Visual Objects : from Reality To EXpression
Head : Hervé Luga
VORTEX aims to form a dynamic group working on visual objects. Our research efforts on visual objects, real or virtual, span a wide range of : still images, videos, 2D and 3D scenes etc. We want to design and develop methods, models and tools to deal with visual objects (real, virtual or mixed) up to give them, by creation or enhancement, elements allowing to adapt their use according to user features, applications, context. For this, we do not want to develop several models but a meta-model or a multi form model (a multi model approach) including several different sub-models, each one with its properties and inner limitations but seen globally, each model covers up drawbacks of the other.
Visual objects are thus central in our research activities to allow to create them, edit them, to give them knowledge, to manipulate then thanks to interaction capabilities and to share them through networked environment.
VORTEX is organized around three axis :
A multi model approach is developed to allow the use of 2D and 3D models ; real, synthetic or augmented ; static and animated ; for the design, the artistic expression, the communication, ...
We want to study models to capture visual object features (shape, motion, texture, ...). Two ways will be followed, the omnidirectional capture of static scenes by dedicated systems and the detection and characterisation of dynamic objects in image sequences : through motion (2D tracking, rigid or not, for instance), through shape and geometry (3D tracking from models, for instance), through texture (object detection by recognition, for instance), and more generally, through the study of visual clues matching.
We also want to provide models allowing building a 3D scene from diverse informations : pictures or videos captured from real, scanned 3D objects, and intuitively modelled 3D objects and synthetic images. We study expression mediums allowing facilitating the interaction between tools and users about sketching and semi autonomous tools for 3D production. We are also interested in real-time visualisation (at least interactive) of 3D scenes, in expressive rendering, in highly dynamic rendering, in merging real and virtual worlds, in lighting, and in complex surfaces properties. The crucial points are : registration ; estimation ; model learning ; model inversion ; information fusion ; model, data structure and algorithms development.
Aimed results are : pictures, videos and 3D objects fusion and manipulation (using sketching) methods, real-time or interactive global lighting and relighting methods, real-time 3D object robust reconstruction methods.
Foreseen applications are : user-friendly environment construction for communication and artistic expression ; product design, augmented reality environment construction.
The goal is the synthesis of virtual worlds in which autonomous or controlled creatures, acting in an evolutionary environment, develop properties inspired from Life (reproduction, co-evolution, self adaptation, self organisation, ...). Our purpose is to go from an individual conception approach for virtual space elements to a global approach allowing synthesizing them simultaneously. Our skills include shape and behaviour morphogenesis using evolutionary techniques, multi criteria optimisation systems acting in evolutionary / dynamic environments. In relation to autonomous entities, our work in the behavioural simulation domain includes aspects such as learning (classifier systems), communication, spatial coordination, decision making. To synthesize complete artificial creatures in complex ecosystems, our work is now oriented to the field of artificial embryogenesis (computational development) i.e. how to connect development and evolution to control cellular growth able to produce fitted organs or assemblies. This is built on massively parallel multi agent systems to obtain emergent system properties.
The fundamental problem is the simulation of behaviour showing properties such as cooperation/competition, adaptation, evolution, learning, anticipation, and memory, up to the synthesis of ecosystems. In connection to other fields, common problems are shape and scene production using advanced tools and coupling real sensors and artificial entities.
Aimed results involve animation in order to animate groups or crowds, to populate artificial worlds of credible autonomous or controlled creatures ; artificial life to synthesize complex artificial systems from properties (physical, structural, visual, functional, perceptual, behavioural,). In synergy with the other fields, results will be to automatically generate product configuration for functional simulation and to provide advanced mechanisms for adaptive (enactive) interaction with visual objects in the context of augmented reality applications.
Targeted applications include behavioural simulation of situated virtual characters (in games, simulation, ...), pedestrian or vehicle simulation, town or population growth up to complex system simulation : major risk management, economical macro simulation, optimisation. Applications also include the help to design, conception and decision (in the scientific, industrial and artistic domains) and the conception of structures and systems for industries and the evolutionary robotic research field.
The goal is the development of methods, techniques and tools allowing the shared interactive use of 2D or 3D models, real or synthetic, static or animated, in virtual reality or augmented reality worlds.
Visual objects managed here may be created or acquired by Axis 1, and may have evolved and been adapted by Axis 2. Objects (called “smart visual objects”) are the active content of our visual environments. We can also find here multi model, multi representation, and multi scale notions as we are leaded to manipulate, share, communicate, and adapt visual objects. Objectives involve : interoperation of heterogeneous interactive cooperative 3D simulations, intuitive and non-intrusive immersive collaborative 3D interfaces and the adaptation of visual objects to heterogeneous exploitation environments.
The main problems tackled here are : distributed cooperative simulations and exchanges between distributed applications, including optimisation of communication between applications as well as links between real and virtual worlds, interactivity, immersion, mobility and intuitiveness in distributed virtual and augmented environments.
Targeted results would be first coupling multi model / multi scale / multi resolution multimedia data to support concurrent uses of distributed environments then enhancing the interaction and immersion for several users thanks to merging virtual reality, interaction and motion capture or merging virtual reality, augmented reality / video and distributed multimedia.
Foreseen applications are distributed VR/AR environments and multimedia applications, with a special interest in massively multi user games/simulations, mobile operators/users support, interactive rooms, up to ubiquitous interactive self-adaptive multimedia information systems. Another targeted application domain is the co-design and the co-simulation, in the context of concurrent engineering with the coupling of design and heterogeneous multi domain simulations on 3D functional virtual mock-ups using collaborative environments (intuitive and non intrusive 3D interaction) for distributed simulation and interactive workspace. More generally, application domains of such complex 3D virtual environment are, for Sciences : experiments “in virtuo” ; for Industries : training, collaborative engineering, virtual prototyping, virtual product lifecycle simulation, for the Society : education, territory planning, communication, marketing, entertainment, ...
Cette rubrique ne contient aucun article.