Simulating The Dynamic Movements And Growing Of Plants

Vegetation and plants are always the important components of natural scene. In such applications as virtual reality/environment, computer animation and flight simulation, the degree of fineness and visual appearance of plants in the outdoor scene attributes to the realism of screen directly and definitively. So far, the representation, modeling and rendering of plants has been one of the prevailing topics. With the rapid development of programmable graphic hardware, researchers have paid more and more attentions on the simulation of dynamic vegetation scene. Simulation of dynamically-varying objects doesn't only require the realism of shape, but also the realism of motion. It is meaning that not only a efficient representation is prerequisite, but also a suitable physical or biological model is necessitated. Based on the physical and biological principles, this thesis aims to simulate the motion and deformation of plants driven by the force or induced by self-growth for the purpose of computer animation. We seek to strike a balance between the realism of motion and appearance and the efficiency of calculating and rendering.The thesis focuses our study on four aspects: first, we combine the advantages of geometric models at controlling deformations and that of image at representing details and explore the multi-resolution representation of plant modeling; Second, we try to construct the physical condition complying with physical laws, seek for the physical model and simplification strategy suitable for computing the deformation of branches, and implement physical-based simulation of trees and forest in the wind in real-time; Third, accurately compute the deformation sequences of a tree in the wind in preprocessing, and by adopting effective motion simplification and geometric compression, implement the efficient algorithm for re-using deformation data. Lastly, explore the fast modeling of flowers, and implement the biologically-based simulation of floral blossom.The main contributions and innovations of thesis are mainly in the follows:We exploit a multi-resolution, hybrid geometry-image representation for plant modeling, which simultaneously meets the requirements of fidelity, efficiency, friendship-to rendering as well as shape controllability. As for level structures, we present an Error Metric Hierarchy (EMH) for geometric levels of detail (GLOD) scheme. Also we develop a parameterized tree modeling toolkit.We implement a physically-realistic wind field using stochastic spectrum analysis. For the efficient dynamically-based simulation of trees in real-time, we propose the simulation levels of detail (SLOD) scheme and achieve a good balance between the realism of motion and efficiency of rendering.We propose a new concept of motion levels of detail (MLOD) scheme to reorganize the deformation data pre-computed in advance according to certain criterion, and the reorganized data could be loaded according to the level of detail. Particularly, we propose a motion simplification method by approximating the original motion based on a tree-like data structure called affine transform hierarchy (ATH).We propose a sketch-based approach to modeling flowers at various status efficiently; based on biological principles, we present a dynamic growth model to describe the continuous deformation of floral components during the process of floral blossom. And by this growth model, we implement the realistic simulation of blossoming originally.