I wonder what the attached price is to that elaborate animations.
Namely I seem to recall, that each and every single subobject needs its own render pass similar to how each and every texture does. So fighters with lots of subobjects = as slow as if it had several textures. Now it's time for a code guru (probably Bobbau) to explain what this truely entails and whether the code can be optimized.
For textures it was pretty self explanatory why it had to be done as is: each new texture meant new data to be sent to the vidcard, whereas data from a sinlge texture could be sent in a sinlge batch.
I was wondering if some tricks could be done with the animation code to allow such batches - namely using 'identical' subobjects and the same texture.
Now I may be entierly in the blind here, but it was not geometry calculation that put the bender in the system but louding and unloading the data from the memory into the GPU.....though now that I thought of it , it will still entail several render passes, as lighting will have to be recalculated for each and every submodel. The renders that need recalculation are shine and enviroment mapping.
