Cloud animation using Machine Learning

Civil engineer thesis topic for game programming students

ANNONSINFORMATION

Sista datum för anmälning: 2020-10-31

Företagsnamn: BTH

Mail till kontaktperson: prashant.goswami@bth.se

Länk till företag: http://www.bth.se

Utbildningsnivå: Magister/Master

Utbildningsområde: Civilingenjörsutbildningar Datavetenskap och mjukvaruutveckling Digitala medier och spelteknik

Background:
Physical simulations play an important role in enriching realism in the domain of computer graphics.
Most of these simulations however, are driven using physical models which place a serious restriction on their efficiency and realism. With the emergence of GPUs and machine learning methods in several fields, we are now confronted with the possibility of exploring algorithms that can mimic existing data patterns.

Aim and objectives:
The aim of this project is to leverage existing animation evolution patterns extracted from the videos, to guide the cumulus cloud animation. In the first step, the problem will be mapped suitably to be able to make learning possible on the extracted video frames. In the next step, the learning step would generate plausible animation patterns for the clouds. Further, in this step the extracted animation would be included within the existing framework for cloud animation. The students will have a reasonable room to experiment and improve the project in the scientific and implementation areas.

Requirements:
Student(s) should be advanced level programmers in C,C++,..
Knowledge in GPU programming required
Knowledge in Python, Matlab needed
Knowledge of Artificial Intelligence and machine learning needed
Interest and adaptability in the underlying physics and simulation

Related links:
[1] Real-time landscape-size convective clouds simulation, Prashant Goswami & Fabrice Neyret, VRIPHYS 2015
[2] Generation of synthetic plant images using deep learning architecture, Ramya Sree Kola, Master Thesis, 2019

Tillbaka