Skinned Mesh Count more or less better?

Small debate in the office. I wonder what is better.
Same model Polycount about 7000-8000 polys. Since i am Responsible for Rigging i like to cut the model up in about 10-20 pieces such as leftshoe,rightshoe,head,shirt, and so on which is easyer to skin or to replace.

In unity inside the Game/Stats the number of Visible Skinned Meshes Reduces for Meshes that you can’t see.
So by my understanding i would say it’s not rendered and should eat up less performance.
I know that more meshes give more Draw Calls.
But wouldn’t it be better as if i make one big mesh which is always Rendered?

This honestly depends on many factors: bone count, bone influences per vert, targeted platform, complexity of the model’s UVs/hard edges, HOW MANY of these objects will get drawn on screen, etc. [for example, if you had one 8000 tri sphere with a single spherical UV mapping on it, skinned to a single bone… unity could render tons of those with minimal issue… add in a bunch of hard edges (which adds vertex after export), lots of UV seams, 20 bones, etc… and Unity can start to choke depending on how many you are ultimately drawing.

In Unity, SkinnedMeshRenderers do not dynamically batch, so adding those extra meshes WILL add the extra draw calls as you say. If you were targeting an iPhone game, draw calls are a HUGE thing and shouldn’t be underestimated, and if it were me I would go with the single mesh solution (but, at 8000 polys, I doubt this is an iPhone character- or you have bigger fish to worry about!) :wink:

Also, triangle culling is also possible (i.e. not drawing the offscreen portion(s) of the model)… but this is another one that might be more expensive than simply rendering a bunch of polys.