1. EVASION Environnements Virtuels pour l’Animation et la Synthèse d’Images d’Objets Naturels Virtual Environments for Modeling, Animating and Rendering Natural Scenes INRIA Rhône-Alpes Équipe du laboratoire GRAVIR/IMAG Future équipe du LJK (CNRS, INPG, INRIA, UJF)

2. Who are we? The team –6 faculties 2 full profs : GP Bonneau (UJF), MP Cani (INPG): Scientific leader 2 assistant profs : F Faure (UJF), F. Hetroy (INPG, sept 2004) 2 CR1 researchers : F. Neyret (CNRS), L. Revéret (INRIA) –4 post-doc, engineer or designer –12 PhD students (8 with MENRT grants) History –Created in January 2003, after the scission of iMAGIS –Basis: Computer Graphics (modeling, animation, rendering)

3. Scientific focus Modeling & Visualizing Nature Fascinating problem (vegetable, mineral, animal worlds) Still unsolved to a large extent Many industrial applications (from realism to real-time) –3D feature films, Special effects, Video games –Virtual prototyping & Pedagogical Simulators (environment, geology, energy, aeronautics, surgery, cosmetics)

4. Modeling & Visualizing Nature Main challenges Extreme complexity – Number of elements, shape, aspect, motion and deformation Re-using models from other sciences is not always possible –Virtual clouds? Fluid dynamics & meteorology study other scales –Hair animation? FEM + collisions not applicable for 100 000 strands Use existing knowledge: Collaborate with other disciplines Combine efficiency and realism? Specific methodology + New fundamental tools

5. Scientific basis Methodology for handling complexity 1.Characterize the observed sub-phenomena 2.Represent them by coupled sub-models –Of different nature : physical model, geometry, texture,... –Applied at different scales 3.Dynamically adapt the sub-models to the needs –By changing their local space and time resolution –By switching from one model to another 4.Validate based on human perception

6. Methodology for handling complexity Example: meadow blowing in the wind 1.Wind : pattern + action 2.Receever : precomputed dynamics 3.Grass geometry : 3 levels of detail … [I3D’01,Computer Animation’03]

7. Contributions 1. New fundamental tools Geometry –New shape representations –Interactive deformations Animation –Motion control from video analysis –Physically-based simulation Visualization of massive data-sets –Multiresolution analysis & adaptive rendering Realistic rendering –Textures, shaders, point-based rendering

8. New fundamental tools Example: Constant volume space deformations Foldover-free space deformation Rings of constant volume « swirls » Applications Modeling virtual clay Animating fluids [Pacific Graphics’04, SCA’05]

9. Contributions 2. Application to specific natural scenes Mineral world –Animation of lava-flows, sea, streams –Simulation of water, smoke, clouds Vegetable world –Real-time rendering of forest –Animating meadows (grass, trees) Animal world –Wild animals animated from video –Virtual humans: hair, skin, muscles, clothes –Real-time organs for surgery simulators

10. Application to specific natural scenes Example: Simulation of Natural Hair New Lagrangian deformable model: Super-helices –Predicts the shape of static hair –Efficient and stable simulation of hair dynamics Identification of hair interaction parameters Bridging the gap between wisps & continuum Interdisciplinary work (cosmetics, mechanics) Industrial partnership (L’Oréal) [EG’05 short, SIGGRAPH’06]

11. Application to specific natural scenes Example: Simulation of Natural Hair

12. 3. Software development SOFA with CIMIT/Harvard, INRIA, ETHZ, CWU An Open Framework for Medical Simulation Multi-institution, international effort Aim: component sharing / exchange / comparison Kernel (release Dec 06) –Communication & interfaces Modules –FEM, Mass & springs, Particles –Rendering algorithms, –Collision detection & response

13. Scientific Collaborations International –Joint team with DGP, University of Toronto (2004-2006) –6 Eurodoc grants: 6 month visit of PhD students to U. of Washington, Davis, Berkeley, Calgary, Montreal –European Network of Excellence: Aim@shape (other joint papers with UBC, ETHZ, U. of Tuebingen, UC Davis) National –Co-advised PhDs: SIAMES, MOVI, APACHE, LMC, TIMC –DEREVE 2 with LIRIS & ICA, MIDAS with TIMC, ICP –ARCs with ALCOVE, EPIDAURE, ISA, Geometrica

14. with Other Disciplines 2003-2005: Collaborations with the fields of –Mechanics (CEMAGREF, LEGI, L3S) –Medicine (IRCAD, TIMC) –Cognitive Sciences (U. of Geneva) –Cosmetics (l’Oreal research labs) 2005-2009: Interdisciplinary research clusters –“Environnement” & “Santé” (Rhône-Alpes Region) 2006-2009: Multidiciplinary ANR Projects –Biomechanics & Neurosciences (project Kameleon) –Botanics (project NatSim) NatSim Kamelelon

15. Industrial grants & transfer Public projects with technology transfer –European project Odysseous –RIAM Virtual Actors & RNTL PARI with Galilea –RIAM projects Vertigo & Prodige with Bionatics and Thales Direct grants from the industry – L’Oreal (contract 2004-2006) –CEA / CESTA (PhD grant 2004-2006) –EDF (PhD grant 2005-2007)

16. Results & Visibility Publications (20 journal, 48 conf, 6 chapters…) Editors: GMOD, IEEE TVCG Conference co-chairs –EG-IEEE Visualisation’2003, IEEE Shape Modeling & Applications’05 Paper co-chairs –EUROGRAPHICS’04, ACM-EG Symp. on Computer Animation’06 PC members –SIGGRAPH, Eurographics, Pacific Graphics –IEEE Vis, SMI, SCA, CASA, NPAR, etc

17. Grand challenge ? Specify and control a full, animated natural scene Creation of digital content, in a difficult case –High number of similar, but different details –Allow user-input / fit specific distributions –Control motion while maintaining realism –Animate and render efficiently Reasons for tackling it? –Real-size tests & interactions between different phenomena –Interactive exploration (GPU, GRimage PC grid) –Validation through the science of human perception

18. Objectives for the next 4 years 1.Creation of Natural Scenes Exploit real images, data, sketching Combine user control with procedural details 2.Animating Nature: multi-disciplinary projects Promote interactive virtual scenes as a support for experimenting and validating hypotheses Model natural phenomena never achieved in CG 3.Efficient Visualization of very large scenes Interactive exploration of hybrid data-masses Fast, realistic rendering of natural scenes

19. Conclusion Computer Graphics group –Competences: modeling, animation, visualization, rendering –Focus: Virtual natural scenes and phenomena Strategic aspects within French research –Combining simulation, visualization and virtual reality –Processing huge data-sets –Applications to Environmental simulations –Applications to Biology/Health-care EVASION

20. Thank you!