August 8, 2013: People of ACM: Marie-Paule Cani
Today's Topic: People of ACM: Marie-Paule
Thursday, August 8, 2013
Marie-Paule Cani is professor of Computer Science at Grenoble Institute of Technology, where she leads a joint research group between Inria, the Centre National de la Recherche Scientifique (CNRS), and three local universities. With a background in Shape Modeling and Computer Animation, she has contributed to several high-level models for shapes and motion, such as implicit surfaces, multi-resolution physically-based animation and alternative representations for real-time natural scenes. Following her work on virtual sculpture, she has been recently searching for more efficient ways to create 3D content, such as combining sketch-based interfaces with procedural models. In 2011, she received the EUROGRAPHICS Award for outstanding technical contributions in recognition of this work. In 2012, she was awarded the Silver Medal from CNRS, which recognizes researchers at the middle of their careers.
Cani has served on the steering committee or chaired programs of several major international conferences (including SIGGRAPH Asia 2010), and recently joined the editorial board of ACM Transactions on Graphics. Director-at-large within the ACM SIGGRAPH executive committee from 2008 to 2011, she now represents Computer Graphics on ACM's publications board. She was recently elected Vice President of EUROGRAPHICS, the European Computer Graphics society. She received a Ph.D. in Computer Graphics at the University Paris XI in 1990.
With the advent of virtual environments, are you optimistic that your vision of a new convergence of Computer Graphics with Algebraic Geometry, Simulation, and Human Computer Interaction will become a reality in the near term?
The dream that drives my research is to develop more expressive ways for people to share and refine their vision of 3D shapes by enabling them to seamlessly model, convey and refine them. In addition to purely creative goals, the resulting tools will provide strong support for cognitive tasks by making visual thinking more generally available. The growing public interest for authoring favorite 3D environments, made easier by the spread of interactive tablets, makes me confident of the strong potential impact of this new generation of tools.
In terms of interface, the new technologies we are aiming at should be as easy to use as a pen. Meanwhile, the underlying models should be able to convey 3D shapes of any dimensionality and topological genus, either static or animated, as well as distributions of such shapes. In my research group, we are developing high-level models for geometry and animation, such as implicit and developable surfaces or frame-based simulation methods, together with interfaces for intuitive content creation. The latter include techniques to create 3D content from 2D sketches and to refine them by sculpting. This coarse-to-fine, intuitive modeling paradigm will be applied to shape design, motion design and to narrative design as well.
What challenges do you anticipate for widespread application of your advances in new modeling and animation techniques to digital entertainment, medical simulators, and virtual prototyping of natural objects and scenes?
Developing tools that can be useful for professionals from various backgrounds is even more challenging than making 3D creation available for the general public:
- Professional artists working in virtual entertainment technologies are trained on specific modeling pipelines. These artists are, however, open to new tools if they enable artists to save time on the less creative aspects of their tasks while still giving them full freedom on the content they are creating. For instance, developing novel deformation methods that automatically preserve the intrinsic properties of a shape, such as the volume behind a character's skin or the developable nature of cloth panels, can make a difference. (See our recent work on Design-preserving garment transfer presented at ACM SIGGRAPH 2012 and in ACM Transactions on Graphics.)
- Our general methodology can also be used for the virtual prototyping of natural scenes or objects. For instance, we collaborated with biologists to develop Structure from Silhouettes, a paradigm that enables multi-resolution, sketch-based design of 3D trees. We used a priori knowledge on plant architectures to infer 3D information missing from the sketches, leading to the efficient design of biologically plausible 3D trees. In the case of medical simulators, colleagues from my group contributed to SOFA, an open-source platform for real-time surgical simulation, and are currently developing MyCorporisFabrica, a visual encyclopedia for anatomical knowledge. In addition, we just developed Anatomy Transfer, a new method for seamlessly transferring anatomical models to different morphologies. Another challenge of medical simulators, which we may address in the future, is to provide intuitive narrative environments enabling specialists to easily design and refine surgical training sessions.
How broad is the impact on Computer Graphics of the geometric method of animation known as Implicit Skinning, which you just presented at the ACM SIGGRAPH conference in July 2013?
The production industry makes heavy use of skinning techniques, i.e., methods for computing the deformations of the skin of a character from the animation of their skeleton. The Implicit Skinning method addresses their specific needs: contrary to physically-based simulation, it runs in real-time and fits in the usual production pipeline. Meanwhile, it is able to capture contact surfaces and bulges observed in the deformation of real skin. This method relies on the combination of standard mesh models with higher-level volumetric representations, namely implicit surfaces. The composition operators used for the skinning application belong to Gradient-based Implicit Blends, a recent advance which solves a number of long-lasting problems in implicit modeling and will hopefully lead to the revival of this research area.
What advice would you give to budding technologists who may be interested in pursuing careers in the virtual world created by computer graphics?
Because of its visually attractive and fun applications, Computer Graphics is one of the areas that most easily attract students to Computer Science. Newcomers should, however, be aware that efficient algorithms, complex mathematical and physically-based models are often hidden behind the nice images. Being a broad scientist who is not afraid of mathematics, and is open to learning concepts from other disciplines is important to success in this area. The second key quality is creativity, since designing virtual worlds gives you the power to create your own virtual universe by even changing the laws of physics if you need to. The last advice I would give is to keep a fresh spirit and remain excited by crazy new ideas, although persistence and hard work will be needed to reach the end of your project.