IPB

Welcome Guest ( Log In | Register )

Circular heritage error
Outlawz
post Nov 25 2016, 11:40 AM
Post #1


Captain
***

Group: Soldiers
Posts: 58
Joined: 19-October 07
Member No.: 3,951



QUOTE
Debug assertion failed!

Version: 1.5.3153-802.0 Build date:
Module: Cluster
File: C:\dice\Projects\BF2Branches\Patch_1_50\Code\BF2\AI\AIDLL\AI\AIPathFinding\Cluster\ClusterNodeManager.cpp
Line: 38

Text: Circular heritage for 8023

Current confile:


I haven't had this in years but last I remember, it was supposed to mean the navmesh is too big (in tricount). Is it really that or can it be something else?

I Googled and found some threads but wanted to see if anyone knows anything newer
http://www.battlefieldsingleplayer.com/for...?showtopic=7888
http://www.battlefieldsingleplayer.com/for...mp;#entry153219


--------------------
Go to the top of the page
 
+Quote Post
 
Start new topic
Replies (1 - 11)
clivewil
post Nov 26 2016, 04:45 AM
Post #2


meh.
Group Icon

Group: Mr. Admin
Posts: 2,008,214
Joined: 24-July 07
From: Perth, Western Australia
Member No.: 3,545



i've had that error many times before on different meshes - if it was only a size related thing, then i shouldn't have seen it on any of the normal sized meshes

to fix it i would usually look at the underside of the mesh and try to spot any polys facing down that shouldn't be (protip: NOTHING should be facing down) - if you have double-sided faces showing, turn them off as they are very unhelpful especially in tasks like this

i found it useful to scan around in the Bottom viewport and find a distance that showed up any bad polys while still being far away enough so that you could cover a reasonable area each time - then save that viewport setting so you could recall it each time you returned after making a repair.

i would typically find between 4 - 6 sections needing repair in any new navmesh being examined.

as you scan across the mesh, some faces will flicker in and out of your vision (which might depend on your gfx driver and other factors) - try not to be too concerned about these; the troublesome ones we are chasing usually do not flicker and remain plainly visible the whole time

good luck and have fun



example of Bottom max viewport (with Visible Edges active) showing one very obvious bad poly



--------------------
Go to the top of the page
 
+Quote Post
Dnamro
post Nov 26 2016, 07:44 AM
Post #3


CinC (for now)
Group Icon

Group: Mr. Admin
Posts: 6,808
Joined: 28-December 03
From: Fort Hood, TX
Member No.: 46



Did you look in your debug folder for 8023?


--------------------
No, my Avatar is not my dog.
Go to the top of the page
 
+Quote Post
Outlawz
post Nov 27 2016, 03:02 AM
Post #4


Captain
***

Group: Soldiers
Posts: 58
Joined: 19-October 07
Member No.: 3,951



QUOTE (Dnamro @ Nov 26 2016, 02:44 PM) *
Did you look in your debug folder for 8023?


Yes, but it only went up to 7000

edit: clivewil's tip might do the trick


This post has been edited by Outlawz: Nov 27 2016, 03:08 AM


--------------------
Go to the top of the page
 
+Quote Post
Outlawz
post Nov 29 2016, 02:40 AM
Post #5


Captain
***

Group: Soldiers
Posts: 58
Joined: 19-October 07
Member No.: 3,951



I looked through the underside of the mesh and there's a lot of bad polys. Is there some more automated way I can find all of these? I tried Face selection with Ignore Backfacing but that selected good polys as well. Am I doing something wrong when navmesh setup/generation? I first thought it had to be one or two statics in the map because the bad polys kept appearing next to their holes but it seems they happen next to other statics too and I never had this problem before except once when navmeshing the exact same map I'm trying to navmesh now.

I also found faces that aren't inverted but intersect with another face


This post has been edited by Outlawz: Nov 29 2016, 02:50 AM


--------------------
Go to the top of the page
 
+Quote Post
clivewil
post Nov 29 2016, 07:22 PM
Post #6


meh.
Group Icon

Group: Mr. Admin
Posts: 2,008,214
Joined: 24-July 07
From: Perth, Western Australia
Member No.: 3,545



QUOTE (Outlawz @ Nov 29 2016, 04:40 PM) *
Is there some more automated way I can find all of these?

not that i know of, unless you are a maxscript guru in which case you might be able to write a tool specifically for the job

it looks to me in your pic like you have your max materials set at 100% opacity - maybe try them set to 50%, so any 'double layered' polys become more visible.

unfortunately i don't have an example of a folded poly nearby but you can see the effect i'm talking about on these multi-floored buildings, it's very obvious where the polys overlap each other.




