Section: New Results
Animating nature
Participants : Alexis Angelidis, Florence Bertails, Marie-Paule Cani, Christine Depraz, Julien Diener, François Faure, Laurent Favreau, Matthieu Nesme, Fabrice Neyret, Laks Raghupathi, Lionel Reveret, Qizhi Yu.
Highly colliding deformable bodies
Participants : Marie-Paule Cani, François Faure, Laks Raghupathi.
We address the question of simulating highly deformable objects in real-time, such as human tissues or cloth. The main problem is to detect and handle multiple (self-)collisions within the bodies. This year, we have focused on the robust response to multiple collisions using two different approaches, one based on Lagrange multipliers and the other based on stiff penalties.
Laks Raghupathi presented his work in a national workshop [30] , and he defended his Ph.D. on this topic on November, 15, 2006.
Robust finite elements for deformable solids
Participants : François Faure, Matthieu Nesme.
We continue a collaboration on surgical simulation with laboratory TIMC through a co-advised Ph.D. thesis. Its purpose is to develop new models of finite elements for the interactive physically-based animation of human tissue.
Two new models of hexahedron-based finite elements have been proposed. The first [27] , [15] is based on octree-based wavelets (see figure 5 ). Given the geometrical model of an object to simulate, we first compute a bounding box and then recursively subdivide it where needed. The cells of this octree structure are labeled with mechanical properties based on material parameters and fill rate. An efficient physical simulation is then performed using hierarchical hexaedral finite elements. The object surface is used for rendering and to apply boundary conditions. Compared with traditional finite element approaches, our method dramatically simplifies the task of volume meshing and increases the propagation of the deformations. This allows us to dynamically concentrate the computational force where it is most needed.
However, the mix of elements at different resolution in the same equation system creates numerical problems. The second model [28] is composed of elements at the same resolution. Starting from voxels at high resolution, we build voxels bottom-up and set the masses and stiffnesses in order to model the physical properties as accurately as possible at any given resolution. Additionally, we extend a fast and robust tetrahedron-FEM approach to the case of hexahedral elements. This permits simulation of arbitrarily complex shapes at interactive rates in a manner that takes into account the distribution of material within the elements. Figure 6 illustrates our results.
|
In the next future, we plan to extend our second approach with multigrid solution methods, allowing us to consider the bodies at different resolutions while avoiding the problems of the wavelets.
Simulation of 1D models and application to hair
Participants : Florence Bertails, Marie-Paule Cani.
Realistically predicting the shape of hair requires an accurate 1D mechanical model, which takes into account the mechanical properties of inextensible, naturally curled hair strands. In the framework of our collaboration with the industrial partner L'Oréal (see Section 7.1 ), we developped a new physically-based method for strand called ``Super-Helices'', which was presented this summer at SIGGRAPH [20] . Super-Helices are a novel deformable model for solving the dynamics of elastic, Kirschoff rods: each strand is represented by a piecewise helical rod which is animated using the principles of Lagrangian mechanics. This results in a realistic and stable simulation, allowing large time steps. We validated this strand model through a series of comparisons between real and simulated hair wisps. We incorporated efficient methods for processing the self-collisions inside hair and with obstacles [7] , enabling us to extend the model to the animation of a full head of hair (see Figure 7 ).
Control of smoke simulation based on vortex filaments
Participants : Alexis Angelidis, Fabrice Neyret.
Based on our last year model of efficient high resolution smoke simulation based on vortex filaments, we developed a new representation adapted to the control of the simulation feature by the user. The idea is to decompose a vortex ring filament into a frame plus harmonic components. The frame is obtained using a PCA analysis (i.e., the average orientation), and is used to represent and control the global trajectory and target of smoke. In this frame, the ring is decomposed into wavelets which represent the animated details of the puff of smoke. This allows us to control the apparence, the simplification with size or distance, and the complexity in a stability/efficiency purpose. This work led to a publication at SCA'06 [18] .
Real-time animation of liquids and river surfaces
Participants : Marie-Paule Cani, Mathieu Coquerelle, Fabrice Neyret, Qizhi Yu.
