A comparative evaluation of CAD-to-VR methods: how to efficiently use 3D CAD models for Virtual Reality applications

The increasing use of Virtual Reality (VR) in engineering requires efficient methods for transferring Computer-Aided Design (CAD) models into real-time visualization environments. Since CAD models are designed for engineering accuracy rather than interactive rendering, their conversion to VR involves trade-offs between geometric compactness, semantic preservation, and runtime performance. This thesis presents a comparative evaluation of fourteen CAD-to-VR transfer strategies using Unity as the target platform. Five mechanical CAD models were evaluated through two experimental campaigns: the first assessed geometric compactness and semantic preservation, while the second investigated runtime performance using frame rate, CPU and GPU frame times, rendering workload, and memory indicators. The results show that the selected transfer strategy strongly affects both model quality and visualization performance. STEP- and SLD-based strategies generated the most compact mesh representations and achieved excellent runtime efficiency. XML-based workflows produced the highest geometric complexity and the lowest runtime performance. The most balanced results were obtained with glTF- and GLB-based strategies, which preserved assembly structure, part names, and material information while maintaining efficient real-time rendering. FBX-based workflows provided strong semantic preservation, although their runtime behavior depended on the specific conversion and import route. OBJ-based workflows showed more variable results, with some routes achieving high runtime efficiency but limited semantic preservation. No single CAD-to-VR strategy is optimal for all applications. The most appropriate workflow depends on the required balance between geometric efficiency, semantic preservation, and real-time performance. The proposed evaluation framework provides practical guidance for strategy selection and contributes to more effective VR-based engineering visualization workflows.