* onboard shader versions
Pushing more computations per-vertex is a better fit for onboard GPUs, with very limited pixel shading power.
The artifacts that result from per-vertex nonlinear computations is something we'll have to work out or put up with.
Onboard variants also avoid multipass algorithms like the plague, using only the absolute minimum passes. Usually this means
multipass lighting is off the table, and planet shaders can only handle one light.
TODO: planet shaders could handle more lights *diffusely* though, but shadows and other phenomena would have issues.
* more consistent looks across shader settings
Increasing/Decreasing shader settings now maintains a consistent look, adjusting only precision in several aspects:
- disabling less influential phenomena (like pass-thru rayleigh scattering, parallax, etc), replacing them
with similarly-looking rough and quick approximations (before it wasn't even approximated, just removed)
- pushing nonlinear but screen-coherent low-frequency computations to the vertex shader, retaining the phenomena
but trading performance for artifacts
- reducing sampling frequency when multisampling
- reducing or removing use of anisotropic filtering
* hidden penumbra shadow artifacts