Beginner Question: Mesh Structure, one chunk, or split?
by Vernon Finch · in Artist Corner · 11/16/2005 (1:20 pm) · 3 replies
Intro: I am doing some beginning modelling to get some basic art in for my game. Getting the hang of the program wasn't too difficult, but what I lack is basic tip's for game model structures, what is bad practice, what is good practice etc.
I have been trying to keep my model all one mesh, with all the vertices linked as I'm an organised sort of guy like that, but it dawned on me that I may be modelling inefficiently, by breaking up polygons into unneccesary pieces just to keep all the vertices linked.
Question: Are there specific advantages / disadvantages to keeping a mesh all one chunk, at perhaps the expense of a greater polycount?
Example: I am modelling a Biplane, for simplicity sake imagine the two wings as two long stretched rectangles/cubes. I want to now add some small support beams between the two wings.
To accommodate these support beams I cut my wings quad into some smaller pieces, so I can extrude the slim support beams up/down to my other wing, which I have also cut up. Then I can weld the vertices, joining them all as one mesh.
Alternatively, I could just create a cube, stretch it out to make a beam, and just rest it between the wings, but not actually linking them, thus not having to split up my wing mesh. This would result in less polygons, but is it bad practice? Is it going to cause me troubles on export or ingame?
Any tips would be most welcome :)
I have been trying to keep my model all one mesh, with all the vertices linked as I'm an organised sort of guy like that, but it dawned on me that I may be modelling inefficiently, by breaking up polygons into unneccesary pieces just to keep all the vertices linked.
Question: Are there specific advantages / disadvantages to keeping a mesh all one chunk, at perhaps the expense of a greater polycount?
Example: I am modelling a Biplane, for simplicity sake imagine the two wings as two long stretched rectangles/cubes. I want to now add some small support beams between the two wings.
To accommodate these support beams I cut my wings quad into some smaller pieces, so I can extrude the slim support beams up/down to my other wing, which I have also cut up. Then I can weld the vertices, joining them all as one mesh.
Alternatively, I could just create a cube, stretch it out to make a beam, and just rest it between the wings, but not actually linking them, thus not having to split up my wing mesh. This would result in less polygons, but is it bad practice? Is it going to cause me troubles on export or ingame?
Any tips would be most welcome :)
#2
Does it matter if the meshes overlap? Or should I position the vertices of one mesh to rest exactly on the edge of a face of another mesh? (I'd prefer that simply for neatness anyway if possible).
I do have the warsparrow pack, and noticed the windows were a seperate mesh, but seemed to be lined up very neatly with the body mesh.
edit: typos
11/16/2005 (1:50 pm)
Thanks for the input Kirby. I'm all for optimising the mesh, so was curious as to the best approach.Does it matter if the meshes overlap? Or should I position the vertices of one mesh to rest exactly on the edge of a face of another mesh? (I'd prefer that simply for neatness anyway if possible).
I do have the warsparrow pack, and noticed the windows were a seperate mesh, but seemed to be lined up very neatly with the body mesh.
edit: typos
#3
11/16/2005 (2:27 pm)
Like Kirby said your mesh doesn't have to be a single object. In fact, you can often save ploys by segmenting the model rather than trying to make one oddball shape fit another. I've found it useful to join edges where they meet but only in the same segment. Not where edges from two different segements happen to fall next to each other. 
Torque Owner Kirby Webber
In the example of your biplane, I think you'd be better served by splitting the mesh up into seperate pices, i.e. not subdividing the wings so that the strust can share vertices.
Fact is, the only time it's going to be obvious that it's a seperate piece is in wireframe... in your modelling application. Once that sucker is skinned, animated and exported into Torque, it shouldn't ever really be apparent that it's not supposed to be that way.
Furthermore, you'd be wasting polygons needlessly IMO. breaking up a smooth surface to accomodate the incorporation of an item that doesn't have to share vertices/ faces isn't exactly "frugal" when it comes to resource useage at runtime... but then again, I'm all about the frames per second.
At any rate, I hope this gives you something to chew on. Again, Torqu won't balk at it either way, so it's really your personal preference.
~ Cheers.