Pros & Cons - DTS Buildings

#1 Pros: More Geometric detail than can be had with typical brush based geometry used in DIFs.
(Think of a "brush" like a child's wooden block or "lego" that the most simple form is a triangular prism or a rectangular prism-- aka block)

Cons:Collision Detection has greater precision and takes less processor time than with DIFs. Plus you don't get the ability of DIFs to determine geometry that doesn't need to be displayed by use of Portals. (portals display only geometry that can be viewed through the portal, kinda like looking through a door, window into a building and only being able to see what is inside the portal. Geometry that is behind other geometry, say objects behind a wall, won't be displayed since they are obscured)

DIFs are specifically designed for building interiors and buildings because doing so with DTS shapes (which are basically just 3D models, like something saved as a 3DS model outta Maya or 3D Studio MAX) would bog down torque to an unplayable crawl because of testing for collision. (if you're doing something like a FPS generally)

Basically collision works (as I understand it) with DIFs as a test between a 3d point in space and a 3d plane in 3d space (the brush).

With DTS shapes, you have the actual model, and then you have the collision model (forget the usual name for it) basically an invisible or non rendered set of polygons used for colliding with the 3d shape inside the game. You can have only a certain number of polys in this "collision shape" since you have to test the player and other objects against each poly of the collision shape. If there are alot of polys in the collision shape, then that means alot of tests the program has to do to see if the player is colliding with just that one DTS model. If there are several DTS models that are colliding with the player, then that means there are that many more tests per DTS model that the computer has to process compared to DIFs which would have far fewer tests, and be therefore be computationally faster.

If you're doing a game where collision isn't much of an issue, or you don't need collision w/ buildings other than real simple collision say w/ an RTS, you could use DTS.
But every sort of 3rd person, or FPS uses some sort of brush or portal based geometry for the majority of their levels with a combination of 3d models for extra detail. TGE may be able to include DTS shapes into DIFs when you're mapping them in Quark in future versions to get more detail than just plain DIFs, like what you see in UT2k4 and others. Otherwise you have to place DTS models inside the mission when you load your DIF into torque.

I'm sure that there are better explainations than mine and if anyone wants to correct my explaination feel free.

#2 Happy birthday to this question!!

#3 So i guess the collision is the issue...

Genrally ... when people model in .. say maya (since it what i'm using) the duplicate the object for the collision mesh.

if i were to model one of the houses in the fps (one with fire in) i could get away with a colision mesh of around 140 polygons

Now the same house is created with in and around 34 od brush blocks if i understand ir corectly .. each block would have 5/6 side it needs for object colision therfore 170 - 204 faces needed for colision testing.

Now wouldnt my colision mesh be way more efficient than the brush one ?? or am i understanding it wrong :-/

(xcuse spel)

#4 Hmm .. could you explain what you mean by "but you wont be able to go through interior level design in DTS files"

:-p

#5 The collision meshes need to be Closed Convex Volumes. Just cloning the mesh won't give you that (unless your mesh is a pyramid, cube, or sphere).