8. Conclusions and future work
The paper contributes to the field by proposing a formalization of the front concept and a new object-aware algorithm for the interactive inspection of huge models which uses a rendering budget and supports selection of individual objects and sets of objects, displacement of the selected objects and real-time collision detection during these displacements. As far as we know, no present algorithm addresses rendering such huge sets of objects with guaranteed frame rates, while allowing for the modification of individual objects during the inspection and attempting to optimize image quality. Our proposal addresses these needs, and is also able to verify on the fly possible collisions when moving objects around. The proposed algorithm has proved to successfully solve a precise need in the design of huge assemblies and is now being used in real ship design environments. In our tests, we have not perceived differences between the quality of the images rendered by our algorithm with respect to the ground truth represented by the input models. Isolated frames may present transient artifacts when the visibility changes abruptly, since the updates exceed the budget, but they very quickly disappear. Although we have not discussed it here, our current implementation supports textured polygons and selection through attributes. To be able to fully support textured models, however, we need to further improve the ORIs to handle them. This is intended as future work. Because of the locality in space afforded by the Kd-trees, the whole pre-processing is amenable to being done out-of-core, and hence our approach should scale well for even larger models and modest hardware. Another avenue of improvement is therefore the implementation of this out-of-core approach in a fully automatic way.
