will offer many possiblities for the viewer.
By adding more cameras, you will increase both the density of the synth and the freedom of movement therein. While in many instances having more than one camera is unnecessary, aerial synths are an exception. Because the kite is moving all around all of the time, it is impossible to create "Nodes" (positions in a synth that are easy for a viewer to rotate their direction) without having multiple cameras on the kite itself.
In short, I think multiple cameras would be great for aerial synths because the ability to turn from a single point, a fundamental part of "Area Exploring" synths, with one camera is difficult or impossible. Ideally, a series of cameras or an ultra-wide angle camera (or ten) would be likely to provide a synth viewing experience that could best meet the goal of aerial viewing and flight-like movement.
Actually, I seem to have misquoted the guide. I reread it and it actually says 25 degrees or less. Even adding a 7th path to the star still leaves a difference of 25.71428571428571 degrees between spokes, so an 8 (or more) path star seems to be where it's at.
The grid that the Bing Maps guys use is good because the angle they shoot overlaps quite a few of the adjoining rows|columns. Any object can be seen from one direction to move from the horizon towards the camera and back out the other side to the horizon on the other side of the plane when the photos are traversed.
As to not being able to create a halo with a kite, it seems to me that if you could somehow control the direction that the camera was facing (not so likely, I'm guessing given the windiness) that if you could fly your kite in a circle around an area and at least collect enough photos to cull the ones that all look inside the circle, you'd have just what you want.