Py3DTilers: an open-source toolkit to create 3DTiles

⚠️ Arrested development ⚠️

Alas, starting on January 2025, the current Py3dTilers development team will move on to new projects. Py3dTilers will thus no longer be developed nor maintained. The current repository owners are thus looking for a community (a set of individuals, teams or corporations) that would be willing to adopt py3dTilers in order to pursue the development activity. This transition could take the form of adding new developers and transmitting the MAINTAINER role (at the repository level). But, as allowed by the LICENSE, the transmission could go as far as relinquishing the repository ownership to another responsible organization.

At this stage, some contacts were made with some public or private organizations (currently using and/or contributing to py3dTilers). If the future of py3dTilers interests you, and you would willing to engage some resources to take the lead on maintaining/developing py3dTilers, please provide in this issue some sketchy declaration of intention and possibly a coarse roadmap.

Deciding the actual form of transmission will be based on the result of the public discussion happening in this issue.

---

Py3DTilers 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.

Py3DTilers uses Py3DTiles python library for its in memory representation of tilesets.

Py3DTilers can only produce Batched 3D Models (B3DM). If you want to produce Point Clouds (PNTS), see Py3DTiles CLI.

An article that presents the tool is available in the ISPRS annals here.

Demo

Find 3D Tiles created with Py3DTilers in this online demo.

CLI Features

• Common features: features shared by all tilers
• 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

See supported Python versions

For Unix

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

apt-get install -y libpq-dev       # required usage of psycopg2 within py3dtilers
apt-get install python3 python3-dev

Install Py3DTilers in a safe python virtual environment (not mandatory yet quite recommended)

apt-get install virtualenv git
git clone https://github.com/VCityTeam/py3dtilers
cd py3dtilers
virtualenv -p python3 venv
. venv/bin/activate
(venv)$ pip install -e .

For Windows

In order to install Py3DTilers from sources use:

git clone https://github.com/VCityTeam/py3dtilers
cd py3dtilers
python3 -m venv venv
. venv/Scripts/activate
(venv)$ pip install -e .

Usage

In order to access to the different flavors of tilers, refer to the corresponding readmes to discover their respective usage and features:

• Common features readme
• 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

Before commiting, please run tests and make sure coding style is respected.

Running the tests

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

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

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