Arnaud Schoentgen

R&D Scientist

Ubisoft La Forge

Biography

I am an R&D Scientist at Ubisoft La Forge. La Forge brings experts from both the industry and academic sector to bridge the gap between theoretical research and game-making process, while contributing to solving real-world problems through scientific publications. At La Forge, I investigate new methods to enhance our virtual worlds with plausible, interactive, and art-directable assets.

Prior to joining Ubisoft, I was a Ph.D. candidate in Computer Science in both Université de Montréal and Université de Poitiers, where my advisors were Pr. Pierre Poulin and Pr. Philippe Meseure. My past experiences include natural phenomenon modeling, physics-based animation, artistic control and procedural generation.

Interests

3D Computer Graphics
Physics-Based Animation
Artistic Control
Procedural Generation

Education

Ph.D. in Computer Science, 2021

Université de Montréal / Université de Poitiers
MAE, General Management, 2016

Institut d'Administration des Entreprises de Caen
M.Sc. in Computer Science, 2016

Ecole Nationale Supérieure d'Ingénieurs de Caen

Experience

R&D Scientist

Ubisoft La Forge

December 2023 – Present Montréal, Canada

Define and develop research within the field of natural phenomenon modeling and physics-based animation.
Collaborate with team members and game production teams to develop new algorithms to generate more believable worlds, and to improve our content creation pipelines.
Stay up to date with the latest research in graphics and attuned to the needs of Ubisoft’s productions.

R&D Engineer

Ubisoft La Forge

September 2021 – December 2023 Montréal, Canada

Build prototypes and turn them into minimal viable products for game development teams.
Validate research conducted in the team via technical prototyping.
Explore new techniques to assist artists in the digital content creation and to create more interactive worlds.

Ph.D. in Computer Science

Université de Montréal

September 2016 – November 2021 Montréal, Canada

Thesis proposal: “Tools for Fluid Control in Computer Graphics”
Investigating new ways of performing efficient and intuitive art-direction over smoke and liquid numerical simulations
Reviewer for the Computer Graphics Forum journal and Pacific Graphics

Publications

Fast Orientable Aperiodic Ocean Synthesis using Tiling and Blending

We introduce a new ocean paving method based on tiling and blending, a procedural generation algorithm popular for real-time texture synthesis. This method also enables us to author the direction of the waves using a flow map. We show that this method is especially fast and can create a fully aperiodic ocean with minimal downsides.

Compact Poisson Filters for Fast Fluid Simulation

Poisson equations appear in many graphics settings including, but not limited to, physics-based fluid simulation. We propose a new Poisson filter-based solver that balances between the strengths of spectral and iterative methods. We derive universal Poisson kernels for forward and inverse Poisson problems, leveraging careful adaptive filter truncation to localize their extent, all while maintaining stability and accuracy.

Tools for Fluid Simulation Control in Computer Graphics

Physics-based animation can generate dynamic systems of very complex and realistic behaviors. Unfortunately, controlling them is a daunting task. In particular, fluid simulation brings up particularly difficult problems to the control process. This thesis presents two projects. First, we introduce a new particle-based liquid control system based on precomputed liquid animations. Second, a tracking solution for smoke upresolution is described.

Particle-based Liquid Control using Animation Templates

It is notoriously difficult for artists to control liquids while generating plausible animations. We introduce a new liquid control tool that allows users to load, transform, and apply precomputed liquid simulation templates in a scene in order to control a particle-based simulation.

A Density-Accurate Tracking Solution for Smoke Upresolution

Controlling smoke simulations is a notoriously challenging and tedious task. We propose a tracking procedure where we optimally modify the velocity field of the simulation in order to make the smoke density distribution closely follow the low-resolution density in both space and time.

Projects

Real-time Procedural Modeling and Rendering of Volumetric Clouds

Arnaud Schoentgen, Patrice Hirtzlin, Jean-Eudes Marvie

Cadomus Project - Place de la République

Nicolas Lefèvre, Charlie Morineau, Stéphane Gorzkowski, Philippe Fleury, Sophie Madeleine, Karim Sammour, Arnaud Schoentgen