OpenArena Message Boards

OpenArena Contributions => Maps => Topic started by: wing0 on March 03, 2015, 01:10:29 pm



Title: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on March 03, 2015, 01:10:29 pm
Hello OA!

So, I made this map and want bots to run around in it.
Therefore I tried to generate an AAS-File, which tells the bots where to run around.

I ran
./bspc -bsp2aas /home/tobias/.q3a/baseq3/maps/wing-sago66.bsp

Which came with GtkRadiant-1.6.4-Linux-x86_64-20131213

The end of the output is

Quote
...
  6650surface 2141 bogus normal vector -0.000000 -0.000000 -nan
t1 = 0.000000 -295147905179352825856.000000 0.000000, t2 = -295147905179352825856.000000 0.000000 0.000000
p0 = 175536873044764000256.000000 -165288685802077814784.000000 -4684887654314016768.000000
p1 = -119611032134588825600.000000 -165288685802077814784.000000 -4684887654314016768.000000
p2 = 175536873044764000256.000000 129859219377275011072.000000 -4684887654314016768.000000
p3 = -119611032134588825600.000000 129859219377275011072.000000 -4684887654314016768.000000
  6651 brush sides
6651 brush sides textured out of 6651
Entity 0, Brush 476: duplicate plane
Entity 0, Brush 478: duplicate plane
nummapbrushsides = 12584
   292 curve brushes
1 threads
-------- Brush CSG ---------
  2368 original brushes
  1909 output brushes
-------- Brush BSP ---------
  1909 brushes
 13441 visible faces
     0 nonvisible faces
 17276 total sides
pthread multi-threading
     1 threads max
depth first bsp building
 66442ERROR: Tried parent
Closed log bspc.log

And the end of the log file is
Quote
...
split not on both sides
split not on both sides
Tried parent

Any ideas?


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: grey matter on March 03, 2015, 03:26:17 pm
Quote from: https://icculus.org/projects/gtkradiant/documentation/Compile_Manual/bspc.txt
Error messages
--------------

Level designers should not worry too much about the following messages and/or warnings. The things reported are non fatal and won't cause any major problems. Some of the messages are just debug left overs.

[..]
"split not on both sides"

Regarding the "Tried parent" (https://github.com/bnoordhuis/bspc/blob/6c11357e6d79a89e88cda2fe0e67c99a8923e116/brushbsp.c#L934) error I have no idea.


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: fromhell on March 03, 2015, 03:56:09 pm
I would try the annoying trial-and-error process of putting a big botclip brush over the whole map and keep shrinking until you found the problem area.


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on March 03, 2015, 05:21:21 pm
Thanks for trying to solve the problem!

I have put the whole map within a 'botclip' brush and still get the same Error-  and no aas.
Do I have to remove the botclip-brushes I already got in the map first?

 :'(


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: Gig on March 04, 2015, 02:18:59 am
What about trying with a solid brush instead of a botclip brush, to make the map "smaller"? (I never tried that)
Note: if you wish to be able to actually play the "small" map, you should let a real area with at least one spawn point...


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: fromhell on March 04, 2015, 03:04:47 am
I don't suppose you've already tried the 'urt' versions of the compiling tools have you? Sometimes they're the only ones that work for me. I don't know why either


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: andrewj on March 04, 2015, 04:28:31 am
Looking at the code to bspc, this error is basically a programmer assertion (and not something inherently wrong with the map).

The error means that, during the process of creating the BSP tree, a node splitting plane was chosen which is co-planar with a plane previously chosen to split the BSP.  That normally cannot happen, since the earlier splitting plane should have sub-divided some section of the map into two halves (either side of the plane) and hence that plane should never be chosen again (since it would have absolutely nothing on one of the sides -- and splitting planes always need something on each side to be valid).

So you have been unlucky I think, something slightly odd about your map is triggering a bug in the bspc code.


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on March 06, 2015, 04:46:20 am
@Gig: I include major parts of the map into huge solid brushes, however the problem persisted.
@fromhell: Urban Terror? What is the program called and where do I get it?
@andrewj: I guess the check is to prevent infinite recursion. I do not want to dig too deep into code. Without having seen the code, I guess that within recursion, already used parts are just marked and ..... I dont want to fix code that is running for everyone besides me. I doubt that this adventure is going to have a good time/usefullnes ratio. Bah, I ma just desperate.


