Re: Height maps, etc.

Jim Terhorst ([email protected])
Tue, 9 May 1995 15:42:26 -0600


On May 9, 4:44pm, Michael Sweet wrote:
> Um, guys, isn't a 200x200 grid going to decompose into a little under 80000
> triangles? You might be missing your target audience just a *little* bit
> with a scene graph that big. :)
The target audience isn't limited to the home user who wants to become
Gandalf in some Virtual Reality elf world. The internet grew in popularity
initially because it was a great way for people to get work done. I know
thats hard to imagine now, but thats really what the internet started out as:
a tool to communicate with. Now, the home user may not have a machine that
can move a 200 by 200 elevation grid in real time, but the environmental
engineer who wants to show his/her cleanup strategy for Rocky Flats to his/her
boss in Washington D.C. may have the horsepower to interact with that complex a
scene.

>
> I'm expecting the average user to own a 486 or Pentium machine with a 2D
> accelerator board of some kind and not a high-end machine (be it Windows NT
PC
> with a 3D accelerator or UNIX workstation).
This won't be true forever. I think its a bad idea to limit VRML based
on the technology that the average user has available to them today. The
level of 3D performance on the average machine will only get better. We
better think about it now. Evans and Sutherland sells a PC card today that
does texture mapping in real time for $2400, so it will probably be here
sooner than most of us could imagine.

Also, VRML can help accelerate this process. If people find interactive 3D
compelling, they will be more compelled to outfit thier machine to handle it.
On the other hand, if 3D seems slow and burdensome, then people will just
wonder what all the hype is about, and download some more flat, boring,
non-interactive, 2D flatlander stuff.
>
> The ElevationGrid idea is a *good* one, but grids of 100x100 are probably all
> your basic PC will handle easily. One nice thing about grids is that you can
> do automatic LODs by sub-sampling the grid.
"will handle easily" today, but what about tomorrow.

jim terhorst (MountainTop::Computing) [email protected]