Game Development Community

"Jigsaw" approach to structural meshes

by Norris Bonner · in Artist Corner · 01/25/2005 (2:09 pm) · 2 replies

I had posted this in the programming section but I believe its better placed here. Ok guys, I gotta bring this topic up again. I have searched everywhere and I found this:

Another area I'd be curious to get opinions about would be using CSG vs. a "jigsaw" approach to structural meshes like buildings which have been modeled to have no interior. I know games like Freedom Force give you destructable buildings that, because they partially collapse in a predictable, similar fashion, appear to be made up of a mesh cut into "jigsaw" pieces. This seems like an absolute pain on the content creation side, but when compared to the learning curve for CSG and suitability on mid-range machines, I wonder which might be the better approach.

The "jigsaw" approach would leave your environment more predictable, which could help in other areas like AI and game design, but it does leave a level looking very similar.


Anyone know ANYTHING about this jigsaw approach? Im clueless on this and I havent seen any resources on it. Think this will work alot better than animating.

#1
01/25/2005 (2:29 pm)
I personally think people overstate the complexity of CSG modeling. I find it much more complex to learn how to deal with a CSG exporter's problems (many of which are simply created by trying to fit a square CSG peg into a round polysoup hole) than I do with a CSG editor. But it has been discussed, but not using the term jigsaw. Mainly, it's just creating meshes and then using the simple collision. A difficulty comes in with accurate shadowing (taken care of with static meshes in the Lighting Pack) and texturing. The more meshes that you create (if they use separate textures) the higher the texture memory it will take, which can cause rather large problems. Plus, there's the zoning issue for large levels.

Personally, I'd create the basic shapes of the levels in something like QuArK. Then, use those shapes to create the actual level geometry. Zone stuff correctly in QuArK. Now, apply the null texture to the interiors. Place everything. You have a zoned DTS structure. Sure, it has an underlying structure that you made in QuArK, but anyone who has taken ten minutes to read the introduction to QuArK can create a simple structure without worrying about texturing and such. It's creating complex arty things that CSG makes difficult. The editors don't make it much easier, but the format itself is the bully most of the time. Creating ugly blocky things is easy in CSG, and more of a complaint since there's not poly-perfect control. But ugly blocky things are the only things that matter in making the basic structure of the level.

I'm not sure how zoning affects DTS models, though. But that's similar to the solutions I've seen when this idea has been discussed before. If you do a search for DTS levels, zoning, etc, I'm sure you'll find quite a few.
#2
01/25/2005 (3:45 pm)
Thank you for your response. This was alot of help. I really didnt know what the term was to even search for it. Thanks again