[FINAL] desctf1 - Overgrowth

[Despair]

Updated Dec. 29 - FINAL!!

Q3 version: http://www.filedropper.com/desctf1
QL version (offline only): http://www.filedropper.com/overgrowth - Updated Jan 4, 2011
-to use in quakelive: place in ql folder with other pk3's, set file to read only, load a practice ctf game on an official map, then /map overgrowth in console.

**************

Hello everyone.

So I've been lurking here for a few years now, while learning to map. However, I've never managed to stay motivated enough to produce anything I deemed worth showing (slight perfectionist ). I've finally managed to stick with a project to an almost finished state, so figured I would go ahead and post it.

Here's a few screenshots.
[lvlshot]http://img14.imageshack.us/img14/8426/shot0007rj.jpg[/lvlshot]

[lvlshot]http://img43.imageshack.us/img43/2547/shot0003sc.jpg[/lvlshot]

[lvlshot]http://img571.imageshack.us/img571/7682/shot0000r.jpg[/lvlshot]

It's a ctf map with a somewhat different layout than you'd usually see in a ctf map. It was all done 'freehand' , so to speak, (no prior planning), so it's a bit of a maze at first. heh . The hope is to have it accepted into quakelive, hence the obvious ad spots. It is mostly done, as I mentioned before, but I feel it could still use some feedback, so any is welcome and appreciated. I don't anticipate making any significant geometry changes, but anything regarding item/weapon placement is quite open. Bots are supported, but do not play it very well atm. I am still working on that aspect, but am still pretty new at bot optimization. Anyway, enough rambling . I hope you guys like it.

[VolumetricSteve]

You're well on your way. The layout and sfx/eye candy integration were very professional. This is one of the best attempts I've seen at giving quake 3 a very...open world, realistic feel. I got some broken textures but I think it's my own damaged version of quake 3, your pk3 should be fine, I'll run more tests later on. For now, this all looks very good and I can't really suggest anything I'd change. I'm sure that's nails on a chalk board to someone looking for feedback, but you have a very strong map here.

If you want to work on the bot navigation a little bit more, there was a great thread I was shown ....this one:

http://www.simonoc.com/pages/design/map ... m_bots.htm

It takes some doing, I have yet to really try that method, but I know one of my future maps will require it and I've been told a few times that dealing with bot paths that way makes it all work MUCH better.

Good luck and keep up the good work.

[praetorian]

Well, I don't know if it's just me, but I am missing a huge amount of textures in your map!

[fKd]

not just you, but from what im seeing its looking very promising! fix the pk3 and give us a real look. a good way to do this easy is to have a secondary q3 install which is clean. throw ya pk3 into it before posting and see if anything is missing.

[Despair]

haha.. oops.. i did go through and manually check that each one was present, but turns out i accidentally copied over a couple .psd's instead of .jpg's. Glad you guys like what you see so far though. The pak is fixed now, original link is updated.

@VolumetricSteve: Professional? Wow, wasn't expecting to hear that! Thanks alot! Yeah, I had to use Sock's method to get bots in. BSPC.exe basically said "haha... no" when i threw my full map at it. I just have to learn the ins and outs of controlling the bots' behaviour better. Given the layout of the map, however, I'm skeptical 'good' bot play is even possible. Can hope though .

[VolumetricSteve]

Oh yeah, aside from the texture thing, you're all set, that map looks like something I'd expect to see in the Team Arena expansion pack.

If I can get this to work with bots:

http://sirventolin.webs.com/razztazzmagoria.htm

If I can recall correctly, without vis or lightmaps, the bsp compile of razztazzmagoria took up somewhere between 22 to 27MB. Bspc was very, very unhappy with me. When I first tested razztazzmagoria with bots, they would run for about 300 yards, find a corner, and just sit...they wouldn't look around, they'd just sit, and never move again, you couldn't even get their attention in-game.

Almost all of the work I do with quake 3 is to see how far I can push the technology until it breaks, then I find work-arounds, then I push it further. As long as you're not on a campaign to drive bspc insane, I have high confidence you can make your map work with bots well.

[v1l3]

It visually transcends any other ctf map that has been put out imo ... I'm far beyond impressed. It's too bad that you'll probably never be able to get the bots to work on it because there are so many routes to take. I was getting lost myself. They either run through a loop from one place to another, or stand still. You would have to do some botclipping to get them to reach the flag ... if that's called botclipping? I can't remember. It's a kickass map though.

[Anthem]

Yeah, it looks very nice. The layout is a bit complex. It would take me a while to get used to it. It's an interesting change, though.

[Despair]

Thanks guys. yea, the bot play is progressing in some spots... but regressing in others . it's definately going to be a project to get working well. I know some areas will never get bot traffic tho, such as the rail area by medkit. That area is just inaccessible they way bots move so. They CAN make the jump to flag though. They did so quite often in the earlier builds. I just have to figure out what causes them to have that behaviour over just dropping down (which they seem to prefer to do now ). The bot map is already pretty much as optimized as possible with bot clip and such, as I pretty much had to do so in order to get bot support to function at all. I just have to find the right balance and placement of item_botroams and clusterportals I think. Which is proving tricky.

[VolumetricSteve]

If I were you, I'd make the most simple bot map I could, even if it ruins their ability to reach everything, even if it isolates them. Then, with each recompile of bspc, I would add another hallway/room, the most simple way you can, literally, hollow out a cube, and leave the complexity at that, then when you've crept your way through out the whole map, you can kinda re-add surface detail into the bot map. I looked over your map again and while it is rich in path way detail, I think you should be able to get it to "bspc" to the degree of functionality you're looking for.

