Requirements for Tyler on the sphere
Opened this issue · 1 comments
- Intersect the viewing frustum with the sphere and retrieve the area of interest
- potentially center the camera in Google Earth style rotation so that you cant translate, then just use FOV and view_direction to compute the bbox from the spherical cap
The second point is done in my GlobeAxis PR to GeoMakie. The first remains to be solved. One interesting resource might be https://stackoverflow.com/questions/13541526/hierarchical-frustum-culling-of-sphere-tiles.
Maybe some plane-sphere intersections can help with this? But we would need to know how dense to make each area of the globe. It's tricky for sure.
A good enough compromise for now is simply getting the viewed lat/long back, and using the "radius" from (0,0,0) to determine how much detail we need. That would work under the assumptions that:
a. The camera always points towards (0, 0, 0)
b. The FOV is circular, i.e., a cone that forms a spherical cap when it intersects with the globe (it's not, but that's an OK approximation - not too wasteful for now)