--------------------
Go to the top of the page
 
+Quote Post
Outlawz
post Dec 5 2016, 09:10 AM
Post #7


Captain
***

Group: Soldiers
Posts: 58
Joined: 19-October 07
Member No.: 3,951



I looked over my past navmeshes and they all seem to have upside-down faces and errors described in post #5 yet I haven't had a circular heritage error on those.

I'm trying to track down the error now, the error message mentions cluster 8233, I can find this in infantry.cls.txt

CODE
-=| Cluster: 8233 |=-
[Elements: 22]
{
    15817
    160960
    98939
    138916
    70884
    31989
    3523
    168178
    94099
    146588
    158327
    152572
    83574
    175277
    170155
    168856
    147427
    148452
    6105
    50248
    44046
    156315
}


and I can find those numbers in infantry.qti.txt but nothing to actually help me find the position of the cluster or poly or whatever it is.


--------------------
Go to the top of the page
 
+Quote Post
Outlawz
post Dec 10 2016, 04:52 AM
Post #8


Captain
***

Group: Soldiers
Posts: 58
Joined: 19-October 07
Member No.: 3,951



After not fixing any of the inverted faces and such and instead shaving off around 10 000 tris, the circular heritage errors stopped. So it seems it was really only tricount size.


--------------------
Go to the top of the page
 
+Quote Post
clivewil
post Dec 10 2016, 09:43 PM
Post #9


meh.
Group Icon

Group: Mr. Admin
Posts: 2,008,214
Joined: 24-July 07
From: Perth, Western Australia
Member No.: 3,545



how many polys did you have before and after?


--------------------
Go to the top of the page
 
+Quote Post
Outlawz
post Dec 11 2016, 11:46 AM
Post #10


Captain
***

Group: Soldiers
Posts: 58
Joined: 19-October 07
Member No.: 3,951



176k before, 164k after, infantry mesh. Vehicle is 87k.


--------------------
Go to the top of the page
 
+Quote Post
clivewil
post Dec 12 2016, 05:24 AM
Post #11


meh.
Group Icon

Group: Mr. Admin
Posts: 2,008,214
Joined: 24-July 07
From: Perth, Western Australia
Member No.: 3,545



QUOTE (Outlawz @ Dec 12 2016, 01:46 AM) *
176k before, 164k after, infantry mesh. Vehicle is 87k.

176k? i thought i'd had higher counts than that, i might go check through some of my older meshes to confirm. i think Bloodymap 1024 was around the 195k mark

i remember when i re-did the nav for Fushe Pass for AIX that it kept on dying (not absolutely certain it was circular heritage errors, but 80% likely it was) and i ended up deleting all but one small section at a time until i isolated the general area that was troublesome, then i re-genned that little section and transplanted it into the main. as far as i could tell, there was nothing wrong in the 'bad' area, the game just didn't like it. maybe there was a thin triangle in there somewhere, those things often cause problems


--------------------
Go to the top of the page
 
+Quote Post
ltfred
post Jan 7 2017, 03:26 PM
Post #12


Private
**

Group: Soldiers
Posts: 30
Joined: 28-July 13
Member No.: 9,714



QUOTE (clivewil @ Dec 12 2016, 06:24 AM) *
QUOTE (Outlawz @ Dec 12 2016, 01:46 AM) *
176k before, 164k after, infantry mesh. Vehicle is 87k.

176k? i thought i'd had higher counts than that, i might go check through some of my older meshes to confirm. i think Bloodymap 1024 was around the 195k mark

i remember when i re-did the nav for Fushe Pass for AIX that it kept on dying (not absolutely certain it was circular heritage errors, but 80% likely it was) and i ended up deleting all but one small section at a time until i isolated the general area that was troublesome, then i re-genned that little section and transplanted it into the main. as far as i could tell, there was nothing wrong in the 'bad' area, the game just didn't like it. maybe there was a thin triangle in there somewhere, those things often cause problems


I always thought based on an old post of yours clive it was around a 100k poly limit for navmesh's since calculating quadtrees being too immense for the engine?

see here post #5 and #6 http://www.battlefieldsingleplayer.com/for...mp;#entry153219
I have struggled wiith this limit as well

This post has been edited by ltfred: Jan 7 2017, 03:30 PM
Go to the top of the page
 
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 25th April 2017 - 10:28 AM