Normal map and smoothing group problems

[I_R_Hopo]

BIG EDIT: Okay, just ignore me.... I went and searched for an hour after posting this and found an identical probelm posted here: http://www.polycount.com/forum/showthread.php?t=71536&highlight=normal+map+problem. Turns out I was doing things right, and I just needed to alter a couple things (Like I figured. Me making a nooby mistake :P) My main problem is now fixed, so just to make this thread NOT useless, I'll ask another question.

Do the amount of smoothing groups make a big difference in your model? For example, I have a weapon I'm making with a lot of hard edges, and about 4-5 smoothing groups per big piece. Would it be better for me to chamfer areas to try and get different parts into a single smoothing group, or would the extra polys be bad? I've read that seperate smoothing groups = more verts, but I don't know how game engines work and what is and isn't important to watch for. Also, say I have a big object on one smoothing group with no errors, but I've split it into two UV islands. Do I put both islands into seperate groups, or will they work properly in one group with no errors between islands?

Thanks guys :) Sorry for making the thread then figuring it out. It seems I always spend hours with a problem, finally work up the courage to post about it, and then I figure it out, haha.

[Mr. Bean]

Would it be better for me to chamfer areas to try and get different parts into a single smoothing group, or would the extra polys be bad?

For game modeling it's all about efficiency and budget. You have to think about what will make it look good but still have the best possible polycount. I wouldn't chamfer all those edges unless it was the high res model for generating a normal map.

You're worrying a lot about smoothing groups; I wouldn't worry too much, instead you might want to put it in a game engine and see how it looks. I don't think having a lot of smoothing groups will hurt anything, but then again it might depend on the engine.

[teaandcigarettes]

tricount can be misleading. Setting each poly to a separate smoothing group in fact doubles the vert count along their borders.

http://www.polycount.com/forum/showpost.php?p=762436&postcount=15

I think it would be nice to see the model first. However, in case of a weapon (I assume it is a first-person-view model) I wouldn't worry too much about the tri-count. After all, how many of these will be rendered on the screen at the same time? Even if enemies use the same weapons as the player, many games have separate 3rd and 1st person models. I would go with whatever looks good.

Then again, I know jackshit about weapons :p

[I_R_Hopo]

Thanks guys. That link made stuff a lot clearer for me :)