Shadowspawn on 6/7/2005 at 14:23
I need to get what TIM file format info I've figured out into the Wiki, but I'm just swamped with RL issues and other projects. I REALLY need to finish the TIM file stuff.
Why not email me your .TIM file, I'll paw through it and see if I can find anything funky? I'm not sure what I can do to help at this point, but another set of eyes never hurts.
rujuro on 6/7/2005 at 16:51
I could do that, but I'm actually more interested in what you could figure out about that mesh that I mentioned in my last post: STMstairsspiralC-rail2.tim .
If we could figure out what's going on there, then I think I'd know how to proceed, since that is the behavior I need to mimic, and the TIM converter finds no Collision mesh. I'll send you my tim file as well when I get home tonight.
Shadowspawn on 6/7/2005 at 17:15
I can't check that file out until the weekend, I'm traveling for work and my laptop doesn't have T3 on it. (It can't even run T2). But I can view the .TIM file stuff.
Send me your file, and I'll check the other one this weekend.
rujuro on 7/7/2005 at 00:18
better yet, I'll send you both! They aren't very big.
Shadowspawn on 7/7/2005 at 18:37
OK, weird stuff;
1.) Thank you, you pointed out a bug in TIMtoE. I was reading a field wrong, which made me lose the rest of the file, which is why there was no collision object for the STMstairsspiralC-rail2.TIM when it was converted to 3DS.
2.) The collision model for that object is just a post at the top, nothing else. I'm not sure how you can avoid jumping off the stairs, maybe the navmesh keeps you from getting off. I don't know yet.
3.) Your collision model for the spiral stairs looks really good. But it might be too complex for T3. I haven't seen any complex models for the collision object yet, so maybe it has a limit somewhere.
4.) I'll send you the re-constructed 3DS model for the rail so you can look at it.
5.) I'll update and post my TIMtoE program this weekend, and incorporate the fix into the full TIM <> 3DS program - as soon as I get time to get back to it.
rujuro on 7/7/2005 at 19:18
Hmm, that is interesting, it's not nav-mesh, or any invisible BSP, because I tested that rail on my level, and the collisions behaved perfectly. It's very strange. Another object I checked had just a single convex step object for an entire flight of stairs, which makes me wonder:
Is is possible there are multiple collision objects, all convex, stored within a single TIM? Is there any way you might be dumping the first collision mesh, but not the rest? Just an idea, might explain the single step, and the single post...
Oh, and thanks for all the help!
Shadowspawn on 7/7/2005 at 19:56
Quote:
Is is possible there are multiple collision objects, all convex, stored within a single TIM? Is there any way you might be dumping the first collision mesh, but not the rest? Just an idea, might explain the single step, and the single post...
Possible, I guess. I'll do some more checking...
Crispy on 8/7/2005 at 03:20
Navmesh is just an AI pathfinding thing. It doesn't influence collisions at all.
rujuro on 31/7/2005 at 23:09
WOO HOO! Breakthrough.
For artist generated collision to work, it must remain convex. BUT, you can have multiple meshes making up the collision of a single static mesh. The hierarchy is put together as mesntioned earlier, you just make all the collision objects children of the collision dummy elements (Collision and High Detail). The pivot points of the collsion objects must also perfectly match that of the static mesh. So, the method I used is as follows:
1. Take your mesh and run the collision maker script.
2. Delete the mesh that is created.
3. Build an assemblage of convex meshes that approximate the shape of the original object.
4. Make their pivot points have the same postition and rotation as that of the static mesh.
5.Make them children of the High Detail object that you can find in the scene object list.
The resulting hierarchy should look like this:
Static Mesh
----Collision
------High Detail
--------Object1
--------Object2
--------etc...
Shadowspawn on 1/8/2005 at 00:02
Could you send me that object? I'd like to see how the TIM file represents the collision mesh!
Thanks