Tim Scheller

I am a PhD student at LIX Ecole Polytechnique, working on creative AI in computer vision and graphics, supervised by Maks Ovsjanikov and Marie-Paule Cani.

Previously, I received my Masters in Computer Science from TH Cologne during an exchange at LIX in 2023, developing generative AI models for 3D scans. In winter 2020, I co-founded PhoGrow to provide digital twins of plant growth, using neural surface reconstructions. From autumn 2018 to summer 2022, I worked at Charamel, to synthesise photorealistic humans for 3D real-time applications, starting as a full-time intern and continuing part-time during my studies.

My passion is to push the boundaries of state-of-the-art technologies, and prototyping solutions for tomorrow's challenges, as in the sustainable computer science Hackathon on Fast Fashion, which I co-organised.

Since 2023, I have been working full-time on creative AI at the Computer Science Laboratory of the Ecole Polytechnique in Paris (LIX). I started as a research engineer in the Vista research group working on deep sketch-based modelling under the supervision of Marie-Paule Cani, and transitioned to the Geometry Deep Learning research group led by Maks Ovsjanikov.

From 2019 until 2022, my responsibility at Charamel was to develop a pipeline for synthesising photorealistic humans, which involves capturing, rendering and animating them in 3D. In addition, my role was to improve the company's web-based real-time graphics engine and authoring system, as well as to develop tools for animating and exporting content.

In 2018 I wrote my Bachelor Thesis during an internship at Charamel, where I implemented a WebGL application to render and animate digital humans in real-time. The former involved modelling subsurface scattering of skin, whereas the latter was based on performance capture, to model facial expressions and wrinkles.

In 2017, we developed a paragliding simulator in VR, where participants sat in a real harness to fly through a virtual environment. My role was to implement an inverse kinematic system for accurate and immersive body interactions, based on the user's movements.

In March 2023, we organised a 2-day hackathon to develop computer vision solutions dedicated to the problems caused by 'Fast Fashion'. Where some of the projects explored vision systems to detect the durability of garmins from simple smartphone-images of clothes. The Hackathon was funded by the European Research Council (ERC) and the Energy 4 Climate Interdisciplinary Center (E4C), to whom we are grateful for their financial support.

In October 2018, we hosted and organised a 2-day game jam to raise awareness about the environmental issues caused by coal mining. About 20 people participated and created 4 different games.