LVIZ author Philip Paar shakes the 3D tree

Philip Paar has been recognized as a geo design, Grassroots GIS, and landscape visualization enthusiast. In 2010, he started an ongoing affair with the digital content creation industry. Autodesk 3ds Max® users of this Blog are invited to check out the Laubwerk Plants Kits Freebie for free trial at

Laubwerk Plants Kits in 3ds Max

Interactive Modeling of Self-Adapting Botanical Trees

Already a few weeks ago, a colleague drew my attention to a couple of new papers on interactive self-adapting botanical tree models, published by the computer visualization group in Konstanz, who had also presented at Siggraph 2011 (see my previous post about Siggraph). This new development in procedural modeling techniques makes tree models possible that will interact with their environment! For example, if you insert a building, the surrounding trees will change their shape and branches will give way to the new object in a most realistic way.

How are these astonishing results achieved? Input is a skeleton-based tree geometry. In contrast to traditional tree growing models, the new technique approximates biologically motivated transformations. Main factor is the light distribution and the amount of resources a tree receives. On that basis, the growth rate for the entire tree and individual branches as well as branch ages are calculated. A complex illumination model makes sure that light conditions are updated for different stages of tree growth. Additional factors are phototropism and gravitropism and I was surprised how realistic the outcomes look.

Because the approach does not require the tree model to be reconstructed from the beginning, it performs much better than previous approaches and is even suitable for real-time applications. That means, you could insert an object such as a building but also other trees and experiment in real-time how the surrounding trees might change their growth in response to each other and inanimate objects.

For more information and the original research paper, please see


SIGGRAPH Asia 2012:

Plastic Trees: Interactive Modeling of Self-Adapting Botanical Trees from Soeren Pirk on Vimeo.









Plant Modeling at Siggraph 2011

Presenting my own poster “Model-based Visualization of Future Forest Landscapes” on the use of Biosphere3D in the Kimberley project at CALP, I got in touch with other researchers in the field. LIAMA, the Sino-French Lab of Computer Science, Automation and Applied Mathematics in Bejing, presented a poster on GreenLab. GreenLab is a program for the stochastic, functional and interactive modeling of plant growth that also considers different growth conditions. According to Prof. Kang, the development of GreenLab goes back to AMAP and Philippe de Reffye, one of the AMAP developers, who is also guest researcher at LIAMA and contributed to the development of GreenLab. The libraries of GreenLab will also contain more Asian species which are still rare in 3D. With its potential for functional modeling, Greenlab may become another promising plant modeler.

Later in the day, I visited the Speedtree booth. Speedtree clearly aims at Game Developers and Movie Makers; therefore it does not need to integrate botanical rules but it has to provide artistic control and “directability”. Furthermore, performance and different LODs are important for game developers and in gaming, SpeeTree is the current state-of-the-art. For landscape architects, it may be over the top with the basic version of SpeedTree Studio selling for $850 while the professional version, including a world construction set, can cost more than $12k.

Siggraph 2011 – Day 3

One of todays` highlights at SIGGRAPH 2011 in Vancouver was the approach of Prof. Deussen`s group in Konstanz and their partners at the universities in Shenzhen and Tel Aviv to use so-called “texture-lobes” for tree modelling from Lidar data. For more information, see

Texture-Lobes for Tree Modelling from Soeren Pirk on Vimeo.

The afternoon was dedicated to urban modeling, starting with a review of the latest literature by Peter Wonka (Arizona State University) and Daniel Aliaga (Purdue University). During the second part of the session, Pascal Mueller from Procedural/ESRI (CityEngine) presented issues encountered in practice and their new Urban Vision project together with the urban planning department of San Francisco and Urban Sim (Paul Waddell). The session closed with a visually very engaging case study by Michael Frederickson from Pixar, using CityEngine for virtual London in Cars 2.

Mangroves of Mexico as Google Earth Outreach example for GE6 in cooperation with CONABIO

Three weeks ago, Google had launched its new version 6 of Google Earth parallel to the Cancun conference, where Google Earth Outreach participated as well. Now, Google Earth Outreach together with the Mexican National Commission for Knowledge and Use of Biodiversity CONABIO launched the first outreach project taking advantage of the newly implemented tree representations (download the kmz into GE here). The trees still look a bit sketchy in comparison to other products but the visualization seems to be based on credible and sophisticated vegetation data.

Source: Google Earth Outreach

GoogleEarth 6 plants 3D trees

Google published version 6 of GoogleEarth and the most important update for our profession is that it comes with 3D trees!

However, not the whole Earth is populated with trees yet but only some areas in a few selected cities: According to TechCrunch 80 million trees were “planted” in Athens, Berlin, Chicago, New York City, San Francisco and Tokyo. Yet, the quality of trees is rather low but Google claims to distinguish 50 species already. Other updates of version 6 include an enhanced integration of StreetView and a more user-friendly overview and navigation through historic aerial images. The new version is a beta and does not update automatically but must be installed from the Google download site. A GoogleEarth showcase can be downloaded here.

The key question for us is how the integration may facilitate the use of GoogleEarth as a tool in landscape related design and planning professions. Sheppard and Cizek (2009) discussed the various technical and ethical issues of previous GoogleEarth versions and identified the lack of vegetation as one of the major limitations. Our tests in the Kimberley Climate Adaptation Project supported the argument that the lack of vegetation distorts the representation of future design scenarios in GoogleEarth and inhibits the otherwise potentially useful tool for landscape planning.

In this context, the new GoogleEarth version allows new user tests and applications. Next steps could be to explore how 3D trees are inserted and distributed, assessments how realistic GoogleEarth forest stands are,  and how user´s landscape perception will change with regard to Google´s “Tree View”. As one of the first new applications, is planning to use the virtual trees to push (real) reforestation campaigns.

Open Source Toolkit for 3D Plant Modelling

Yesterday I came across an articel in Graphical Models (71), pp. 1-21. It is about PlantGL, an open-source graphic toolkit for the creation, simulation and analysis of 3D virtual plants. The software is based on three components: a geometric, an algorithmic and a GUI library with an interface to Phyton, which enables a modeller to develop scripts and procedures in Python. Furthermore, importers and exporters for data exchange with several modelling and visualization systems, such as AMAPmod/VPlants and Pov-Ray, exist. The approach combines several methods to represent plant structures at different scales, ranging from tissues to plant communities. One of the features are parametric envelopes for the representation of crown shapes, which sounds to be a very promising technique to control the shape of 3D plant models. Accurate representation of the crown shape will be important for representing plants in real world models (e.g. 3D city models) for visibility analysis.

According to the paper the librarys seem to be very powerful and several examples are presented reaching from 2D and 3D tissue models to branching systems and plant communities. Moreover, a quick overview over current plant modelling approaches is given including L-Studio/Vlab and GroGra based on L-systems, the AMAP system, and Xfrog and extensive references to related work are given. Thus I like to encourage everybody interested in plant and vegetation modelling to read the paper and look at the website of the Virtual Plant project team. The site shows several examples, links to research papers and software, and explains the methods apllied in the research.