Python 3DTiles Tilers

p3dtilers is a Python tool and library allowing to build 3D Tiles tilesets out of various geometrical formats e.g. OBJ, GeoJSON, IFC or CityGML through 3dCityDB databases

p3dtilers uses py3dtiles python library for its in memory representation of tilesets

CLI Features

• ObjTiler: converts OBJ files to a 3D Tiles tileset
• GeojsonTiler: converts GeoJson files to a 3D Tiles tileset
• IfcTiler: converts IFC files to a 3D Tiles tileset
• CityTiler: converts CityGML features (e.g buildings, water bodies, terrain...) extracted from a 3dCityDB database to a 3D Tiles tileset
• TilesetReader: read, merge or transform 3DTiles tilesets

Installation from sources

NB: py3dtilers installation requires a Bash shell. If you're using Windows 10, see how use the Ubuntu Bash shell on Windows.

In order to install py3dtilers from sources use:

apt install git python3 python3-pip virtualenv
git clone https://github.com/VCityTeam/py3dtilers
cd py3dtilers

Install binary sub-dependencies with your platform package installer e.g. for Ubuntu use

apt-get install -y liblas-c3 libopenblas-base # py3dtiles binary dependencies
apt-get install -y libpq-dev                  # required usage of psycopg2 within py3dtilers

(Warning: when using Ubuntu 20.04, replace liblas-c3 by liblaszip-dev)

Proceed with the installation of py3dtilers per se

$ virtualenv -p python3 venv
$ . venv/bin/activate
(venv)$ pip install -e .

(Note: On Windows (without bash shell), replace $ . venv/bin/activate with $ . venv\Scripts\activate)

Caveat emptor: make sure, that the IfcOpenShell dependency was properly installed with help of the python -c 'import ifcopenshell' command. In case of failure of the importation try re-installing but this time with the verbose flag, that is try

(venv)$ pip install -e . -v

and look for the lines concerning IfcOpenShell.

Usage

To use the a tiler, check the corresponding readme to know the usage and the features of the tiler.

• CityTiler readme
• GeojsonTiler readme
• ObjTiler readme
• IfcTiler readme
• TilesetReader readme

Useful tutorials:

• CityTiler usage example
• GeojsonTiler usage example
• Visualize 3DTiles in Cesium, iTowns or UD-Viz
• Create 3DTiles from OpenStreetMap data
• Host CityGML data in 3DCityDB

Develop with py3dtilers

Running the tests (optional)

After the installation, if you additionally wish to run unit tests, use

(venv)$ pip install -e .[dev]
(venv)$ pytest

To run CityTiler's tests, you need to install PostgreSQL and Postgis.

To setup PostgreSQL with Postgis on Windows, follow the first step (1. Download PostgreSQL/PostGIS) of 3DCityDB tutorial.
For Ubuntu, follow this tutorial.

Once the setup is over, update the config files (e.g test_config.yml, test_config_2009.yml and test_config_2012.yml) with the right password. Ignore the config files to avoid commiting them in the repo

git update-index --assume-unchanged tests/*/test_config*

Then, in a shell, run:

> psql -c 'create database test_city_tiler;' -U postgres
> psql -U postgres -d test_city_tiler -f tests/city_tiler_test_data/test_data.sql
> psql -c 'create database test_temporal_2009;' -U postgres
> psql -U postgres -d test_temporal_2009 -f tests/city_temporal_tiler_test_data/test_data_temporal_2009.sql
> psql -c 'create database test_temporal_2012;' -U postgres
> psql -U postgres -d test_temporal_2012 -f tests/city_temporal_tiler_test_data/test_data_temporal_2012.sql

Coding style

First, install the additional dev requirements

(venv)$ pip install -e .[dev]

To check if the code follows the coding style, run flake8

(venv)$ flake8 .

You can fix most of the coding style errors with autopep8

(venv)$ autopep8 --in-place --recursive py3dtilers/

If you want to apply autopep8 from root directory, exclude the venv directory

(venv)$ autopep8 --in-place --exclude='venv*' --recursive .

Developing py3dtilers together with py3dtiles

By default, the py3dtilers' setup.py build stage uses github's version of py3dtiles (as opposed to using Oslandia's version on Pypi. When developing one might need/wish to use a local version of py3dtiles (located on host in another directory e.g. by cloning the original repository) it is possible

1. to first install py3dtiles by following the installation notes
2. then within the py3dtilers (cloned) directory, comment out (or delete) the line reference to py3dtiles.

This boils down to :

$ git clone https://github.com/VCityTeam/py3dtiles
$ cd py3dtiles
$ ...
$ source venv/bin/activate
(venv)$ cd ..
(venv)$ git clone https://github.com/VCityTeam/py3dtilers
(venv)$ cd py3dtilers
(venv)$ # Edit setup.py and comment out py3dtiles reference
(venv)$ pip install -e .
(venv)$ pytest

Concerning CityTiler

• For developers, some design notes
• Credentials: CityTiler original code is due to Jeremy Gaillard (when working at LIRIS, University of Lyon, France)

Configuring your IDE

When configuring your IDE to run a specific tiler, you must indicate the module you want to run (e.g. py3dtilers.CityTiler.CityTiler) and not the path to the file (i.e. not ${workspace_root}/py3dtilers/CityTiler/CityTiler.py), otherwise python will not be able to resolve the relative import of the Tilers to the Common package of py3dtilers. An example of launch configuration in VSCode:

{
    "version": "0.2.0",
    "configurations": [
        {
            "name": "<launch_config_name>", // e.g. "CityTiler" or "bozo"
            "type": "python",
            "request": "launch",
            "module": "<tiler_module>", // e.g. py3dtilers.CityTiler.CityTiler
            "args": ["--db_config_path", "${workspaceRoot}/py3dtilers/CityTiler/<my_config_file.yml>"],
            "console": "integratedTerminal"
        }
    ]
}

Profiling

Python standard module cProfile allows to profile Python code.

In code

Import modules:

import cProfile
import pstats

Profile the code between enable() and disable():

cp = cProfile.Profile()
cp.enable()  # Start profiling
   
# code here

cp.disable()  # Stop profiling
p = pstats.Stats(cp)
p.sort_stats('tottime').print_stats()  # Sort stats by time and print them

In command line

cProfile can be run in the shell with:

python -m cProfile script.py