Splitting up long objects

#1

We all know a game’s camera will only bother rendering & calculating what it can see. So I am wondering, say I have a river from one end of my level to the other. The whole thing is animated. Should I split it up into sections? For example, if my character is at one end… won’t the engine have to still simulate the uv animations and the rest of it on the entire object all the way down, as long as part of it is visible to the camera? Or will most engines automaticaly divide it up into sections to handle this automaticaly, to save on resources?

#2

Many engines do split up the scene into what is currently visible vs. what is not. Most will split the polygons for you.

Most FPS games work on the BSP or portal principle… there’s a twisty cooridor at the start and end of a room, which helps the engine throw those polys away when you leave, and the cooridor makes sure you won’t see them pop off.

There were a couple great articles in Game Developer magazine recently by Guillaume Provost (of Pseudo Interactive) about managing graphics performance. “Beautiful Yet Friendly” parts 1 and 2, June and July 2003 issues.

Well worth the search if you can find them. Good pictures and lots of essential info for the game artist who wants to learn about optimization issues like this.

Also, your title makes it sound like you may have long thin triangles… if you do these are generally frowned upon. But I have an inkling that you know about this already.

#3

> won’t the engine have to still simulate the uv animations and the rest of it on the entire object all the way down

if this is for a video game, i don’t see how animated textures on a single large poly representing a water surface would make a difference. animated textures are just a sequence of textures so whether you’re using index 1, 2, 3, or however many textures are part of your animation, it’s no different than if you use one texture.

so is the question you’re asking is it faster for OpenGL or DirectX to render a single large polygon or many smaller ones? Well with frustum culling done by OpenGL, that large polygon will eventually be cut anyways, making it smaller. I’m sure that if you’ve played Ghost Recon or Line of Sight: Vietnam with large outdoor areas with rivers, the far frustum plane used in those levels cuts off the trees, rivers, etc. So in my opinion it would be best to leave it as one.

But I’ve never seen a 2-triangle mile-long river in a game :p.

#4

Hopefully someone wrote a book on this issue :stuck_out_tongue: Seems most the time the only optimization stuff for 3d games I can find is tri-stripping issues.

#5

Here are the links to those two articles I mentioned.



You can get a GDMag subscription for free if you’re a developer. Won’t help you with past issues, but the Post Mortems are great each month. Lots of gems in them.

This describes a lot of the issues in the 3d pipeline…
ExtremeTech 3D Pipeline Tutorial

Here’s something I wrote awhile back, might help you…
An Artist’s Real-Time 3D Glossary
http://www.planetquake.com/polycount/resources/glossary2/index.shtml

#6

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.