I will now try to use a variation of Gigs idea and "substract" parts of the map with huge brushes ... annoying :/


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: Gig on March 06, 2015, 06:58:20 am
When mapping, it's a good idea to:
- Keep multiple backup copies of old versions of the map... in case one needs, one may rollback to a previous version, or may copy-paste portions of the map from a version to another...
- Try to compile .aas file by now and then (e.g. once a day at least) even when you are not yet focusing on bot optimization, to avoid the risk to get a fully 100% perfect map, only to discover .aas does not compile. So, if you know that the edits you did "today" broken that up, you have an idea of what may be the cause and try modifying the things you made "today" (or recover from the backup of "yesterday").

Unfortunalety, id tech 3 mapping tools are not perfect, and we have to deal with that in some way...


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on March 06, 2015, 07:38:49 am
@gig: the working title of the map is wing-sago68 ... 68 because it is the 68th backup. I have an aas for "version" 34, but the map is quiet different.



I would be glad to upload the bsp to get some of your input (everyone's input - critique!).
I will upload a bsp soon!


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on March 07, 2015, 06:02:07 am
So, I found the bogus normal vector.
Surface 2141 roughly belongs to brush 358 , as 2141/6 = 356.
Note the offset of two, which I guess comes from two func brushes (doors).
After deleting that brush, the program still tried a parent split but the bogus vector was gone.

However with a modification of Gig's idea (I deleted brushes with huge brushes) I found out that a triangle brush in the skybox was  causing the problem and voila - I got the aas...


HOWEVER ...


Now the aas file which I just generated was "out of date" ?!?

Can I just md5 patch the aas or does anyone have a clue?


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: andrewj on March 07, 2015, 07:13:11 am
You re-compiled the bsp too, right?

I don't understand what you mean by "md5 patch".

Normally just have aas file with same base filename as bsp file and put in same place, like:

maps/mymap.bsp
maps/mymap.aas


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: Neon_Knight on March 07, 2015, 08:09:25 am
If you made modifications to any brush in the map file or added new ones, regardless of their function, then you have to recompile both the map and the aas. :/


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on March 07, 2015, 08:55:33 am
@andrewj: As you went a little bit more into detail with some tech stuff, so did I here.
To check if a bsp and an aas belong together, a so called 'hash' is being checked.
The 'hash' can be seen as a summary. the hash that ioq3 uses is "md5".
My idea was to just "patch" (modify) the aas file to work with that bsp.
I came around techniques where people compile aas files for modified versions of their maps to 'optimize' bot-paths.
I dont understand too much of this topic myself ... we will see!!

@Neon_Knight: Hey, nice to see you! If I compile them in a pipeline, it does not even work for me :/
It just says 'aas not up to date'.




Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: andrewj on March 07, 2015, 11:28:29 pm
To check if a bsp and an aas belong together, a so called 'hash' is being checked.
The 'hash' can be seen as a summary. the hash that ioq3 uses is "md5".
My idea was to just "patch" (modify) the aas file to work with that bsp.
OK I understand what you mean.

But it is not a good idea.  Compile your map to bsp with q3map2, then compile aas with bspc, place both files together and it should just work.

Maybe the engine is finding an older aas file -- as a test you can rename your bsp and aas files to something odd (e.g. call them kkk123.bsp and kkk123.aas) and see if it works then.


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on March 10, 2015, 11:52:15 am
@andrewj: Thanks for the ongoing help and your reply, but I already tried that plus, I use version-numbers, so after some changes the file has a new number anyway (at the moment wing-sago73).


My search does go on .....  ;D :-\


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: grey matter on March 11, 2015, 12:51:14 pm
You can use q3map2 -fixaas <mapname> to overwrite the checksum in the .aas to match the given .bsp. Note that this is pretty much a hack you should only be using if you didn't change geometry, e.g. when you modified worldspawn only.
Another use would be adding better .aas for e.g. untouched id Quake 3 maps using additional bot clip in a decompiled .map or such.


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on March 13, 2015, 08:52:09 am
That was what I was looking for!
It changed four digits only and I got no errors anymore ... but still it does nothing :(

Someone from #ioquake3@freenet told me to use another version of gtkradian ... I will try that soon.


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on February 18, 2016, 04:48:24 am
I am able to generate an aas file from the bsp, but when loading the map it says:
"aas file is out of date"

Any ideas how to fix that?


Title: Re: AAS Problems (no bots in map) "tried parent"
Post by: wing0 on February 18, 2016, 05:06:44 am
As suggested by Pan I downloaded this version:

https://s3.amazonaws.com/gtkradiant/GtkRadiant-1.6.5-20151224.zip

It worked ! <3