LAUSR

Realistic and interactive visualization of individual trees is a desirable functionality in numerous applications for landscape planning, ecosystem simulations, and forest management. However, achieving a persuasive visualization of extensive forests while maintaining an interactive experience remains a challenge. This paper introduces a new framework for a convincing and interactive visualization of large-scale forests originating from forest growth simulation. First, the GPU-based self-organizing tree synthesis algorithm is adapted to produce detailed tree models on-the-fly with the desired level of detail and at interactive rates. Next, the algorithm is enhanced to generate tree models corresponding to the forest simulation results and local growth conditions. Finally, a forest succession model, based on single trees, is linked to the tree synthesis algorithm, to produce detailed tree models that concur with the results of forest simulation. The results demonstrate that the generated trees follow predicted height from forest simulation, are adapted to their neighbors properly, and retain the typical form of the corresponding species. A decimation technique, integrated directly into tree geometry construction, lowers memory requirements for interactive visualization of forests containing thousands of trees. Finally, a combination of the GPU-based tree synthesis and load balancing enables interactive tree synthesis in-between individual frames.