las2dem error

[Raúl Casado]

Hello,

I'm using las2dem to create several dtm tif from a point cloud that I previously turn into tiles of 100 m with 20 m of buffer.

When I try to create the dtms with this command

las2dem -i cloud_merged_*_clas.laz -o mdt\ns_ndre.tif -keep_class 2 -step 0.05 -temp_files D:\tmp -ocut 5 -odix _mdt -otif -cores 6

I get this error in some of them:

llx/lly unspecified. set to lower left bounding box 535180/4.12958e+006.
ncols was unspecified. set to 2800.
nrows was unspecified. set to 2800.
ERROR (TINclean): failed malloc for 40610234 TINtriangles.
ERROR: cannot alloc enough TIN triangles to triangulate 20305117 points.
       contact in...@rapidlasso.de for help on what to do.
llx/lly unspecified. set to lower left bounding box 535180/4.12968e+006.
ncols was unspecified. set to 2800.
nrows was unspecified. set to 2800.
ERROR (TINclean): failed malloc for 53759966 TINtriangles.
ERROR: cannot alloc enough TIN triangles to triangulate 26879983 points.
       contact in...@rapidlasso.de for help on what to do.

...

Does anyone know what might be going on or give me a clue as to where I need to go next?

Greetings.

[Jochen Rapidlasso]

Hi Raúl,
this is an “out of memory” issue, which means that the process you are running cannot allocate enough memory to perform a huge operation. This is not necessarily related to your total computer memory.
It has more to do with the memory that a single process can allocate in your system.
You can try using the 64-bit version: lasground64 has the option to handle larger files.
if you have a license for las2dem you can try to use blast2dem, which will process the data "on-the-fly" and can handle much larger files.

In any case: you can easily avoid this problem by splitting your input files into (smaller) tiles.
Tiling is the most important feature of all tools.
It allows you to process very large files and speeds up processing thanks to the ability to perform multi-core operations.

We recommend this article about tiles and buffers:
https://rapidlasso.de/use-buffers-when-processing-lidar-in-tiles/

Your input filename looks also like you use merged data as input. Please avoid this.
Optional you can set your output dimension to a smaller size (ncols, nrows) or set a larger step size.

Best regards,

Jochen @rapidlasso

[Raúl Casado]

Hello Jochen,

I'll try both methods (tiling and blas2dem)

Thank you.