University of Konstanz
Graduiertenkolleg / PhD Program
Computer and Information Science

Sören Pirk

Research Student in the PhD program since 01.09.2008.

advisors

  1. Prof. Dr. Oliver Deussen
  2. Jun.-Prof. Dr. Dorit Merhof

organisational data

Room:Z 706
Tel.:4091
E-mail:soeren.pirk ( at ) uni-konstanz.de
Other Resources:Alternate Member Page
picture

project description



Rendering of natural scenes with vegetation as rich as in the real world has been a motivation of computer graphics research ever since. As frequent objects in our daily life, vegetation is almost part of all virtual sceneries, ranging from single plants and trees to huge outdoor landscapes. This thesis proposal outlines open research questions in the problem domain of modeling and rendering botanics and presents recent advances in these areas.

The complex visual appearance and the inhomogeneous structure of botanical objects makes rendering of large scenes a challenging task that extends to this day. The obvious main reason is the tremendous amount of geometry that is needed to represent trees and plants. Even though complex virtual environments are routinely used in applications like games, movies or urban visualization, storing as well as rendering such objects with full detail is beyond the capabilities even of modern (graphics) hardware.

The visual appearance heavily depends on the underlying geometry and complex material properties (i.e. translucent leaves). Targeting photorealism requires complex models and algorithms for approximating the transport of light. Shading and shadowing, both by themselves complex areas of interest within computer graphics, need to be adopted and developed for achieving a high level of reality when rendering trees and plants. Especially shadowing is an expensive task in real-time graphics, requiring multiple render passes of the tree’s geometry. Often, vegetation is only a minor, however, important detail when defining virtual scenarios. Content creators like artists, designers or architects are in the need to define these details without spending a majority of their time. Algorithms and models for editing and modeling trees are not yet satisfying these requirements. Complex procedural formalisms (i.e. L-Systems) or tedious modeling by artists are the current approaches for more natural reality in applications.

Animation of vegetation and the interaction of trees and plants with their environment are rather unaddressed. While high level physical interaction models found their way into today's applications (e.g. simple wind animations), more sophisticated approaches are necessary to describe the complex behavior observable in nature. The visualization of biological motivated processes like growth or effects of tropisms are barely researched. The simulation and visualization of biological processes - either at microscopic or macroscopic level - of plant behavior and development is yet only in its minority part of computer graphics research. Especially the real-time domain defines a hard set of constraints expressed by the need to process several frames a second. Modeling and rendering trees in this domain therefore requires efficient algorithms and well defined paradigms within all afore mentioned areas. Realism therefore depends on particular characteristics of vegetal scenes represented by the following areas:

  • Geometric Complexity - The complex nature of trees and plants at the human scale of perception translates to massive amounts of geometry that needs to be handled on computer (graphics) hardware. Algorithms and paradigms need to be researched allowing to represent botanical objects in a less complex way.
  • Shading and Shadowing - Shading, shadowing and complex lighting highly increase the visual appearance of involved objects. Current approaches require multiple render passes to achieve visual quality. More efficient algorithms and render techniques are necessary.
  • Editing and Modeling - The geometric complexity disables the interactive modification of predefined models. More sophisticated approaches are necessary to allow editing and modeling in real-time.
  • Animation and Interaction - Visualizing physical interaction and biological processes are barely addressed in this context. New approaches are required to enable rendering of these effects.


We currently strive to contribute to all of the mentioned areas by proposing solutions to the addressed problems.

publications

The following list of publications covers only those, which are or were published during participation at the Graduiertenkolleg / PhD program.

Articles in Journals

2014201320122011
2014
2013
2012
2011

Conference Papers

20122010
2012
2010

Phd Theses

2013
  • Pirk, S., Efficient Processing of Plant Life in computer Graphics, University of Konstanz, October 2013. Link abstract