Making a Map with true 3d Models

How much lag would I get by making a map made entirely with 3d models using the standard map to keep from "nocliping"

Basicly picture a hollow box model with the map built around it for my concept just more fancy. XD

I can convert the dff files from san andreas into obj files and then in turn into md3 files. thinking about making a map that uses a bunch of dff chunks as a test.

Prepare to be disappointed if you try this kind of things.



There isn't any Doom engine to my knowledge that supports using models for map geometry. Any model in the map will be an actor.

Every actor is in a sector. Even if they're outside the map geometry. The entire space is partitioned into a horizontal 2D binary tree, where each point is either to the left or the right of any of the partition line. (Remember: BSP means "binary space partition".) That means that any point in the map will actually belong to a subsector that belongs to a sector, even if it is outside sector boundaries. The engine has no actual concept of "outside sector boundaries", it simply does not understand its data this way.

So every point is in a subsector, every subsector belongs to a sector, every actor is positioned as a point, therefore every actor is in a sector. No exception.

Here is where all that boring technical stuff becomes relevant: the Doom engine only renders actors if the sector they are in is visible.

What this means is that if you have a huge 3D model that you use as a "box" in which the level is entirely contained, if the player is noclipping away from the playable area, the player will also likely turn his back to the sector that'll be roughly in the center and in which the actor with the huge 3D box model is logically situated.

Which means that the huge box will not be rendered at all.

A slight correction; whether or not a 3D model is drawn depends on the renderer logic of the sourceport. The logic you describe above only really applies to GZDoom. Doomsday for example uses the model geometry to determine whether its visible, not merely the center of the actor to which it is "attached".

Doomsday also supports replacing the sky with any arbitrary 3D model, so that is another option.

That said, you'll probably encounter so many other issues that you'll have to kludge around that its just not worth the hassle. It might be viable for some limited use though.YMMV.

Maybe you out to map for Vavoom, but that engine has an extremely small audience.

Vavoom is pretty much dead; its forums have had only three posts made in 2014, the last was in April. It's kinda depressing actually.

Gez said:

Vavoom is pretty much dead; its forums have had only three posts made in 2014, the last was in April. It's kinda depressing actually.

That's a shame, I have not visited their forums in a while. Hopefully the inactivity is only temporary.

Why not just try for Doom 3 instead?

It'd be a clever thing to do if any living source port was built in a way that allowed for it elegantly, though!

Is there actually a port that cares about the model geometry for collision detection?