thanks for the answers, satan.
I will do the bounding spheres trick and possibly use the idea of multiple bounding spheres. I will take spheres over boxes because they seem to be easier to work with and will just do for my needs.
Now for last: For testing intersection, i have come across the suggestion on partitioning the model with BSP trees. Following that, i could easily discard several objects that would definately not be being hit by the ray. That would be used on looking for whole models (bounding spheres) hits, aswell as, after hitting a sphere, looking for which triangle was (if any) hit.
I have implemented binary trees before but no BSP trees. I suppose it should be like implementing a standard binary tree and a buildBSP() method or something like that.
Is this solution really worth it or not? If so, is it worth implementing BSP tree for searching BOTH model hits and then specific triangle hits (as described in paragraph 4)?
It seems to me that i would have to rebuild the BSP tree at each frame depending on objects and “camera” position: again sounds like big overhead! Reading the arrays, building the triangles and sorting them to the tree? Is that how it’s supposed to work?
It just came to my mind a possible solution as sorting the vertex arrays as a Heap structure, would this be an employed solution by any chance?
Thanks in advance! Sorry if it’s somewhat confuse, i tend to have a brainstorm while writing.