Thanks, but no thanks!
Also, pitch movements in baseball would become more obvious.
I am a fan of basketball and football, and do know that the innovative camera angles introduced so far fail the pub test; whenever I'm at a bar and the TV switches to one, people either complain or say nothing. I've yet to hear any support.
However in most sports, the ball spins rapidly, anywhere from one rotation every few seconds to 50+ times / second.
Personally, seeing that on camera would probably make me barf.
Also, it would be almost impossible to compensate for this, since they usually rotate around more than one axis.
Could you elaborate?
However, the axis on which the ball spins may change during the flight. The ball can also spin on more than one axis at the same time.
The combination of (1) the difficulty associated with determining which axis the ball will spin on, and (2) preventing the ball from spinning on another axis, along with (3) putting all this hardware in the ball without changing the physical performance of the ball, would make this idea difficult to pull off.
However, your idea might work if we're talking about still images. Then you only have to worry about creating a ball that performs the same as traditional air-inflated balls, but has a camera inside.
You would think that at the north pole the sky above you won't actually change, and that it would only rotate around you. However, if you stand there long enough (several thousand years), you'll notice that the earth rotates more slowly along a second axis. This is why the "true" north star changes over time.
See: http://zebu.uoregon.edu/~imamura/121/lecture-2/precession.html
Apply this to sports balls, and if there's more than one axis, it would probably be hard to stabilize the image with software. One second you're looking at the ground, then the crowd, then the sky, then the foot of the player, etc.
You could put several cameras in the ball, but since most balls are inflated with air - you'd probably significantly change the ball's weight by doing so. My guess is that there are very few situations where a league would be willing to play with a heavier ball.
You could also have one camera with just an extremely high frame rate, and then use image processing to select the few frames every second where the camera's actually pointing in the right direction. However, you'd still want the camera along the major axis, which would be very hard to predict before-hand.
My guess is that if you did this (with only one camera), you might get a few good frames, but not enough to create an actual video. Maybe that's OK - I could see you getting some stellar shots.
I have no idea indeed, but I love how nearly everybody responded to the question by pointing out technical difficulties :)
If you put a camera on both ends, you get a forward/backward view for the most part. Then write something that takes the video and orients the field to the bottom of the screen or something like that.
This reminds me of a post on reddit a loong time ago. Camera software for use with blurry surveillance video that sharpened the image by adjusting it with frames seen immediately before/after. Anyone by chance remember that?
Also what about having flying robocams? How would they work?
(I suspect you'd want the view of wherever the ball was projected to land.)
Perhaps another approach would be to have a bunch of cameras around the stadium, motion-track the ball in 3d, then stitch the ball's POV together using a process sort of like Google street view, or that dumb 3d revolving-subject thing that was in all the commercials five years ago. Then you could calculate a 'line of best fit' for the ball's trajectory and render video 'just-in-time' for a quick, smooth replay.
As others note, though, placing cameras in balls may not be as interesting to fans as other options, like player cams. (Beyond Justin.tv, what about LeBron.tv or Peyton.tv?)