Skip to Main content Skip to Navigation

Synthèse de terrain à l'échelle planétaire

Yann Cortial 1, 2
2 Origami - Origami
LIRIS - Laboratoire d'InfoRmatique en Image et Systèmes d'information
Abstract : Allowing the real-time exploration of huge, detailed, heterogeneous synthetic terrains remains an unsolved challenge in computer graphics, despite forty years of research. In general, existing methods only handle terrains of limited extent, defined over a planar topology. In this thesis, we explore terrain modeling at maximum-scale, i.e. at planetary scale - a scale which exceeds by four orders of magnitude the extent of the domain of classical terrain synthesis. However, this problem has been largely neglected, only a couple of modeling methods have been proposed in past research. To tackle this challenge, we propose a novel decoupled solution, capable of generating true-to-scale planets. Our architecture takes first into account the geology of terrestrial planets by deploying a guided plate tectonics simulation. This simulation allows the generation of large-scale planetary features such as continents, oceanic relief, islands arcs and mountain ranges. This macroscopic model is then forwarded as an input to an hyper-amplification method capable of producing continuous, detailed real-time views of the terrain. To handle the size of the domain, the procedural amplification method relies on the pre-structuration of the tectonic model, allowing on-the-fly generation of the final detailed model, restricted to the current camera view. The model is produced entirely on the GPU by relying on a massively parallel stochastic subdivision scheme, guided by level-of-detail dependent rules. Overall, our method better handles certain problems tied to previous fractal methods, such as the self-similarity of homogeneous landscapes and the lack of user-control, by producing planets that show more variety, appear more realistic and can be more efficiently designed by artists.
Document type :
Complete list of metadata
Contributor : Abes Star :  Contact
Submitted on : Wednesday, March 31, 2021 - 12:38:21 PM
Last modification on : Tuesday, June 1, 2021 - 2:08:11 PM
Long-term archiving on: : Thursday, July 1, 2021 - 6:32:18 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03186765, version 1


Yann Cortial. Synthèse de terrain à l'échelle planétaire. Modélisation et simulation. Université de Lyon, 2020. Français. ⟨NNT : 2020LYSEI094⟩. ⟨tel-03186765⟩



Les métriques sont temporairement indisponibles