[QGIS-Developer] R: New QEP: Point clouds in QGIS (Martin Dobias)

[Martin Dobias]

Hi Niccolò

On Tue, Aug 11, 2020 at 10:24 AM Niccolò Marchi
<sciurusurbanus at hotmail.it> wrote:
>
> Reading the description of the QEP I didn’t see great considerations about LAS & LASlib.
>
> From my experience, LAS format is almost a standard for ALS LiDAR data sharing and, even if LAZ is still not well known, the size can be shrinked to 1/10 thanks to it. It has a 2D index that can be easily created through the open source tools of LAStools and it’s definitely well exploited using the streaming filters provided by LASlib (at least in IO operations). For what concerns the visualization, here you can find how it has been solved for the R language (horrible for memory handling) making use of OpenGL.

I have considered LASlib library as well, but for our use case, PDAL
is a superior choice for multiple reasons:
- PDAL supports various different formats and protocols (like e.g.
GDAL does) while LASlib only deals with LAS/LAZ and maybe three more
file-based formats
- PDAL comes with plenty of algorithms for processing of point cloud
data while LASlib does not
- PDAL is being developed as an open community effort while LASlib is
mostly a single-person project

Regarding format support, both LAS & LAZ (among other formats) are of
course going to be supported as PDAL already supports them.

The 2D index from LAStools is useful for processing of data, but not
so much for visualization: if you are rendering point cloud data, you
often need to show the whole extent of the dataset but only using a
small fraction of the dataset. This is where that index from LASlib
does not help as far as I can tell: you want to have a different index
structure where you have data stored in internal nodes so you can
fetch those points quickly without having to traverse large parts of
the dataset.

Thanks for the pointer to PointCloudViewer app - I have not known it before.

Regards
Martin