Fixing Buggy Tracks
#1
Posted Jun 16 2009 - 09:36 AM
I was looking at this thread and was reminded about Indy 67 having issues. Does someone have the knowledge to fix it?
#2
Posted Jun 16 2009 - 11:43 AM
High Risk (To be avoided for points races)
Mallory Park - Easter Trophy (mallory)
Montjuic Parque (to be re-evaluated with new version)
Oulton Park V2
Oran Park
Paul Ricard 1988
Sepang
Wigram Airfield 1968 (65mod + 66 mod so far)
Zeltweg 1970
Medium Risk (To be tested in fun races)
Bathurst Mt. Panorama
Blue Mountain (65mod?)
Croft
Daytona RC (freeze is a car cuts in front of you at the chicane: clipping bug.)
Goodwood
Imola
Indy 67
Pic Long
Suzuka
Anyone else know of problematic tracks or have a better idea about the above tracks?
#3 Border Reiver - guest
Posted Jun 16 2009 - 04:05 PM
The "fixed" Montjuic is about as good as it can get. The mean time to failure is much longer than it was, and 80%+ cars finish now in full grid online races from the tests we ran. There are still occasional glitches, but these might be more GPL and instantaneous processor load related than the track itself. There is nothing wrong with the track in terms of mismatches now so it is about as good as it can be.
Zeltweg is in a similar camp to Mallory. I did look at it for a cln file and realised it needs a lot to sort out. The problem with most of these things is that without the source files it is a really hard job to rebuild and recompile tracks. We did it for the update to John Black's excellent Oulton Park track, but it took absolutely ages to reconstruct all of the build files needed to allow recompiling.
I have been meaning to look at Wigram again as it is one of mine. Personally I have never had a problem there, which you will appreciate makes it very hard for me to test if any changes I make to see if they help or not. Again it is in the Montjuic camp that there is nothing materially wrong with the 3do or the trk file that should cause lockups. I have notes from a recent league race showing where people had lockups that I will look into when I have time to see if there is a clear pattern. Something else I wonder about with some of these is if people run with various graphics options turned off. Depending on the object classes chosen certain odd things can happen at some tracks if not all options are enabled. It is a long time since I made it, but I vaguely remember some sort of oddness about Wigram to do with object classes for some of the outfield stuff that was used to mask some clipping. As I recall turning on all objects got rid of this, but as I say it has been a while....
Of the others, Goodwood is another one with a known mismatch. Indy I think has to do with sliding on the banking in the turns. I saw something similar with Brooklands that always when the car got to a certain drift angle in a certain place it would lockup. I think I fixed it for Brooklands by doing something to the fb and set files when compiling, but again I would have to go back to my notes. Either way, that is also probably a rebuild job.
The only other thing I can think of for some are if they are rasteriser problems. I know that some sizes or numbers of sub images can cause problems for some rasterisers. (I need to look up what the limitation is, but I know that Targa needed a fix to make it work in OGL by changing a texture). It could be that this is also coming into play on some of these other tracks.
For debugging and trend searching it is always worth knowing what track version anyone was using, what their rasteriser is, what graphics options they have turned on/off, what detail bias they have, what corner/location on track it happened at, what else was going on around them, i.e. did some cars just appear ahead, or did someone just have a high impact crash etc. All of these can help to find a trend and ultimately a solution if it is evident that there is only one problem area on track, or perhaps even only one problem texture or object.
Ultimately for most of these things it is a real pain though, because obviously the problems didn't show up during original beta testing or they would have already got fixed before release, by their nature these faults are intermittent which is very frustrating and as I said above, often the original source files are not available so it is a long hard slog to do anything about things if the track.3do is involved in the problem.
Rob
#4
Posted Jun 17 2009 - 02:55 AM
we also have a track blacklist at our GPLRACER league (in German & English):
https://forum.gplrace...opic.php?id=50
Cu
Stefan
Edited by Stefan Roess, Dec 27 2018 - 06:20 AM.
#5
Posted Jun 17 2009 - 03:23 AM
Thanks in advance.
Claudio
#6
Posted Aug 31 2009 - 04:46 AM
Bob Simpson, on Jun 16 2009, 12:43 PM, said:
I will add a few others from the GPFun blacklist. We run 90 minutes races with full grid, so I guess we have more chances to get into bad situations
Not to be used in official races (several CTDs over the years, every time we used them):
Salzburgring 74
Virginia (V.I.R)
Sandown
Risky and thus avoided (several CTDs over the years, yet not systematically):
Riverside
Glitches:
- Oulton 60's version (cars "jump to the side" for a couple of frames on the back straight; manageable)
- Laguna 67 (cars jump backward on the start/finish line; the issue was known and I don't know if it was fixed).
- Keimola (this is probably fine, just a bad event with several CTDs)
- Buenos Aires, BA15 (several CTD's, but all in all manageable and we use it)
- Road Atlanta (you can be thrown into the air when exiting the pits, manageable)
Bye,
Luca
#7
Posted Aug 31 2009 - 05:01 AM
Thanks for your list.
