Semantically Controlled LMV Techniques for Plant Design Review

Inspecting large industrial plants in a virtual walkthrough environment has proven to be a valuable tool in Plant Design. Many CG techniques, such as various LOD and culling methods, have been developed to visualize complex models in VR environments. These techniques decide solely based on geometric properties how to optimize the scene. In this paper we introduce the concept of semantically controlled selection of those techniques and show how semantic considerations can enhance the CAD to VR conversion process for large model visualization (LMV) walkthroughs of Plant Design models, improving the performance and adapting the visualization to the users’ needs. A taxonomy, together with semantic considerations coming from the relationship between user, model, and resources is the basis to decide which rules should be applied for a specific visualization technique. By extending a LMV walkthrough system we are able to reduce the complexity of large industrial plant models by a factor of two. On a common workplace PC the semantic preprocessing takes only 10-20 minutes for models with 10⁶ to 10⁷ polygons. Our approach is orthogonal to commonly known CG techniques and can be combined beneficially with those approaches.