Manuel Noronha Gamito
INESC, Rua Alves Redol, 9, 2º, 1000 LISBOA E-Mail: mag@inesc.pt
In this article we present a method for the animation and visualization of turbulent fluid flow. The method is simple, fast and stable. It is based on well know methods from the field of Computational Fluid Dynamics, treating the fluid as a vorticity field. Vorticity is transported by a particle system. A uniform grid is also used to calculate a velocity field that moves particles to their new positions. Such a mixed method where free particles move over a fixed grid has very small computational costs, making it suitable for Computer Animation.
The method simulates the behavior of fluids in situations where the contact between fluid masses with different velocities gives rise to turbulence phenomena. It is suitable for the animation of gaseous fluids like smoke. Unlike previous algorithms, it is possible to generate turbulence over all scales, ranging from the macroscopic to the microscopic level. The algorithm is controlled by a small number of intuitive parameters, enabling animators to quickly take maximum advantage of it. The algorithm can also be parallelized easily owing to its particle nature.
Mário Rui Gomes
INESC, Rua Alves Redol, 9, 1000 Lisboa E-mail: mrg@inesc.ptIn this paper we present a work on human face simulation. The purpose of this work was to develop a system capable of simulating realistic facial expressions using personal computers. This work is part of a project which consists on implementing a coding algorithm for videotelephony communications. This project aims at transmitting digital model data instead of images to allow data transmission at a much lower bit rate. Since videotelephony communications are mostly used for transmitting human face images, we need to construct a human face model to simulate the behaviour of the real face.
The system described here uses a single frontal image of the human face that we intend to simulate. This image is then used for texture mapping on a three-dimensional model capable of adaptation to different people's facial images. For this adaptation we need to recognize the position and attributes of the most relevant facial features such as the mouth, eyes, eyebrows, nose, chin, etc. In this article we assume that this recognition is manually specified by the user of the system. However, we have already developed a system capable of automatic facial feature recognition. Based on the position and attributes of facial features, we can adapt the 3D model and through texture mapping, the system is capable of simulating the same face with many different expressions and positions by changing a set of parameters (about 50). Our system can also simulate movement by synthesizing transitions between different expressions specified as keyframes.