func_wall usage for mappers

I'm suggesting 2 add a rule for mappers 2 use original "worldspawn" insted of "func_wall" as more as possible. Why? :

1. Shadows don't give lags - thows all shadows is the mapmodel itself (plus map will look better itself).
2. Clients will have less lags, cuz all objects what aren't "worldspawn" aren't sended 2 client (not like other players) (plus, cuz it all will be only 1 model - clients will have better fps).
3. Servers will have better cpu, cuz they aren't sending all thows objects to all clients each frame (yes, cuz "worldspawn" cannot be changed under any condition - it isn't sended to client - client loads it from .bsp hemself).

P.S. Use "func_wall" only for invinsible objects to avoid shadows.

wtf? for invisible walls it is better to use the clip texture, it doesn't fps it doesn't create a entity, it will only be add to the bsptree. And also it doesn't create shadows.

Model != not Worldbrush
Entity != Model
Model != Entity
Worldbrush != Model

The Worldbrush it self isn't a Model its only the bsptree.
So i wouldn't recommend mappers that they shouldn't use func_walls for the stages or for the jumpstones, because there will be some issues.

#1 Issue: It will increase the compile time for mappers(mostly for beginners because of the bad brushwork)
#2 Issue: Without func_walls maps with big rooms will get lowfps because they are now worldbrushes. (It isn't recommend for older machines. Most of the community users have older machines where they get lowfps on maps)

Extranote:
func_walls & func_illusionary can make shadows too
and also it func_walls & func_illusionary are good when the mappers are using them wisely (max 3-5 objects in a func_wall & func_illusionary)

Now explain this: http://nuuk.nl/a/func_wall.jpg

Left pilar: func_wall
Right pilar: not tied

Originally posted by dEMolite

In this part you are write.

Originally posted by NumB

Here you go a plugin, to see how much fps it wold save: source (*.sma) & compiled (*.amxx).

Command: say /nowalls

Belive me, you'll be amaized how much fps it will save if your pc is slow.

// edit:

That picture btw will only take more cpu while compliting map. In game it has absolutely no effect. Or does - map models "*#" wont be precached and analized, so API wont work that hard, while "wordlspawn" is the main model traceline scanner what cannot change and requres as much cpu as I made 255longjumps (NONE!).

Worldbrushes do affect the performance ingame. Just check a map with gl_wireframe and check the leaf threads that are created by worldbrushes. Those raise the polycount and will reduce the fps. There is a reason why every mapper is using func_walls cause those wont create those leaf threads.
And not everything is a model ingame, like you said, cause models in hl1 are only build out of e polies, triangular, and world brushes are w polies. When you take a look at r_speed you can see the amount of both. If everything would be a model you would only see e polies.
Maybe we force the engine to analyze more enitities with func_walls, but its better than having too much polies. The task a mappers has is to find the right balance between models, brushes and enitities ;)

Bla^ we aren't talking over the cpu usage.
We are talking about the hlengine with the fps issues.

Even if it takes more power 2 load the map - map will use less resurcess while playing! I wold rather connet 5 minutes then play with 14fps or 1000+ ping! Althow it is imp anyway (i'm obout 5 min connection w/o no downloads).

lol NumB & SchlumPF aren't right. If you think without func_wall it takes less cpu makes a stresstest over a week. But without func_wall you will get lowerfps than with func_walls!
And if you don't believe us mappers take the carebear and learn how hlengine is drawing the leafs and how the compiler tools are creating the leafs during the vis compilestage

I know, that you mappers are right in mapping, but omg. WorldSpawn will take less cpu than thet func_wall. Making worldspawn insted of func_wall you are just making wordspawn model a bit more bigger, plus you are making ping lower and server+client side fps higher. Yes it takes mby it takes more time for valve hammer 2 complite the map, but omg... Wold you like if all plugins of the world were uber lagy and wold give uber big ping, only cuz scriptors are 2 lazy 2 and want to compile source as fast as possible?

Download that my plugin and enable your anti-virus scan, so your pc will be lagy. Then check how much fps /nowalls will give you. Please, mappers. Even if someway, somehow it all dosn't work by models (and i'm 100% sure its wrong), it will take only max 5 more seconds 2 connect as an exchange for perfect fps and lower ping.

I cannot understand what you dont wanna do it. It's only the source of the map. And already proved, that all good stuf have hard sources.

// edit:

At the end of each server frame there is a scan. It starts from 1 and gows to hl-entity limit. Not much that that scan takes much cpu (it's mby 1380*player_number loop), but if entity was found it forcess the dllfunc to add that entity 2 full pack for specific player (and that dllfunc uses not a little power). Bounch of all thows entities are func_walls what you created. Althow that scan dosnt check wordspawn, cuz it's unchangable object and is loaded only once for each player and server! Not like other objects - they are loaded each frame and scanned each traceline and other map objects scans, when worldspawn is the main object and evrything is scanned by it.

// edit2:

Can anyone make 2 test maps? Absolutely the same, only in one - the way you are already doing it all. And another with as less func_walls as possible?