Making of virtual laboratory : evaluating AI and software engineering tools to minimize manual work
Leghari, Mariyam (2026)
Diplomityö
2026
School of Engineering Science, Tietotekniikka
Kaikki oikeudet pidätetään.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe20260728112862
https://urn.fi/URN:NBN:fi-fe20260728112862
Tiivistelmä
Virtual laboratories (VLs) are increasingly used in education and training to provide accessible, interactive learning environments. However, creating realistic three-dimensional (3D) virtual spaces is time-consuming and often requires manual modelling or adaptation of repository assets. Recent advances in Artificial Intelligence (AI), particularly image-to-3D generation, offer opportunities to automate parts of this process, but their practical value within complete asset-production workflows has not been fully evaluated.
This thesis investigates whether AI-assisted image-to-3D tools can reduce the asset sourcing and integration workload involved in creating a SketchUp-based 3D space intended for virtual laboratory use. A repository-based workflow using SketchUp Free and 3D Warehouse was compared with AI-assisted workflows using Hunyuan 3D and Rodin AI. Six representative assets were evaluated. The comparison considered active user time, total pipeline time, visual resemblance, post-processing and integration effort, accessibility, and practical workflow usability.
The results show that AI-assisted workflows approximately halved the active user time required for asset production compared with repository-based sourcing, while achieving higher visual resemblance for several distinctive objects. However, the generated models required inspection, file conversion, import, modification, and placement before they could be adapted within the SketchUp Free model. Therefore, AI-assisted image-to-3D generation provide partial rather than complete automation of asset-production workflow. Repositorybased sourcing and AI-assisted generation should be treated as complementary approaches, with a hybrid workflow offering the most practical balance between workflow efficiency, visual resemblance, software compatibility, and asset availability.
This thesis investigates whether AI-assisted image-to-3D tools can reduce the asset sourcing and integration workload involved in creating a SketchUp-based 3D space intended for virtual laboratory use. A repository-based workflow using SketchUp Free and 3D Warehouse was compared with AI-assisted workflows using Hunyuan 3D and Rodin AI. Six representative assets were evaluated. The comparison considered active user time, total pipeline time, visual resemblance, post-processing and integration effort, accessibility, and practical workflow usability.
The results show that AI-assisted workflows approximately halved the active user time required for asset production compared with repository-based sourcing, while achieving higher visual resemblance for several distinctive objects. However, the generated models required inspection, file conversion, import, modification, and placement before they could be adapted within the SketchUp Free model. Therefore, AI-assisted image-to-3D generation provide partial rather than complete automation of asset-production workflow. Repositorybased sourcing and AI-assisted generation should be treated as complementary approaches, with a hybrid workflow offering the most practical balance between workflow efficiency, visual resemblance, software compatibility, and asset availability.