Appearance preserving simplification of 3D CAD model with large-scale assembly structures

Abstract

Currently commercial 3D CAD systems which had only been used in upper stage of design scenes become widely used in lower stages such as rough sales purposes, model exporting to external users or e-commerce on Web. Usually, in the design stage, solid models even of very fine parts or full-detailed shapes have been built in the CAD systems, and the assembly models tend to have a huge number of parts and very complex inner structures. Moreover, for achieving the light-weight and strengthened parts, the inner structures of the housing such as ribs or bosses have had very complex geometries. However, when they are used for browsing, styling review and sales purposes, there is hardly the occasion where is full-detailed assembly models are required, and the primary purpose of the systems is often to fast render external shapes rather than to render detailed inner structures. Appearance preserving simplification of large scale assembly model available to the commercial 3D CAD systems is strongly needed for these purposes. Therefore, this paper proposes several appearance preserving simplification methods of 3D CAD model with large-scale assembly structures. Three simplification methods are proposed in the paper; (1) only by removing invisible parts from the assembly, (2) by removing both invisible form features from the part surface and invisible parts themselves from the assembly, and (3) by removing both form features and parts which are invisible even when position and orientation of movable parts change in the assembly. Our methods are based on an algorithm which can directly detect invisible parts or features by pre-rendering the models from multiple view directions and reading the rendered results from the frame buffer. Our algorithm can be carried out regardless of CAD systems. Thanks to using the current GPU, invisible parts or features detection is robust and fast in the algorithm. If needed, geometric dependency among the features in the assembly can be kept even in the simplification. The performances of these simplification methods in model size reduction and the processing time are examined.