The objective of the joint laboratory AllegoRI is to iniate a joint research activity in 3D science and technology between Telecom ParisTech and the Allegorithmic SME. This activity aims at developing new milestones in 3D digital content creation (3D DCC) which plays a major role in emerging media, such as video games, VR, digital fabrication and web3D, by exploiting the scientific expertize of Telecom ParisTech on one side, and the industrial establishment of Allegorithmic on the other side. Over the last few years, this company has given rise to a new hybrid methodology for the dual control of digital materials and textures which cover virtual 3D scenes, with on one hand a programmable control of smart materials, adapted to engineers workflows, and on the other hand an interactive control of the material model, accessible to digital artists.
The goal of AllegoRI is to generalize this approach to other computer graphics pillars, such as shape and illumination modeling. In this context, the main challenges are:
1. the growing amount of 3D scenes to design, for which existing tools do not scale and require a significant amount of manual work ;
2. the content reuse (animation parameters, illumination, dynamic shapes) from one scene to another is hard, often impossible, since 3D models are tightly linked to the spatial structure of the 3D environment in which they have been designed ;
3. the growth of use cases for these technologies implies that more diverse professionals, with different skill sets, have to collaborate aound a given 3D model, and to contribute to it using a specific interface, adapted their particular capabilities, from engineers to artists, through decision makers.
Three research directions are proposed.
1/ "Meso-structure modeling". Just as color maps for appearance (micro-structure), geometric meso-structures (e.g., wicker) are self-similar, often tilable and can be described independently from the macro-structures that carry them (e.g., chair). This direction aims at developping a new representation model for geometric meso-structures. This model will provide the same control primitives as classical texttures (procedural generation, multi-scale representation, tiling, mapping) but will be able to reproduce geometry to its full extent, including the complex underlying topological components which make these operations highly non-trivial and, sometimes, ill-posed.
2/ "High-level interface for macro-structure control". Within the growing market of 3D technologies (games, VR, design), offering non-expert users the ability to exploit advanced 3D modeling mechanisms through a simple and natural interface has become critical to reach success with a large and diverse panel of user profils. The major challenge here boils down to the semi-automatic relationship one can establish between a number of geometric parameters for a complex shape and a compact set of control primitives.
3/ "Reusable illumination templates". The illumination of a virtual 3D scene is the result of light transport simulation on the geometry of the scene. To that respect, although its design may require a significant amount of work, it is hard to transfer a given light setup from one scene to another. The ability to define procedural rules allowing to "program" lighting in a scene with a simple visual programming language, as well as the capability to "paint" light interactively are two important missing technologies to bring lighting design to a new era. To develop them, approximation models in light simulation caching will be exploited and enriched to establish them as programmable and and interactive components of the scene.
Monsieur Tamy BOUBEKEUR (Institut Mines-Télécom - Télécom ParisTech)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
LTCI - TELECOM ParisTech Institut Mines-Télécom - Télécom ParisTech
Help of the ANR 300,000 euros
Beginning and duration of the scientific project: January 2017 - 36 Months