tin2dem

Tool for fast rendering of TIN (Triangular Irregular Networks) surface in LandXML format into DEM (Digital Elevation Model) raster files in GeoTIFF format using OpenCL on CPU or GPU

Running in Docker

Two Docker images are available: CPU (with POCL) and GPU (vendor-agnostic, requires host GPU runtime (not tested for ADM)).

Using Pre-built Images

You can use pre-built images from Docker Hub without building them locally.

CPU Image

docker run --rm -v "$(pwd)":/data lekkks/tin2dem:latest-cpu /data/input.xml /data/output.tif

GPU Image

To use GPU acceleration, you need to configure Docker to access GPU devices. For setup instructions, see:

• NVIDIA GPUs
• AMD GPUs

# For NVIDIA GPUs
docker run --rm --gpus all -v /etc/OpenCL/vendors:/host-ocl-vendors:ro -e OCL_ICD_VENDORS=/host-ocl-vendors -v "$(pwd)":/data lekkks/tin2dem:latest-gpu /data/input.xml /data/output.tif

For AMD GPUs, you'll need to manually run with appropriate device access (NOT TESTED):

docker run --rm --device=/dev/kfd --device=/dev/dri --group-add video -v /opt/rocm:/opt/rocm:ro -v /etc/OpenCL/vendors:/host-ocl-vendors:ro -e OCL_ICD_VENDORS=/host-ocl-vendors -v "$(pwd)":/data lekkks/tin2dem:latest-gpu /data/input.xml /data/output.tif

Diagnostics

Both images include clinfo for OpenCL diagnostics. To check available platforms:

docker run --rm -it --entrypoint clinfo lekkks/tin2dem:latest-cpu
docker run --rm -it --gpus all \
  -v /etc/OpenCL/vendors:/host-ocl-vendors:ro \
  -e OCL_ICD_VENDORS=/host-ocl-vendors \
  --entrypoint clinfo lekkks/tin2dem:latest-gpu

Installation

Requirements

• python3
• GDAL library with python bindings
• OpenCL runtime

Linux

Install dependencies

sudo apt-get update
sudo apt-get install python3-pip gdal-bin gdal-data python3-gdal ocl-icd-libopencl1 clinfo 

Install OpenCL driver

Here is the list of OpenCL implementations: https://www.iwocl.org/resources/opencl-implementations/
Install runtime corresponding to your GPU

You can use universal runtime POCL and run render with CPU:

sudo apt install pocl-opencl-icd 

Check if you have runtime installed

clinfo -l

Windows

Install Python 3, GDAL (e.g., via conda: conda install -c conda-forge gdal), and OpenCL runtime (GPU drivers or POCL).

Install package using pip

pip install git+ssh://git@github.com/lekks/tin2dem

or checkout from git and run test

pip install pytest
pytest tests

Usage

usage: tin2dem.py [-h] [--pixel PIXEL] [--epsg EPSG] [--chunk CHUNK]
                  [--margins MARGINS] [--surface SURFACE] [-a]
                  input_tin output_tiff

positional arguments:
  input_tin
  output_tiff

optional arguments:
  -h, --help         show this help message and exit
  --pixel PIXEL      Pixel size
  --epsg EPSG        EPSG code
  --chunk CHUNK      Processing chunk size, optimal value may depend of your
                     GPU memory. Default is 256
  --margins MARGINS  Output DEM margins
  --surface SURFACE  Surface to render if multiple surfaces is found
  -a, --autocad      Autocad compatible output (shift on 1/2 pixels)

set PYOPENCL_CTX environment variable if you don't want to choose runtime every time, for example:

PYOPENCL_CTX=0 tin2dem tin.xml dem.tif

Examples

Render example files from landxml.org

wget http://landxml.org/schema/LandXML-2.0/samples/Carlson%20Software/Olympus_Subdivision-2.0.xml
tin2dem Olympus_Subdivision-2.0.xml Olympus_Subdivision-2.0.tif

Multiple surfaces:

wget http://landxml.org/schema/LandXML-1.1/samples/BLUERIDGE%20Analytics/siteops.xml
tin2dem siteops.xml siteops.tif --surface=4