I've got portaling down for buildings, and have the knowledge that too many portals will cause problems because the game will take longer on what to draw.

So, for this huge map I'm working on, I'm at the point of where outdoor portaling needs to be done to help FPS.

Does this looks like sufficient, or do I have the wrong idea? I mainly concentrated on the areas where there's lots of buildings and such and without going wild.

I remember finding another thread awhile back where someone did the same thing I'm doing now, posting a pic of their map with what they had in mind.

Any info appreciated!

(copy the image address and paste in a new window to see a bigger version. The red outlines are a lot easier to see)

just looking at what you have - you are still going to end up with one huge cell that encompasses your entire map

not knowing where your fps issues are - i would assume that you may still see spike/dive issues when popping from one cell to another, and some cells as outlined may be entirely ineffective...

Would it help more if you properly joined all the portals shown going through the very center of your image possibly with the shapes shown and just leave the ones in the upper right and left and lower left alone without portals?

Basically just use portals through the very middle section of the map from left to right as that is where most of the action and potential fps issues will occur. The sections more along the top of the picture and bottom of the picture dont look like they will be that much involved in the map as compared to the center so leave them alone.

I could be way wrong on this but perhaps someone much better than me as portals can run with it from here.

LPRay, thanks for the reply, that's exactly why I left some areas unboxed intentionally.

--

And Stryder,

I boxed the more "crowded" areas because FPS goes down when looking towards that area from another part of the map. That's why I didn't box everything up or connect the cells together.

So what you're saying, I should cut up the whole map into individual cells kind of like an unequally divided piece of graph paper?

yes - that is more effective than the first pic.

if you have fps 'wells' (drop areas) then double check that items are detailed where possible. try and reduce brush counts and or complexity

see if the fps drops are fx or model related by removing excessive or unneccessary models in the area

portalling will only fix so much.

Keep in mind your portals are only as good as the positioning of the structural brushes they touch. It's hard to tell by those screens, but if you're trying to do a "grid" portal system without worldbrushes or other structural brushes, the portals may not fail during compile but they will serve no purpose in-game because they won't be blocking any tris.

Sweet Stryder! Thanks!! Yes, I tried to make as many detailed brushes as possible. All bldgs are made minimal structural brushes and the portals for the bldgs compile just fine.

Now onto what I hope is the last question, which is replying to cornrow's mention. I plan on making extending the portals from top to the bottom of the skybox since the skybox is structural (outlined in blue).



Is that ideal? Or should I divide it into 3 sections with the middle section having individual cells?



photoshopped, not MSPaint'd

Oh yeah, and the FPS drops aren't really all that bad at all. I'm just trying to optimize my map.

Dividing your map into open cells like that won't really do anything for you. It's the structural brushes (not the portal brushes themselves) that block tris, the portal brushes just tell the game where the portal boundaries are.

Think about this: if you make a large square portal from the bottom of the skybox to the top, it might not fail in the compiler, but it will not block any tris from view, hence not changing visible tris in any way. However, if these 4 portal brushes are between buildings, when you stand outside of the square, in front of one of those buildings so only the building is in sight, nothing behind that building will be rendered, hence reducing visible tris.

But before you go through all of this trouble, does your map really need it? CoD 2 LOD's are set very low so tris often aren't the problem as much as FX are. The only maps I've seen with performance problems in CoD 2 are those with tons of FX or a huge amount of tree models in a small area.

Given that this map is so huge that our "LR Gun" mod won't reach the other side of the map, I think it needs some portaling.

It does have a couple areas of concentrated trees, but it's not really that "busy" I guess you could say. I'm not using any FX at all either.

I've always thought that big maps would need portaling, but now I'm wondering when and if it is really necessary in my case?

I'll post an updated version when I can. Might be a few days or so though.