One Object of not?

unscodst

I have a very dumb questions considering that I've been modeling on and off for a few years now. Recently I was playing video games and took a look around and the objects as kind of a study. It occurred to me that I have been complicating my modeling based on how I thought / learned how to model.

For video games, are objects with multiple pieces all attached or are they all build as one object? For example:




Assume this is a telephone pole with an electrical box. The box on the left pole is cut to match the pole, then all the vertices are welded. This makes the pole and box one object. This is how I learned to model, I don't exactly remember why and I probably started learning around 2005.

The box and pole on the right are two separate objects. Technically they could be attached to be one object but the box is just "floating" on the pole. This is what I have been seeing recently in videos games which I thought I learned was the wrong way to do things.

ZacD

Depends on the object, game engine, if someone could tell, how modular you want the project to be. Intersecting is fine but can cause artifacts, there's also some game engines were if a model is made up of separate parts, it will count as extra draw calls, which can be really bad for performance.

axxic3

When I first started modeling I had a similar thought. ZacD said it perfect, however that might confuse me a little on just starting into modeling. You dont want to model like legos, as I did with gmax. Build as much as possible in one mesh. These videos helped me get an understanding.

http://vimeo.com/10941211

tynew

ZacD what kind of artifacts do you mean? Are you talking about subobj edge aliasing when zooming out in engine?

ZacD

z-fighting, lightmap seams, vertex painting or effects might look worse, to name a few.

Logithx

I use the method you showed on the right a lot, especially on simple props. It doesn't create any problems for me in-game (UDK or UE4) unless you have the lightmap resolution set really low. An exception to using the 2nd method might be where you have a really large flat surface with a small object (like a cylinder) sticking right in the middle of it. If you don't merge the objects properly in this scenario then it's possible for the lightmap to look strange as the flat face goes 'under' the cylinder. The part that's under the cylinder becomes black, this may leak onto the lit part of the flat surface which kind of looks like terrible AO that's not supposed to be there. Again, this could be solved by using a higher lightmap resolution.

Using the first option may give you some extra UV space if you cut out a big hole where you can put smaller UV islands into.

Anyway this is just my personal experience from UE3/4, feel free to explain to me why I'm wrong.

unscodst

Hey everyone thanks for the replies. Glad to see I'm not the only one who may be confused. axxic3, thanks for the video, i'll be watching it later.

As far as everything else goes that's exactly what I was thinking. Lighting would obviously be a problem depending on how it was build. Assuming I build everything like the example on the right, everything floating in place, would it be best if I attach them before exporting to they "make" one model, or can I keep them as separate objects?

Building everything as one object is definitely helpful as you can keep everything as on UV and same memory. The hardest part is working with rounded objects due to having to cut other objects properly to incorporate all the edges. But if you have the time in patience I am sure it would make a much better model than one where everything is separate.