Declarative Modelling of Virtual Environments DEM 2 ONS PROJECT 2 - PowerPoint PPT Presentation

Declarative Modelling of Virtual Environments DEM 2 ONS PROJECT 2 ONS PROJECT DEM (Declarative Multimodal ModeliNg System) (Declarative Multimodal ModeliNg System) (Visual Objects: From Reality To Expression) VORTEX Team (Visual Objects: From Reality To Expression) VORTEX Team IRIT – Toulouse III IRIT – Toulouse III

VORTEX Research topics Complex shapes & Behavioral Simulation 3D environment modelling Declarative modelling Declarative modelling Constraints, Optimisation Constraints, Optimisation Animation of characters Distributed & cooperative simulation Meshes Points based modelling Collaborative virtual Medical imaging & Visualization on prototyping Related applications very big screens Rendering & visualization Virtual Reality 2

Declarative modelling: Review of our works DEM²ONS Project Constraints Application solvers prototypes prototypes experiment platform of multimodal modeller ORANOS (1998) DEMONS’93 (Numeric CSP) non robust isothetic layout 3

DEMONS_ORANOS  Dynamical and hierarchical N-CSP [Kwaiter 98] Isothetic Final Scene with orientation fixed by the designer(~25 objects) 4

Declarative modelling: Review of our works DEM²ONS Project Constraints Application solvers prototypes prototypes experiment platform of multimodal modeller ORANOS (1998) DEMONS’93 (Numeric CSP) non robust isothetic layout MANHATTAN (2002) (Geometric CSP) isothetic layout more efficient than Oranos 5

DEMONS_Manhattan: Geometric CSP  Search algorithm: backtracking + dynamic filtering + heuristic into a discrete space [Le Roux 02] Isothetic ! 28 objects, 24 possible orientations in 3D, description by script ~10 sec. of generation + rendering time (ray-tracing) 6

Declarative modelling: Review of our works DEM²ONS Project Constraints Application solvers prototypes prototypes experiment DEMONS-GA (99- platform of 03) multimodal (genetic algorithm) modeller ORANOS (1998) non-isothetic layout DEMONS’93 (Numeric CSP) non robust isothetic layout MANHATTAN (2002) (Geometric CSP) isothetic layout more efficient than Oranos 7

DEMONS_GA  Genetic Algorithmes [Sanchez, Le Roux 03] Non-Isothetic 88 objects, description by script ~2 min. of generation + rendering time (ray-tracing) 8

Declarative modelling: Review of our works DEM²ONS Project Constraints Application solvers prototypes prototypes DEMONS- LE (2003) ORANOS (1998) (Numeric CSP) DEMONS-GA (99-03) non robust (genetic algorithm) isothetic non-isothetic layout layout MANHATTAN (2002) (Geometric CSP) isothetic layout more efficient than 9 Oranos

DEMONS_LE  Metaheuristics from local search [Larive 03] Non-Isothetic « drag and drop » from the interface towards the scene (interaction + generation in a few seconds) 10

Review of our work in declarative modelling DEM²ONS Project Constraints Application solvers prototypes prototypes experiment ADMUNSEN (2003) DEMONS- platform of (Numeric CSP) LE generative non-isothetic (2003) processes ORANOS (1998) layout DEMONS’03 (Numeric CSP) non robust DEMONS-GA (99- isothetic layout 03) (genetic algorithm) non-isothetic MANHATTAN (2002) layout (Geometric CSP) isothetic layout more efficient than 11 Oranos

DEMONS_Admunsen: a tenacious explorer  Numeric CSP ... and non-isothetic ! 12

DEMONS_Admunsen: Object position tags  Example: the in front of in front of constraint Characteristic points Quadratic location zones 13

DEMONS_Admunsen: Tags  Object position tags The chair against against the table On the chair Under the chair In front front of of the chair On Under In 14

Conclusion  Critical steps: – Interpretation of properties – Efficient generation with guaranteed results (CSP) or not (Metaheuristics) – Take into account the direct modifications, but keep the consistency with the result of the generation step  Generalization to the complex shapes: – Features of complex shapes ? – Sketching – Relevant combination of textual and gestural interactions  Need to collaborate with qualified people: – artists, architects, designers, etc. 15

Declarative Modelling Related applications Related applications (Visual Objects: From Reality To Expression) VORTEX Team (Visual Objects: From Reality To Expression) VORTEX Team IRIT – Toulouse III IRIT – Toulouse III

Declarative modelling of virtual tows [M. Larive]  Automatic generation of realistic digital mock-ups of towns: – Geographic or social maps: • Population density • Zone maps (residential, commercial) • Street patterns • Elevation or hydrographic maps • etc. 17

Wall grammar for automatic building generation [M. Larive]  Generation for any building footprint, convex or not, even with holes – 2.5D wall grammar based on a set of rules – Availability of various kinds of roofs independently of the footprint complexity – Used in a commercial terrain modeler 18

Wall grammar for automatic building generation  Urban area (17 362 buildings, 920 182 faces) – Exemple of generated buildings (generation in 7mn 55sec !) 19

SIAMA project (Système d’Imagerie et d’Analyse pour le Mobilier Archéologique) [C. Maïza]  Symbolic classification of archaeological vessel – Qualitative e Qualitative e valuation (declarative) of the similarity • Allow a first sort in a huge solution space – Quantitative e Quantitative e valuation • Rigid matching • Computation of the geometrical distance between two forms • Elastic matching (excessive tolerance of the archaeological data) 20

Symbolic segmentation of the brain [F. H. Andriamanankoavy]  Data fusion, Symbolic Symbolic matching of volumetric data – Knowledge (about the brain) • Reference data bases (healthy patients, phantoms) • Expertise Properties  Properties – Applied to medical imaging and palaeontological data (bones) 21

Approximative shape modelling [Anca Alexe]  Sketching – Intuitive 2D Sketching – Easy 3D adjustment (skeleton or boundary)  To do: Constrained manipulation 22