melinda rackham wrote:
> 
> someone told me today that when making a multiuser vrml world its
> best to keep it below 5,000 polygons, 10,000 at the max..
> is this true?
> melinda rackham
> 

 Don't count polys. There are other things that make much more
difference, but what those other things are depends on so many
details that you'll go crazy trying to figure them all out.

 Instead, set a performance budget. Pick some miniumum framerate
for the absolutely lowest-end hardware you want to support. Then
go try things out on that hardware.

 If you know who the audience is going to be, then get ahold
of whatever HW they have. Otherwise, you're going to have to
choose. A middle-of-the-road non-HW-3D Pentium 200 box might
not be a bad low-end, but even that's going to mean some 
large number of lusers fall off the low-end. 

 If you own a low end computer, you're set. If not, and
you just can't get ahold of one, try this: turn off HW 3D
acceleration, try out your world, then divide the resulting
framerate by how many times faster than a Pentium 200 your
computer is. For example: if you've got an 800Mhz intel chip,
and your low-end target is a 200Mhz intel chip, divide your
framerate by 4 to get a (wildly inaccurate but generally
correct ballpark) estimate.

 So you blow your budget and have half the framerate you
need. Don't touch your polys yet! Turn off every last visual
effect you can

 The killer for non-HW-3D accelerated machines is not polygons,
it's textures and dynamic lighting. That's why the "5000 polygon"
rules are completely useless: all polygons are not created equal
when you're using software rendering.

 As a last resort, of course, you can start chopping polygons,
but unless you really loaded them on in the first place, then
other things make much more difference.

 One last note: some dork with a 30,000 poly texturemapped
avatar with it's own orbiting dynamic light can ruin your
whole day. So provide a good set of very simple non-texture
mapped non-dynamically-lit avatars for visitors to use.



-cks

Reply via email to