Hundreds of millions points rendering

[Jean-Romain]

The branch rglcontains an early draft and experimental version of a new point-cloud viewer algorithm integrated into lidR based on (and thus compatible with) rgl. With the new improvement users can now display hundreds of million points relatively smoothly (in theory) while it was almost impossible to display more than 10 millions with the former version based on rgl::point3d.

The rough idea is that it only displays 3e6 points by clipping and densifying the point cloud when the user is zooming in. It removes dynamically areas of the point cloud that are invisible and densify the ones that are in the frustum.

The main concept is working fine but important optimizations are needed. It still uses too much memory (but does not blew up the RAM like the regular viewer) and could be made much faster. It is optimized for ALS data. It won't work with TLS

At least it is a proof of concept available as pre-alpha test for who is interested. I tested it with 95 million points.

remotes::install_github("r-lidar/lidR", ref = "rgl")

@bi0m3trics @tgoodbody @ptompalski @Saadi4469 @jfbourdon @Rpouliot