I remember the problem from Laguna67 with flipping cars on S/F-line every lap by 10 - 15 meters. But this problem is fixed. We had more then one league-races there without showing the old problem. But unlucky i don't know where the fix is linked. Maybe other drivers can give you a hint.
Greetings
Achim
#9
Posted Aug 31 2009 - 01:33 PM
And Stu; thanks for starting this thread. I think we should build a page in the GPL Wiki for this.
Fletch
#10
Posted Aug 31 2009 - 01:52 PM
It seems Ardmore is a bit critical near the turn 1 too, as seen offline.
Marlboro 1967 in replays and Speed Valley when spinning in turn 2 are very critical in terms of CTD, both even without a touch in any wall.
Edited by Otavio Silveira, Aug 31 2009 - 01:52 PM.
#11
Posted Sep 22 2009 - 05:57 AM
Got fixed!
Download here: https://pb21.autosimsport.net/
Edited by Stefan Roess, Nov 14 2017 - 05:07 PM.
#12
Posted Oct 04 2009 - 10:51 PM
Stefan Roess, on Aug 31 2009, 11:27 AM, said:
Looks very critical, also regarding freezes. In 2008 I experienced too my first freeze/disconnection retirement in league, right there.
#13
Posted Oct 05 2009 - 01:16 PM
Otavio Silveira, on Oct 4 2009, 11:51 PM, said:
Stefan Roess, on Aug 31 2009, 11:27 AM, said:
Looks very critical, also regarding freezes. In 2008 I experienced too my first freeze/disconnection retirement in league, right there.
that track had these issues back in 2003 already - we had lots of disco/freezes back in the GL 67-only days at Zeltweg, which made us ditch it from future seasons.
#14 Border Reiver - guest
Posted Oct 06 2009 - 08:02 AM
I am sure I asked you before, but did you find a good shortcut to make it easier to create the .tex file in order to rebuild tracks?
Rob
#15
Posted Oct 06 2009 - 08:40 AM
Border Reiver, on Oct 6 2009, 04:02 PM, said:
I am sure I asked you before, but did you find a good shortcut to make it easier to create the .tex file in order to rebuild tracks?
Rob
Rob,
In what way do you need to rebuild the .tex files?
If it can be automated i'm willing to have a look at it
#16
Posted Oct 06 2009 - 08:48 AM
Ginetto, on Oct 6 2009, 11:20 PM, said:
zeltweg_GINfix_1006.7z
This fix is based on high res version in the download main page
install instructions:
Backup your zelweg folder to be on the safe side
Just extract the 3 files included into your zeltweg folder
Please try it online and report back if still issues
Hope it works...
Ciao!
Thanks Ginetto
Much appreciated and good timing as we have a MGPL league race there this Sun and will be testing tomorrow night.
Is this fix Ok to use after applying the Caballero update (SRMZ button on ATDB)?
Will let you know how it goes.
Cheers
David
#17 Border Reiver - guest
Posted Oct 06 2009 - 11:46 AM
MECH, on Oct 6 2009, 03:40 PM, said:
Border Reiver, on Oct 6 2009, 04:02 PM, said:
I am sure I asked you before, but did you find a good shortcut to make it easier to create the .tex file in order to rebuild tracks?
Rob
Rob,
In what way do you need to rebuild the .tex files?
If it can be automated i'm willing to have a look at it
Most often to fix "broken" tracks you end up having to recompile the circuit which means you need build files. With other tools it is possible to get the gtk, tso, fb files, but the tex is a bit more awkward as you need to match up which texture is assigned to which sub-type. It is possible to do by hand by looking around the gtk file and comparing that to the 3do to see which sub type matches with which texture name. I did this for Oulton Park and it took me a fairly long time to find them all, so for anything complicated it would take ages.
The big thing is to match up the list of sub types to the mips that these call for. Usually you can make good guesses for the mapping values, so those do not matter so much, but a quick way to know which texture should be put onto which wall would be helpful.
Rob
#18
Posted Oct 06 2009 - 12:01 PM
I get the txt file via trk2doZ
open it in excel
order by unk3
order by walltype
now you have all the list of all the wall types and all the specific subtipes, with the section and walls you can find them.
With this list you can already prepare the txt file with only miss the mip names and the correct mapping values.
Now I open the track.3do with 3dosimEd and the gtk in 2 different windows to quick know where are the sections on the graphic 3do and how is originally mapped (you can tell watching it in 3dosimed)
Find the wall, right click have the mipname
Easy enough
I didn't do it for zelweg though, I just fixed a couple of walls that were bothering me in the track.3do
Thank You Frenchy, I'll wait for your feedback
Edited by Ginetto, Oct 06 2009 - 12:06 PM.
#19 Border Reiver - guest
Posted Oct 07 2009 - 03:43 AM
I think I follow that.
The list in Excel tells you how many types and subtypes you have to find and also which section to look for them in the gtk/3do. With that you can then just go straight to them and do not need to check every wall, (which is what I did before to check that I did not miss anything).
I have printed your instructions, so I will put that in my track editing notes folder.
Rob
#20
Posted Oct 07 2009 - 06:37 AM
Thank You Rob for explaining my thoughts in a proper English
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users