I am thinking about adopting Unity 2.5 for a multi-player baseball game.
I was reading that the Unity character models needed to be one mesh. Our models are already skinned and have 5 different meshes. From a poly point of view they are 5000 poly which is OK.
Is the multiple meshes a no no for decent game play in unity ? If so that means we have to recombine the meshes and re skin which can be a lot of work I think.
Heres a picture of our catcher for instance:
It may be a lot of work, but it’ll be worth it.
Five meshes wouldn’t be too bad for, ay, one central character. But you’ve got to draw two baseball teams. That’s 18 players * 5 meshes = 90 skinned meshes to draw - which is basically asking for framerate trouble.
I would probably tone down the poly’s too, unless you’re aiming for really high-end machines. Again, for a single main character 5000 polys is doable. But do the math - 18 * 5000 = 90,000 polys, which is a lot to ask of a graphics card, especially if they’re all skinned.
And all this isn’t even including the diamond, stadium, crowd, bat, ball, umpire(s), bases, or GUI.
So if you want your game to run on anything that costs less that $3000, you’ll rebuild that model. :-/
I am not sure what is your meaning of “polygon” (polygon = 2 triangles?) but current graphics card can process over 1 mtriangles/s.
Also, you can use LOD.
Minus, you must consider the shader, textures, lights and so on any other fx that can impact in the performance.
But i think a 5’000 vertex model is fine if not considered as low poly.
Thanks for the replies ! The manual actually notes that 5000 polys per character is OK. Since we are not targeting machines with integrated video this seemed OK. We are going for a realistic game. The issue seems to come from the fact our char models have multiple meshes and the following paragraph from the manual.
"Your character should use only a single Skinned Mesh Renderer! There is usually no reason to use multiple meshes for a character. Unity also has optimizations related to visibility culling and bounding volume updating which only kick in if you use one animation component and one skinned mesh renderer in conjunction. If you care about performance, multiple skinned meshes per character is NOT an option. If you use 2 skinned mesh renderers for one character instead of 1, the time spent on rendering the character will most likely double!"
It seems, base on this, we need to combine our meshes and therefore re skin. I was just looking for confirmation.
Interesting.
I’m wondering. If you’d have character models that need to be able to wear a large variety and combination of clothes, combining meshes is not a good option.
What would be decent triangle-counts and texture sizes for characters with an additional 5 meshes for clothes? Let’s say you have about 10-25 characters on-screen at the same time here.
Well we are redoing ALL of our models. They will all be 1 mesh with 1 texture, except for a few detachable items. In those cases it will be at the most 2 mesh and 2 more texture.
Unity is going to be great but its performance is likely to be lower overall then say Torque. So we are planning on doing everything we can to optimize for the lowest common denominator.
Regarding your question on other meshes the way we are planning on doing it is switching out the entire model. That way we can still have 1 mesh and 1 texture. Its a bit more work but the performance payoffs I hope are going to be worth it.