Agree on orbiter orientation

[vorg]

Currently orbiter is using latlon-to-xyz and xyz-to-latlon to convert spherical coordinates to xyz position. How those 2 relate to each other is matter of convention. The convention assumed in the current implementation assumes that lat:0, lon:0 is point [0, 0, 1] therefore orbiter is located at +X axis looking towards -X axis. This is to make it compatible with lat/lon coordinates of a textured sphere from https://github.com/glo-js/primitive-sphere often used for globe rendering. The problem is that this is agains default OpenGL convention where default camera is positioned +Z axis looking towards -Z axis.

[vorg]

What can we do:

1. nothing, just add docs and use orbiter.set({ position: [x, y, z] }) for orienting the camera.
   • Problem: expected minLon, maxLon is from e.g. -PI/2, PI/2 but in this case the correct values are -PI, 0
2. change pex-renderer Orbiter only but adding +-PI/2 to all xyz2latlon transformations
   • Problem: now orbiter lat/lon doesn't match geo position (gps) lat/lon
3. Add new theta/phi parameters that behave in "WebGL way" (0 deg is +Z)
   • Problem: conflicting set of constrains would be possible if we add minTheta/maxTheta on top of minLon/maxLon

[vorg]

Current pex-renderer behaviour. Longitute 0 is on the right +X to make it match longitue 0 on a equirectangular earth texture (given primitive-sphere uv coords)

[vorg]

Three JS

[vorg]

The conclusion is not to touch lon and lat and introduce theta = lon - 90deg and phi = lat

[dmnsgn]

v4: move this to pex-cam