BTH students present AI-based visualization and virtual production for the regional industry
- Post by: Tobias Larsson
- 14th June 2025
- No Comment
2025 marks the first year of BTH students working in Virtual Production Studio Lab since its inauguration, who have now presented their BA thesis work in collaboration with and in support of regional industry.
“Collaborating with industry enables students to approach real-world challenges, and to practically prototype, demonstrate, and evaluate new ways of working to improve industry-relevant workflows. The students demonstrate that photorealistic real-time rendering, based on procedural and AI-assisted visualization, proves to be useful across various sectors, opening up new application possibilities and workflows for industries.”
Andrea Nordwall, BA student thesis supervisor, PhD candidate in Mechanical Engineering, focusing on visualization, XR, and virtual production.
Building on strong industry interest in AI-assisted visualization, real-time game engines, and virtual production, students have been exploring new possibilities in technology here is a short summary of the works:
Satellite-driven environment reconstruction in Unreal Engine: A simplified process for widespread adoption
Riis André, Johansson Jimmy
A user-friendly tool in Unreal Engine that has been developed to automatically generate large-scale 3D environments from satellite images and elevation data, using machine-learning segmentation and procedural content generation. It reduces the need for expertise and manual labor, and shows potential for time saving and further automation.
The method can be used for visualization, virtual production, or for generating realistic training data for autonomous vehicles such as drones, which the students demonstrated in a previous course in collaboration with The DEMINE Foundation to protect Ukraine:
Designing Effective UI for Remote Operations: Integrating UI Elements for Enhanced Teleoperation
Alvin Högman Lindblom
The study is conducted in collaboration with Volvo CE to integrate Dig Assist in teleoperation via 2D screens. An experimental method, iterative design, and agile project management are employed to evaluate and refine the interface in Unreal Engine 5. The result visualizes real-time critical information for tele-operated excavators.
Evaluation of AI models Flux and Adobe Firefly for matte painting in film
Olga Murinskaya, Caroline Nellerup
The purpose of the work is to investigate different AI models commissioned by the film company Istudios VFX. An assessment was conducted to determine the suitability of the models for professional use with minimal human creative intervention.
Dynamic Lighting in Virtual Production: Matching Foreground and Background for Filmmaking in LED-Based Virtual Production Studios
Caela Landin, Tim Jegård
The work explores new lighting methods for virtual film production, aiming to bring the foreground and background closer together in LED-based studio environments in collaboration with Istudios VFX. We developed methods using DMX and tracking and combined these to illuminate both the virtual environment and the physical scene simultaneously.