This year, a PhD student (Qizhi Yu) obtained a European Marie-Curie funding (Visitor programm) to work on this topic. The purpose is to obtain a realistical detailled appareance of landscape-long animated rivers in real-time, with user-editible features. The idea is to separate the river simulation into 3 scales, corresponding to different specification and simulation tools: macroscale for the topographic shape and global flow characteristics (relying on simple CFD at coarse resolution), mesoscopic scale for the local waves patterns (relying on dedicated phenomenological models), microscopic scale for the details (relying on textural procedural schemes). Note that this topic is included in the scope of the NATSIM collaboration (see Section 8.2.2 ).
The PhD of Mathieu Coquerelle, co-advised by Georges-Henri Cottet, explores the use of vortex particules for animating liquids and gases and to simulate their interactions with rigid solids.
Motion capture of animal motion
Participants : Lionel Reveret, Laurent Favreau, Christine Depraz, Marie-Paule Cani.
The motion of animals is still a challenging problem in 3D animation, both for articulated motion and deformation of the skin and fur (see Figure 9 ). The goal of this project is to acquire information from the numerous video footage of wild animals. These animals are impossible to capture into a standard framework of motion capture with markers. There are several challenges in the usage of such video footage for 3D motion capture : only one 2D view is available, important changes occur in lighting, contrast is low between the animal and foreground, etc. Currently, a method has been developed to first extract a binary silhouette of the animals and then, to map this silhouette to pre-existing 3D models of animals and motion thanks to a statistical prediction. This work has been selected as one of the best papers of the Symposium on Computer Animation 2004 (SCA'04) and an extended version has been published to the Graphical Models Journal [13] .
In several domains of character animation, footsteps are one of the most important constraints. It guarantees one of the main aspects of a realistic animation of locomotion. This task, when done manually, is even more complex for quadrupeds. Being able to automatically predict the footsteps information from a video footage is thus an important contribution. The method developed is based on the design of a dedicated image filter to detect the pattern of animal legs. Along the time range of the video, the positive filter responses are clustered so that a single trajectory point is given per leg. As 2D images are considered (profile view), there exist ambiguities in the prediction of each individual foot position when side views of legs are crossing each other (typically left and right side of the animal, and front and back legs for higher velocities). A motion model has been developed to take into account this problem. This work has been done in collaboration with the University of Washington, in Seatle, USA.
The activity on this topic for this year has been dedicated to the ANR project Kameleon. Database of Xray video of a rat has been collected and is currently under processing. In addition, 3D surface reconstruction has been investigated. The approach is based on structured light scanner technique. This work has been done by a PhD student of the University of Montreal, Jamil Drareni. This work has been sponsored by an INRIA Internship grant for a 5 months stay at INRIA Rhône-Alpes. First results are available on the website of the project http://www-evasion.inrialpes.fr/people/Lionel.Reveret/kameleon .
Motion capture of tree motion
Participants : Lionel Reveret, Julien Diener, Fabrice Neyret.
This work investigates how the complex motion of plants and trees under wind effect can be analyzed from video and retarget to complex 3D model to create realistic animation. A first method has been proposed and published at the SCA conference in 2006. The particularity of the method is to be based on statistical clustering only, without the need of any physical model of the tree. The key idea is to retarget hierarchical clustering of moving features as the automatic building of an animated 3D geometrical strcuture of the branches. This work has been done in cooperation with Pr. Eugene Fiume from the University of Toronto, thanks to the "Equipe associée" grant I-MAGE between EVASION and the DGP laboratory of the U. of Toronto. This work will be continued in cooperation with the laboraties from INRA dedicated to the physiology of trees and the fluid mechanics laboratory of "Ecole Polytechnique", within the ANR project Chene-Roseau to be started in 2007.
Evaluation of 3D facial animation
Participant : Lionel Reveret.
Techniques for rendering of the skin surface have now achieved a highly realistic level. However, human subject are highly trained to perceive other faces and require 3D animation of faces to be accurate in terms of amplitude and timing of motion of facial features to be believable. We are conducting experiments with experimental psychologists to evaluate the naturalness of the synthetic control of facial features motion, using an exhaustive search over the tunings of the control parameters. Two rendering techniques are considered: a 3D geometrical modeling of the face including texture map (in collaboration with David Sander, from Experimental Department of the University of Geneva), and a 2D image-based approach using re-synthesis of video of real faces (in collaboration with Edouard Gentaz, from UFR de psychologie expérimentale, Université Pierre Mendès-France, Grenoble).