[Eraser]

Can't really comment on gameplay as I only ran through it for a bit. All I can say is that it looks gorgeous. A+ for that!

[Despair]

Yeah, I'm currently using a completely seperate bot map, as Sock did for PoM, stripped of everything. No lights, no textures, no details, etc. Nothing but the caulk hull, botclip, and neccessary entities. So bspc does run just fine, it's just the results that aren't up to par just yet. I'm going to go through the bot map again today and see if i can't find any more areas i can simplify, or even remove, and play around with item_botroam and clusterportal placements some more.

[VolumetricSteve]

...really? that doesn't....I....*runs through your map again*

[Despair]

heh yea.. i have a feeling i've just put too many/too strong of weight of item_botroams in the map and it's confusing the bots. i have already found more to simplify in the bot map, so i'm going to try a compile w/o the botroams and see how it does this time.

[fKd]

also you need to take into account the bots preferences to items. each have their own settings. ahhh q3 ctf bots. heh do the best you can. they are fully retarded in the ctf game type unfortunately.

[VolumetricSteve]

I tried to run back through your map, but something is really wrong with...all 3 of my quake 3 installs, I can't run your map in windows or os x. Otherwise I'd have helpful suggestions on how to make the bots respond better. It sounds like you have it well in hand though, so good luck, from what I vaguely recall it'll be a hefty project, but should be entirely possible.

also, bots in ctf are pretty retarded, it's true.

[Despair]

Heh yea, they are retarded indeed. Basically all i'm after is that the bots actually go for/can get the flag, and don't just follow the same path over and over. I don't think anything beyond that is realisticly possible.

[fKd]

remember, bots cant jump, you will need a bot clip brush as a ramp to the flag

[VolumetricSteve]

bots....absolutely jump, they have to jump in my maps and they do it all the time, what are you talking about?
also, bspc console output:

2151 walk
0 crouch
11 barrier jump
96 jump
0 ladder
626 walk off ledge
119 swim
0 water jump
0 teleport
0 elevator
0 rocket jump
0 bfg jump
0 grapple hook
0 double jump
0 ramp jump
0 strafe jump
7 jump pad
0 func bob

[fKd]

if you have a item over a gap with no other way to get to it, they cant get to it. its a fact. its another issue with the q3 bots. been doing this longer than you. i know what im talkin about

[Despair]

I know there is no way to explicitly tell them to jump a gap. however, as i mentioned before, in the earlier iterations of the .aas, the bots most certainly did jump the gap to the flag and retrieved it. Now i just have to figure out how to reproduce that behaviour.

[VolumetricSteve]

if you have a item over a gap with no other way to get to it, they cant get to it. its a fact. its another issue with the q3 bots. been doing this longer than you. i know what im talkin about

I was just responding to the outrageous claim that "bots cant jump" which they clearly can, as they jump all the time. What they can't do is have their AAS paths linked to things that aren't built into the tree of reachable items which is established by objects being in contact with the ground. If the object were isolated on a podium within reach, they could rocket *jump* to it. Also, you don't know how long I've been doing this (since 2004).

@Despair

Having dealt with the code of bspc for a while now, the reason you got the bots to jump(and the same reason I did in my maps) gaps of a path was because the compiler sees an object, and essentially tries to draw a line to it and if the compiler sees a jumpable gap, it'll tell the bot to jump when it reaches the gap when trying to reach that object. If the object is suspended, it's never picked up by the bot compiler as something the bots can reach - I don't know why Mr.Illusive didn't code object reachability differently, but.....it could be rewritten, you really only need to have the compiler look for floating objects, and determine if they could be reached by a short vertical jump, at which point the aas file would tell the bot to jump straight up when it reaches a certain area...but that's all theory, It might actually not be that hard to code in....now I'm curious, I'll look into it.

In other words, you can make bots jump gaps as long as the object they're hungry for is sitting on the ground(term used loosely) somewhere. They won't jump for it if the object is suspended because they won't know it's there. If you're a tricky mapper, (look at q3dm17) you can put jump pads in places, and make a suspended object be "in the way" of the jump path, and they'll blindly smack into it mid-jump, creating the illusion they're jumping for that item.

[Despair]

Well, I seem to be making good progress on the bots. They now navigate fairly well and consistently make the jump to flag. They do tend to take the same routes into base, and out with flag, but I think that's pretty typical for bots in ctf. They can now access all areas, though for most beyond their normal route, only if chasing an enemy. Again, pretty typical. My only concern now is, with 7 bots, it can cause quite a bit of stuttering (from the game doing bot calcs I assume). For the most part, it's not too bad, but if more than a couple bots are in view, or in a larger area, it can get fairly choppy. I have a pretty powerful cpu (i7 930) so, I'm worried how bad this will be for people with weaker cpus. Anyone know ways of increasing performance in this regard? I've tried adding more clusterportals to create smaller clusters, but that didn't help much (if at all).

[VolumetricSteve]

Like everything else the Quake 3 engine does, the AAS is pre-compiled, so I don't know why there's stuttering on an i7.....it just doesn't make sense. Can you post the output of your bspc process console? That'd help me help you a lot.

[Despair]

It's not making much sense to me either. The .aas tells the game what areas are accessible from where, but doesn't choose which to goto or when, right? I'm guessing it's struggling to pick which areas to travel to, but why, i don't know. The problem is reduced/eliminated with fewer bots so. Here is my last bspc.log, http://pastebin.com/Rd3h62v8