It actually creates alot of extra triangles to do this in some cases, if the func_details are particularly complicated. I don't believe this is the best practice, because it can mess with the z-buffer, but I've had good luck with sinking func_details 1 or 2 units into the world brush as well. This is acceptable if the detail brushes would be noticeable if they were floating a unit up. I've also mixed in func_brushes and took the performance hit for that to gain less splits in the world brushes. Use mat_wireframe 1 to see where your brushes are getting split. You have to look close sometimes though, there's usually a lot of tiny triangle strips. Usual culprits are things like staircases and brush handrails. Anything with alot of repetitive geometry.
On Thu, Jun 17, 2010 at 2:23 PM, Mike Zimmermann <rem...@gmail.com> wrote: > I've been out of mapping for a while but if I remember correctly, > waterindices are created when func_details touch world geometry. (The name > comes from GoldSrc when they were used to smooth edges of water into world > geometry. In Source the same effect is applied to func_detail, func_lod, > etc.) > > I wouldn't worry about it too much but an easy solution is to raise > func_detail 1 unit off world geometry to prevent it from performing the > "smoothing". A clean solution is to create your func_details so they have > the least amount of area touching world geometry as possible. > > -Mike > > > On Thu, Jun 17, 2010 at 1:48 PM, Benjamin J Dunbar < > bdunb...@student.scad.edu> wrote: > > > yikes, thanks for the info. i was mainly concerned about the > > "waterindices" and I was thinking using func_lod would reduce the "very > > full" "waterindices" warning on the compile log. > > > > ----- Original Message ----- > > From: ics <i...@ics-base.net> > > Date: Thursday, June 17, 2010 12:35 pm > > Subject: Re: [hlmappers] Optimization Questions > > To: Discussion of Half-Life Mapping <hlmappers@list.valvesoftware.com> > > > > > Avoid func_lod, do not use a lot of those within a map. It > > > consumes much > > > of precious fps when rendered compared to func_detail. You might > > > not > > > notice the difference on high end machine but low end it shows > > > pretty > > > clearly. I have used func_lod on my own projects only once or > > > twice per map. > > > > > > As an example (not from l4d2 but from cs source), i encoutered a > > > map > > > with a friend that had used func_lod on brushes, instead > > > func_detail > > > which should have been used. So with permission from the owner, > > > we > > > updated that map and fine tuned it a bit. Fps became very steady > > > and > > > increased with 100+ from what it was. Func_lod is very expensive > > > to use, > > > if used a lot. > > > > > > -ics > > > > > > 17.6.2010 19:20, Benjamin J Dunbar kirjoitti: > > > > Hey man, that's an enormous help for me. Thank you. > > > > > > > > It looks like func_lod is worth trying out. > > > > > > > > Question about "waterindices": Does it say anything about > > > there being a limit or breaking point for them? I'm only > > > halfway to the brush limit but the "waterindices" is up to 85%, > > > do you think the engine will break if I ended up going much > > > higher? The map itself has great frame rates. > > > > > > > > The other wierd thing is that I normally get those "glimmering > > > edges and cracks" in my brushes no matter what I do. They > > > usually appear in dark areas and at distances greater than 500 > > > units. I have learned to deal with it and map in such a > > > way that gives very few long-distance views in darkness - ei > > > dark tunnels. > > > > > > > > Thanks again > > > > > > > > ----- Original Message ----- > > > > From: "Kenneth B. Swisher"<capt...@qis.net> > > > > Date: Thursday, June 17, 2010 10:44 am > > > > Subject: Re: [hlmappers] Optimization Questions > > > > To: Discussion of Half-Life > > > Mapping<hlmappers@list.valvesoftware.com>> > > > > > > > >> Hey Ben, > > > >> > > > >> Here's what VDC has (seein' as how you can't access it). > > > >> > > > >> > > > >> func_lod > > > >> > > > >> > > > >> From Valve > > > Developer Community > > > >> > > > >> Jump to: navigation > > > >> <http://developer.valvesoftware.com/wiki/Func_lod#column- > > > >> one>, search > > > >> <http://developer.valvesoftware.com/wiki/Func_lod#searchInput> > > > >> > > > >> This is a brush entity > > > >> <http://developer.valvesoftware.com/wiki/Brush_entity> > > > >> available in all > > > >> Source games. > > > >> > > > >> > > > >> Entity Description > > > >> > > > >> Brush-built > > > >> > > > <http://developer.valvesoftware.com/wiki/Entity#Brush_entities> model > > > >> that fades out over a specified distance. Useful for creating > > > >> world > > > >> detail that doesn't need to be drawn far away, for performance > > > >> reasons. > > > >> * Note:*The fading is not always consistent; brushes may > > > >> suddenly "pop" > > > >> into view depending on where the player is looking. This happens > > > >> because > > > >> of a hardware issues with rendering translucent materials. > > > Low- > > > >> end cards > > > >> will fade nearer than the distance specified, and smaller > > > >> objects fade > > > >> nearer than a larger object with the same fade distance. > > > >> * Note:*The overhead required by this entity is substantial. Use > > > >> it only > > > >> for special effects or when fading very dense geometry, and one > > > >> could > > > >> then argue, shouldn't that better be a model > > > >> <http://developer.valvesoftware.com/wiki/Model>? > > > >> > > > >> As for waterindices, here's as much as I could gather: > > > >> "t-junctions are edges with more than two collinear verts. This > > > >> happens > > > >> when dissimilar geometry meets sharing an edge, vbsp > > > >> automatically adds > > > >> all verts along the edge to each polygon sharing the edge to > > > >> avoid > > > >> cracks. vbsp has a -notjunc option that will skip this process, > > > >> but that > > > >> will leave you with cracks in your level. You probably want the > > > >> t-junctions fixed. Cracks usually look like sparkling edges > > > >> (bright > > > >> pixels in the skybox showing through). > > > >> > > > >> There are several references on t-junctions online as it's a > > > >> general > > > >> problem when processing geometry for any rendering engine. It's > > > >> not > > > >> unique to anything specifically in Source." > > > >> > > > >> Basically, it looks like it has to do with when you have two > > > >> brush edges > > > >> with vertices along them that don't line up, so vbsp does some > > > >> work to > > > >> prevent unsightly cracks from showing along the edges. > > > >> I'unno, that's > > > >> my best guess. > > > >> > > > >> And, thanks for your service (Army). > > > >> > > > >> > > > >> > > > >> Benjamin J Dunbar wrote: > > > >> > > > >>> Hi, > > > >>> > > > >>> I am asking my questions here on the mailing list because I do > > > >>> > > > >> not have access to the VDC at this time...it's considered by the > > > >> US Army to be a "gaming" site, so they have blocked it on our > > > >> machines.> > > > >> > > > >>> What exactly is func_lod, and do you think I should use it > > > >>> > > > >> instead of func_detail? I have a high amount of > > > >> func_detail in my map and after looking at Valve's maps I > > > >> noticed some func_lod where I would have assumed to use > > > >> func_detail. Can someone clear up it's purpose and usage > > > >> for me? > > > >> > > > >>> Also...same map as above...when I compile, I get 85% > > > >>> > > > >> "waterindices", which the compiler calls "VERY FULL". But > > > >> I have no water in my map at all. Can someone tell me > > > >> what's up with that? What are "waterindices"? > > > >> > > > >>> Thanks very much guys. > > > >>> _______________________________________________ > > > >>> To unsubscribe, edit your list preferences, or view the list > > > >>> > > > >> archives, please visit: > > > >> > > > >>> http://list.valvesoftware.com/mailman/listinfo/hlmappers > > > >>> > > > >>> > > > >> > > > >> > > > >> _______________________________________________ > > > >> To unsubscribe, edit your list preferences, or view the list > > > >> archives, please visit: > > > >> http://list.valvesoftware.com/mailman/listinfo/hlmappers > > > >> > > > >> > > > >> > > > > _______________________________________________ > > > > To unsubscribe, edit your list preferences, or view the list > > > archives, please visit: > > > > http://list.valvesoftware.com/mailman/listinfo/hlmappers > > > > > > > > > > > > > > > > > _______________________________________________ > > > To unsubscribe, edit your list preferences, or view the list > > > archives, please visit: > > > http://list.valvesoftware.com/mailman/listinfo/hlmappers > > > > > > > > _______________________________________________ > > To unsubscribe, edit your list preferences, or view the list archives, > > please visit: > > http://list.valvesoftware.com/mailman/listinfo/hlmappers > > > > > _______________________________________________ > To unsubscribe, edit your list preferences, or view the list archives, > please visit: > http://list.valvesoftware.com/mailman/listinfo/hlmappers > > _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlmappers
