Counter-Strike 2

Counter-Strike 2

De_A.R.K.
⮞⮞ ReDrAgoN ⮜⮜  [developer] 25 Nov, 2013 @ 5:28am
Bugs and fps drops
If you find bugs\ nav errors and elements causing fps drop, please, let me know.
Remember - map is still in beta.
Known bugs:
Bots try to avoid stairs to B site on CT side
Bots can stuck on elevator.
Reflective envmask on tile texture in hall can lower fps, lowering shader settings from very high can solve this.
14 player\bots maximum, at least 1 above it can case entity limit crash!
< >
Showing 1-2 of 2 comments
I have tried the map. First things, when I saw it I was quite impressed by the images and the appearance of all this custom content. I appreciate all this effort for new content and it's nice to see something different than all the usual content being scattered around the workshop, which really is a pity IMO... Great work!

I liked the watefalls, the normal maps on the textures and especially the cubemaps reflections on the floors and walls (although some are really heavily exposing).

I'm not a professional at gameplay so I will be adressing all the graphical elements here.


First most important problem is optimization. Actually, everything I'm about to tell is going to be about that. The map runs, horrible. Sorry, I can't say different. The weird thing is that the map is in fact good optimized. I'm not sure if you implemented max and mindistances on props or if you used hints or skip, but those seem to be right.

Unfortunatly, your problem lies within a few deciding factors: Shaders, textures and props. First thing I noticed are the watefalls. I understand refraction is a nice shader, but know that this is one of the most complex shaders to render and having many of those across the map just kills framerate. You are better off making a plance as a model and have the refracted texture be put on that model, and then fade the model once it's out of distance, otherwise you're gonna have a big time trying to fix FPS.

Secondly, although physics props are nice (breakable physics props with gibs), you're having more than a hundred of those in your map... Valve almost has none in each of their maps for good reasons. They're expensive, they're bad for latency issues on server, and you're having hundreds,hundreds of them. Even the SDK docs mentions you should only have about no more than 'tens' of them, meaning around 10-20 should be the max. My suggestion is to replace EVERY glass railing as a static prop and your FPS will rise like never before. Nobody is going to notice that, also no one will be waisting ammo to break them. It's not a problem if that extra luxury (unnecessary) feature is there. Breakable props with custom gibs should only be used on the most important physical props, like windows that allow a passage to an important area (see militia).

Moreover, your map has an amazingly large area outside of the complex, which is by logical means bad for gameplay, but also bad for visleafs, as you're having loads of them which needs to be calculated for rendering visibility, making FPS drop if you don't group them with something like func_viscluster. Add func_viscluster where a large area with unbroken visibility to assign them all together for VVIS, it will also compile much faster then, and improve FPS.

Bumpmaps and reflections also take a toll on the FPS. Consider which textures are best used with bumpmaps. Don't bumpmap everything you have, just use it on textures with some depth in it.

With 5vs5 I get 40-50fps in general, this should be higher for standard CSGO maps. There is much you can do to improve it, with the above being the most important ones. If you didn't add max and min distances on props, please do so.

It's a long list, but as someone else I know always says, if I thought it hadn't any potential I wouldn't review it :)

Good luck with optimizing the map, and enjoy the mapping process!


⮞⮞ ReDrAgoN ⮜⮜  [developer] 29 Nov, 2013 @ 5:20am 
Wow, thanks JorisCeoen, its really help. Gonna soon make lite version with better optimization and new game modes.
< >
Showing 1-2 of 2 comments
Per page: 1530 50