Heightfield vs. mesh was RE: AC2 was RE: [MUD-Dev] Total Annilation of Downtime

brian hook brianhook at pyrogon.com
Sat Dec 14 14:29:14 CET 2002


(Fear my subject header changing kung fu!)

[Marc Fielding]

> I'm curious about the peculiarities you mention. Would one be the
> challenges in smoothly transitioning between heightmapped and mesh
> terrain?

The biggest problem is that you have two different fundamental data
representations for your 3D objects.  You have a heightfield and
then you have polygon soup.  The types of tools and operations you
perform on them can be very different.  Heightmaps have a lot of
nice attributes about them, but algorithms targeting that particular
data structure don't necessarily carry over to a generalized mesh.
The opposite approach sometimes work -- use a heightfield strictly
for optimized storage/retrieval, but convert it to a specialized
mesh representation in your tools and/or game.

So there's a lot of duplicated work, and merging the results at the
seams can be problematic.  For example, you would expect a mesh zone
to be hand textured, but a heightfield zone is almost always going
to have some kind of procedural texcoord generation, and where they
meat will be a texel mismatch.  Same with lighting, collision, etc.

It's far from impossible, but it's definitely not as straightforward
as saying "My world is a heightfield, and everything else are
objects".

Even a pseudo-compromise solution where you have a heightfield with
3D objects placed on it to give a feel of "real 3D" doesn't feel
right.  It looks like, well, a heightfield with a bunch of 3D
objects shoved into it.

Brian


_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev



More information about the mud-dev-archive mailing list