Compile taking days

I need help with Hammer. But first, some back story!
A while ago, I was having trouble with compiling my map. It would take about 5 days to compile. Not very good for testing. So I figured out that if I set VVIS to run on fast mode, it would only take about 5 minutes. But in-game the map would crash a lot, mostly when looking at the whole map in no-clip.
A couple months later, and I feel that I am done with the map. So I want to stop the crashes by doing one last compile running VVIS on normal. It’s been over a week now and the last two lines of the compile are:


BasePortalVis:     0...1...2...3...4...5...6...7...8...9...10 (338)
PortalFlow:        0...1...2...3.

It has looked like that for 6 days. (That’s part of the VVIS process.) If it were to continue in this pattern, it will be several CENTURIES before it’s done.
And yes I know that complicated geometry causes long compile times, but my map really requires complicated geometry.

So what I’m wondering is, how could I speed up this compile or reduce vis cluster crashes or whatever?

optimize your map

OPTIMIZE, my friend. apply the nodraw texture to faces the player should never see, make detail brushes func_detail, don’t have an excessivly huge skybox.

This has basically no effect, but if you insist on doing it, do it to faces the player cannot ever see.

Oooh. Can I do it?

It’s been established that the map needs to be optimised, but posting an in-hammer screenshot of it would certainly help us tell you exactly what needs to be modified.

func_detail anything that is ‘detail’. Don’t func_detail it if it is touching the void.

Also add areaportals and hints and shit.

I’m surprised you actually wasted days compiling this.

Not enter able buildings though. Otherwise you would render everything outside the building.

What?

It has a huge effect on FPS.

No its not. VBSP culls backfaces automatically if they are touching another brush face or touching the void.

Then what would be the point of the tool?

Afaik roofs you will never see over and that kind of stuff.

VRAD will not generate lightmaps for nodraw-textured faces or bounce lights off of them, so using nodraw on unseen faces (roofs, etc.) will speed up VRAD and reduce your file size.

Dude, Make sure you dont use tons of blocks for one thing that can be less blocked.

Make theese nodraws…

Make areaportals

And"func_detail" In area stuff, Make sure it wont collide with the skybox.

Also, You can compile it with a “fast” vis for the start, Noclip on you’re entire map and look for useless places that can be nodrawd, Like under the floor textures that you wont see as a regular player.
Try to find the views that gives you less fps which makes you lag more…
That will direct you more.
I’m not good at optimizing myself but i hope it’ll work for you better than mine.

Who told you that you can’t touch it?

That’s what a second computer is for. :stuck_out_tongue:

That’s my problem. The map is a spaceship I made for a comic, and just about every where in it is enter-able.

Well, again, it’s in space. It needs a large skybox for other ships to be floating there and stuff. How do spacebuild maps work? Would that help here?

I don’t think a single screenshot would be enough information. I can barely tell what I’m looking at while I’m editing… there’s just too much.

I can try making a lot more brushes func_detail, does anyone have suggestions? Railings, chairs, wings, etc.?
Would making the skybox much smaller and adding a 3D skybox help?

Thanks for all the help, guys!

it helps with FPS optimization, for slower computers.

[editline]11:36PM[/editline]

yes, make all the railings, chairs, and other details “func_detail”.
you basically want to make everything that dosn’t really obscure your view very much “func_detail” (smallish pillars, doorframes, etc)
although having a 3d skybox adds a little to the compile time, a 3d skybox would be a good idea for adding background planets, etc.

also, to speed up compiling:

us func_visclusters for large, open areas. even in semi large rooms.

(its not actual optimization, i know)

But I would be shrinking the regular skybox quite a bit, so would it be worth it?