OGRE rendering performance

16 views
Skip to first unread message

doeko....@gmail.com

unread,
Mar 14, 2009, 7:30:42 AM3/14/09
to realXtend
Hiya,

I ran into a bit of a complication with the performance of the OGRE 3d
engine here.

The scene I am using is basically made up of:

-5 1024x1024 textures (trees and shrubs)
Instanced in 300 tree objects across the scene
No complicated geometry. The landscape mesh they are placed on is
about 400 polygons.

FPS in OGRE-mode: 7
FPS in SL mode: 60

System:
Intel Core2Quad Q8200
Nvidia GeForce 9800GTX
8gb RAM

I don't even want to imagine how this runs on lower-end systems. Is
there any way to fix this?

(My made up solution, create all the trees in a 3d mesh and make sure
the textures are applied correctly and so forth, that way you only
have one texture that is tiled across the entire mesh of trees, but
that should not be necessary...)

doeko....@gmail.com

unread,
Mar 15, 2009, 7:59:24 AM3/15/09
to realXtend
update:

I think the rendering process of primitive-type objects is REALLY
inefficient somehow. I got the following:
http://www.marlinstudios.com/products/treefarm/treefarm.html

These trees look A LOT better (seriously they look awesome) and should
be far more intensive on the graphics driver of any computer. Here's
the kicker though: I put about 160 10k-ish poly trees right next to
each other and they are running at 50 fps. huh?

I've had a similar problem once with my demo project (you can see the
movie on the homepage at www.visibuild3d.com). I originally built out
the area with prims (the houses) and got about 15 fps out of it (on my
old laptop) in rex mode and 30-ish on SL-mode. I then proceeded to
model all the houses in 3ds max and the framerate shot way up...

So is rendering primitives just inefficient or what is the deal here?
I think there should be a huge warning sign on that before anyone
tries to undertake any serious project with realxtend that uses a lot
of primitives...

On Mar 14, 12:30 pm, "doeko.cass...@gmail.com"

Lc

unread,
Mar 16, 2009, 2:00:18 AM3/16/09
to real...@googlegroups.com
Hi

Could you tell us how many trees did you set, the RAM, how many objects in your scene ?

Sacha

doeko....@gmail.com

unread,
Mar 16, 2009, 7:10:03 AM3/16/09
to realXtend
Hi Sacha,

Between 200-400 resulting in a total of 400-800 prims. I've replaced
the 2d trees with better looking 3d trees which perform better as
well, so I couldn't answer the question for RAM. I don't really
understand why this is relevant though, as the realXtend viewer
should be able to handle numbers like these just fine, like the SL
client can.

Regards,
Jules

On Mar 16, 7:00 am, Lc <lcc1...@gmail.com> wrote:
> Hi
>
> Could you tell us how many trees did you set, the RAM, how many objects in
> your scene ?
>
> Sacha
>
> On Sun, Mar 15, 2009 at 12:59 PM, doeko.cass...@gmail.com <
>
> doeko.cass...@gmail.com> wrote:
>
> > update:
>
> > I think the rendering process of primitive-type objects is REALLY
> > inefficient somehow. I got the following:
> >http://www.marlinstudios.com/products/treefarm/treefarm.html
>
> > These trees look A LOT better (seriously they look awesome) and should
> > be far more intensive on the graphics driver of any computer. Here's
> > the kicker though: I put about 160 10k-ish poly trees right next to
> > each other and they are running at 50 fps. huh?
>
> > I've had a similar problem once with my demo project (you can see the
> > movie on the homepage atwww.visibuild3d.com). I originally built out

Lc

unread,
Mar 16, 2009, 7:19:32 AM3/16/09
to real...@googlegroups.com
The problem could be serverside. If your server is low on Ram, it could produce some off results.

sm

doeko....@gmail.com

unread,
Mar 16, 2009, 2:46:57 PM3/16/09
to realXtend
Framerate is purely client-side and the server has 8gb of ram so that
is entirely out of the question.

On Mar 16, 12:19 pm, Lc <lcc1...@gmail.com> wrote:
> The problem could be serverside. If your server is low on Ram, it could
> produce some off results.
>
> sm
>
> On Mon, Mar 16, 2009 at 12:10 PM, doeko.cass...@gmail.com <
Reply all
Reply to author
Forward
0 new messages