Section: New Results

Motion & Sound Synthesis

  • Scientist in charge: François Faure.

  • Other permanent researchers: Marie-Paule Cani, Damien Rohmer, Rémi Ronfard.

Animating objects in real-time is mandatory to enable user interaction during motion design. Physically-based models, an excellent paradigm for generating motions that a human user would expect, tend to lack efficiency for complex shapes due to their use of low-level geometry (such as fine meshes). Our goal is therefore two-folds: first, develop efficient physically-based models and collisions processing methods for arbitrary passive objects, by decoupling deformations from the possibly complex, geometric representation; second, study the combination of animation models with geometric responsive shapes, enabling the animation of complex constrained shapes in real-time. The last goal is to start developing coarse to fine animation models for virtual creatures, towards easier authoring of character animation for our work on narrative design.

Physically-based models

We proposed a survey on the exhisting adaptative physically based models in Computer Graphics in collaboration with IST Austria, University of Minnesota, and NANO-D Inria team. Models where classified according to the strategy they use for adaptation, from time-stepping and freezing techniques to geometric adaptivity in the form of structured grids, meshes, and particles. Applications range from fluids, through deformable bodies, to articulated solids. The survey has been published as a Eurographics state of the art [13].

In collaboration with the Reproduction et Développement des Plantes Lab (ENS Lyon), we proposed a realistic three-dimensional mechanical model of the indentation of a flower bud using the SOFA library, in order to provide a framework for the analysis of force-displacement curves obtained experimentally [12].

Simulating paper material with sound

Figure 6. Left: Geometrical deformation using our geometrical model from [5]. Right: Various paper deformation and type leading to different synthesized sounds [24].

We developed within the PhD from Camille Schreck a dedicated approach to model a real time deforming virtual sheet of paper. First we developed a geometrical model interleaving physically based elastic deformation with a dedicated geometrical correction and remeshing. The key idea consists in modeling the surface using a set of generalized cones able to model developable ruled surfaces instead of the more traditional set of triangles. This surface can handle length preservation with respect to the 2D pattern, and permanent non smooth crumpling appearance. This geometrical model published in ACM Transactions on Graphics in Dec. 2015 [5] has been presented at ACM SIGGRAPH this summer and is currently under investigation to be part of Inria Showroom. This model has then been extended to real time sound synthesis of crumpled paper within the collaboration with Doug James (Stanford University). This method was the first to handle real-time shape dependent sound synthesis. During the interactive deformation, sudden curvature changes and friction are detected. These sound generating events are then associated to a geometrical region where the sound resonates and defined efficiently using previous geometrical model. Finally, the sound is synthesized using a pre-recorded sound data base of crumple and friction events sorted with respect to the resonator region size. This work has been published at Symposium on Computer Animation [24] and received the best paper award.

Human motion

Figure 7. Live tracking and visualization of a plausible anatomical skeleton following the pose the subject [17].

Armelle Bauer defended her PhD in November, co-advised with TIMC (Jocelyne Troccaz as principal advisor), on Augmented Reality for the interactive visualization of human anatomy. This is one of the main achievements of the Living Book of Anatomy project, funded by Labex Persyval. This work was partly published at the Motion in Games conference (MIG 2016) [17]. It served as a basis for the follow-up ANR project Anatomy2020 involving Anatoscope, TIMC and LIG laboratories, and Univ Lyon 2.