Parented Meshes On Skeletal Animation - Virtua Fighter Style - Any Hidden Issues?

Hey all,

We’re building an RTS style game in a 3D environment. The bipedal characters are driven by a simple skeleton, about 18 joints. With the scale and art direction we can absolutely visually get away with very simple character meshes that are individual ovoids and shapes that are simply parented to the skeleton. Like an old Virtua Fighter model. Then the skeletal mesh can only have the ground shadow which makes them as fast as they can be.

I’ve already verified that if we do this and use one texture atlas for the material on all of those pieces it is only 1 draw call which is awesome.

However, I’m concerned I’m missing a drawback. This turns each character into at least 19 GameObjects. Thus far I’m not seeing any drawbacks and the rendering payoff seems worth.

Can anybody thing of any issues I’m missing? Anybody tried this before?

-L

I would guess the answer is yes, you don’t have to worry about it.

A simple test to reassure yourself would be to keep adding empty GameObjects to your scene until you see an impact on performance. If that number is much higher than what your character(s) will be adding then don’t worry about it.

Personally, I would question your first assumption: That using your setup (no skinned mesh renderer?) gives any performance gain at all. If you like the visual style or it’s much faster for you to make that’s fine, but don’t stick to it because of an assumed performance gain.