I am currently working on a build, my goal is to keep it under 5k parts. The build is pretty high detail, it has elements of both low-poly builds and realism, somewhere in the middle.
I’m about 1/5 of the way through the build, already above 1K parts. This will be a pretty large-scale map, a lot of space. There are a lot of parts that I can group together and convert to meshparts to reduce the part-count. However, I don’t know how effective that tactic is. Will doing this make a noticeable change? Isn’t the game still having to render the same amount of faces in these meshes/parts anwyays? I know at least by converting to mesh, I can turn automatic render-fidelity on those meshes. I’m curious to know if this method is actually effective though.
Regular parts are supposed to be very optimized. Converting a bunch of parts to a mesh its only practical if:
A. There are a lot of unnecessary (internal) faces that players will never see.
B. You’d like to be able to easily resize, place, and maneuver the chunk of parts multiple times elsewhere.
A good amount of parts? Or one extra mesh to load at the beginning of the game. I would say for just a couple conversions, the performance difference is negligible (after initial loading).
At some point, meshes might be the better option for large chunks of parts that could be optimized further. But then again, a lot of unique meshes means more to load on join, so its a bit of a balancing act.
EDIT: When I say converting to a mesh, I mean exporting to blender and cleaning up extraneous faces, and then re-importing. Exported parts will maintain positional data, so re-importing is easy. Rotation and position will be flipped though, so just reverse the values and it will be back where it was initially.