About OpenSpace

OpenSpace is open-source software that aims to visualize the entire known universe in accurate three-dimensional space. It is mature science visualization software that can be used by anyone to tell stories about the cosmos and our place in it.

The Project

OpenSpace relies on a flexible scripting system to describe and control visualizations of astronomical phenomena. This system uses assets defined in Lua — text-based configuration files that describe how celestial bodies, trajectories, spacecraft, or other elements are rendered and animated in the virtual universe. Building these assets, however, requires significant technical knowledge of the internal data structures and scripting conventions used by OpenSpace. This thesis project explores whether Large Language Models (LLMs) can be used to assist in the creation of these assets — an “Asset Builder” powered by AI. The Asset Builder would function as an intelligent code-generation assistant, capable of producing valid Lua code for OpenSpace assets from natural-language descriptions such as “Create a scene showing the Voyager 1 trajectory with labels for each planetary flyby” or “Add a Mars globe with an orbiting satellite showing live telemetry data.” The main goals of the project are:

• To investigate methods for using LLMs to generate syntactically and semantically correct Lua asset definitions.
• To explore approaches for validation of the generated code, ensuring that it is compatible with OpenSpace’s scripting engine and visualization pipeline.
• To design a workflow or interface through which users — from novice students to experienced visualization designers — can iteratively refine the generated assets. This is a highly sought-after and strategically important project for the OpenSpace community. A successful Asset Builder would significantly lower the barrier to entry for new contributors, speed up the creation of complex visualizations, and open new pathways for combining AI-assisted design with scientific storytelling. Students undertaking this project will gain hands-on experience with modern generative AI techniques, scripting-based visualization systems, and interactive tool development for scientific applications. Collaboration with the OpenSpace research and development team will be part of the work. The thesis will take place at Visualiseringscenter C — a unique, cutting-edge environment where public outreach meets advanced visualization research, featuring technologies such as a state-of-the-art 3D dome cinema.

Application

Please send an email indicating your interest for this or other thesis works to ylva.selling@liu.se including your CV/LinkedIn and a Ladok transcript of courses. Also include an explanation of why you are interested in this specific project. If you are already a pair wanting to work on the thesis together or you are interested in multiple thesis works, a single email will suffice.

Applications are accepted on a rolling basis.