Originally posted by Bobboau
you realy think that haveing two 1000+ polly versions of the same subobject that are identical exept one has a subobject and the other dosent is more effectave use of file space than simply makeing another subobject to replace it?
Now that's twisting my point. The example I'd use is Nico's Ezechiel. The cockpit, all of the little grill-like extrusions, and all extra submodels and textures go away at LOD1, which takes the detail from at least 2000+ polys across 4 textures (counting the cockpit, the number, the pilot, and the hull) to about 1000 across 1 texture. LOD1 is detailed enough to pass scrutiny at everything but short range, so ALL of the extra detail goes away, along with three additional render passes. Plus it's the first of three LODs, which I would in no way advocate giving up. LOD1 on fighters
has to look good anyway, so all you're going to peel off of it regardless of whether you do subobject LODs or not is going to be the close-up detail that can't be seen from more than a few fighter-lengths away anyway (i.e. the cockpit). Therefore, having a second cockpit model in the mesh is redundant. IMHO.
Unless, of course, the target window no longer renders LOD1. In which case, the LOD system may be somewhat redundant in and of itself, in the manner in which it is currently used.