[Proposal - PointCloud] Add Eye-Dome Lighting (EDL) feature in pointclouds

This issue is a feature proposal. Feel free to upvote (with :+1: ), comment and provide your use-cases if you're interested by this feature.

Context

Eye-Dome Lighting (EDL) is a non photo-realistic, lighting method designed to improve depth perception by grouping objects that are close together and shading their outlines. This is implemented as efficient post-processing passes on the GPU which relies solely on depth information.

This algorithm was first proposed by Boucheny et al. in 2009 (in french, chapter 4, pages 115-127) and implemented in Cloud Compare.

A variation of this algorithm for point clouds was proposed by Schultz (section 4.4, pages 38-41) which uses logarithmic depths and different defaults and implemented in Potree.

Expected behavior

A picture is worth a thousand words!

Autzen stadium point cloud rendered by potree

Without EDL:

with EDL:

Proposal

Add EDL post-processing in itowns either:

as a postprocessing pass after rendering all layers. Easier to implement but EDL will be applied to all rendered layers.
as a postprocessing pass only on point-cloud layers. Harder to implement since we'll need deferred rendering for those layers (and sort rendered objects).

Identified use-cases

IGN LiDAR HD viewer

iTowns / itowns