PDA

View Full Version : Game environment mesh?


Gaboon
07-25-2005, 04:10 AM
I'm trying to do a game environment for my portfolio. I'm doing a castle and surrounding area. I'm wondering if, like characters, the castle has to be one solid mesh. Cuz if it does, I'm using a lot of polys to make it so. Anyone know?

capone_adam
07-25-2005, 08:39 AM
I was taught to believe everything should be single mesh. Its good to learn...but no, in the games industry now most engines support intersecting geometry.

Heres a great link...

http://www.cgchat.com/forum/showthread.php?s=&threadid=18021&highlight=intersecting+geometry

gasmask
07-25-2005, 04:22 PM
gave enviroments are probably never single meshes if they are going to be used as levels, u will find you will run into tons of problems if you try to do that

baaah888
07-25-2005, 04:47 PM
the main thing you definatly have to keep solid is any area a player can jump to or onto.

for example, a floor that meets a wall, if theres a gap the player models bounding box could get caught on it, or even worse the player could fall into the evil darkness ala mario kart 64 if you broke the game and got up the sides :)

Basically if the player can touch it using normal methods it should be solid and "crack" free. if its purely environmental and astehtic, i.e the roof of a building you can never get onto. then it doesnt need to be welded in.

Remember theres a difference between props and environement. so you don't need to weld book cases and such into the floor as they would be classed as props even if they never move.


Im more a character artist so this is just what ive picked up from research.

Hourences
07-25-2005, 05:15 PM
Its easier to make the castle in 1 big mesh to start with but later on when youre almost done you might be better off splitting it. Depends on polycount, size etc. Bad for optimization else

Youre also better off making the castle for an actual game so you can see it real time in game than just Max only. Thats also what a lot of companies preffer to see.

And the amount of stuff you weld depends on loads of stuff for which you first have to know what kind of engine you have. For example if your engine uses vertex light you cant just stick everything in to eachother without welding or making extra cuts

Again, first choose engine and game, and then make something for it so you got all the info on what to do and what is best for that specific engine

Gaboon
07-25-2005, 05:42 PM
Thanks guys for the many replies. I think I know where to go with it now :)

Matt
07-25-2005, 07:25 PM
the main thing you definatly have to keep solid is any area a player can jump to or onto.

for example, a floor that meets a wall, if theres a gap the player models bounding box could get caught on it, or even worse the player could fall into the evil darkness ala mario kart 64 if you broke the game and got up the sides

That's not true at all. Every game engine I've heard of supports invisible clipping planes. If the engine runs on an advanced physics engine, it would be absolute insanity for hit detection and player movement to be based on a high density mesh... What this means is, if a player is running on top of a sphere, there is a lower CPU/net load if there is an invisible clipping model of a lower polygonal density encapsulating it.

Answering the original poster's question:

For building an environment you only need to respect the specifications of the engine, and the texture space. What this means is, if the engine can only handle 200 models in a level, then you shouldn't model every tiny detail as different meshes. Most engines now support more than thousands of models and hundreds of thousands of polygons visible. Whatever you do, regardless of the engine's capabilities, try to model smart. If you can accomplish something at a low poly level and it looks as good as it does at a high poly level, then use the low.

If you think it's low enough, think again. I've seen many games that use three levels of polygonal detail when only one was necessary. For example, a stop sign in a poorly made game:

LOD1: 6 triangles, 512x512 texture
LOD2: 2 triangles, 512x512 texture with alpha transparency
LOD3: 2 triangles, 256x256 texture with alpha transparency

Unless the player is going to be an inch away from the stop sign, there's no reason for it to use such a large texture space. A lot of texture artists try to keep textures 'scale' to one another, but that isn't always the best approach. For example, they may want to make a grenade texture about the same size as the texture for a hand. That works because the hand and grenade will both be touching and if one was really sharp while the other was blurry from mipmapping it would be an obvious distraction... But on the other hand, if you were texturing a cashew (I know, odd.) and the hand was done on a 256x256, you wouldn't necessarily have to keep the cashew at the scale 32x32, you could probably get it done with a 2x2 flat color. That was a bad example, but I hope it helped a little.

Texture space is important to the environment. Make sure you're friendly to the idea of tiling textures because it's not really logical in modern engines to break up a model into small parts and create unique textures for every part of it. It would be prettier, but the tiling is better unfortunately.

CGTalk Moderation
07-25-2005, 07:25 PM
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.