 |
|
|
|
|
|
 |
| ||||||||
 |
 
|
|
|
|
|
|
|
|
|
| ||||||||
|
|
| 
| User list | Rules | You are not logged in.
Don't know if this makes sense or not.
If I shoot a bunch of panos, and I can see from one tripod position to another, then there should be a matching area in the two panos, one larger than the other.
Could SIFT or some other algorithm be used to find hotspots this way?
Would this be worthwhile, or is it just to easy to place hotspots instead of letting SIFT run?
Offline
This is theoretically possible. However computing position of cameras from different point of view of a common scene is just too time/brain consuming and not a mature enough science considering how easy it is to do manually! 
Thanks for the input.
Offline
Don't know that you'd have to compute the position of either camera.
Just imagine two camera positions, A and B where A can see B. Camera A will take an image that is sort of a cone with a rectangular cross section with camera B at its center.
The image from position A will have all the information that the image from B has, but in less detail (and possibly with some obstacles that are in front of A but not B.)
APP could find matching CPs on these two images if there is less than a 4:1 ratio between them.
APT could do some similar sort of matching and setup hotspot links instead of CP links.
It may not need too much processor power, and processors are getting more and more powerful.
I think automatic hotspot generation is possible, but is it useful?
Offline