Tracks TT Circuit Assen 0.8.2

[NightEye87]

The same issue happened with some rally track a few years ago, and for that one it was easy to replace all the movable objects as they were in their separate kn5. Replacing that file with a placeholder had the track go from over 10 GB of RAM to less than 2, and the kn5 itself was tiny. Its something connected to movable objects, a bug in the engine.

I could put the moveable objects in their own KN5 for the next update. Would make bugtesting easier, too

EDIT: They already are; see my post below.

 

[cheetocatto]

Thanks for reporting. It's weird, I've never had any issues. It could be the moveable objects but the autosport layout has almost as many moveable objects and works fine, you say.
Also, "back in the days", those things were really hard on a system because of the collision detection. But these use a very crude collision mesh underneath a higher detailed visual mesh, and I hadn't heard of performance issues before regarding them.
I even did a test once on the Junior Track with ~200 of these cones and all worked fine.

A quick test on my end:
34 grid MX5's on Autosport: 1.97 Gb RAM. ~42 fps. ~49% CPU.
34 grid MX5's on North: 2.00 Gb RAM. ~41 fps. ~49% CPU.

I know physical objects remain static untill touched. Maybe you knocked some over and it causes the RAM to fill up? So I did a test:
34 grid MX5's on North, but after knocking over every cone and tyre stack there is: 1.95 Gb RAM; same.

Could it perhaps be that you're driving in different weather conditions?
I tested it with Sol's clear skies, scattered clouds and snow; all the same.

It might be the cars? But if you're testing for a league, then I guess you tested with the same car.
I tested with M3 GT2: same results.
Type C mod: same.
Dodgy Saab 99 mod: same.

Amount of spectators? They are equal. And with that I mean they are scattered throughout the track as if it were the Autosport layout, with the exact same numbers (the lazyness of me, I know..).
LODs are the same, textures are the same, etc. The only real difference between the layouts are the names of the surfaces, the AI line and the 3% or so increase in moveable objects. I don't think any of those should triple the amount of RAM used.

Maybe some left-over CSP config files somewhere? I really can't think of anything else.. I wish I could help

Many thanks for taking a look at it.

edit: Kinda narrowed down to replay size, reduced quality to medium and size to 460MB and ram usage fell to 4.2 GB (instead of 10, 11 GB). Now... still only happens with north loop layout. It happens in all layouts.

I had one mate that had 20GB of ram usage (out of 32) and the other one had the same numbers as you for ram usage. On my end (and the other guy who got the memory filled) issue started with the session loading memory was already full on car setup screen. He removed some of the .kn5 and had a drop on usage of 1 or 2 GB, apparently the one causing issues was the one that cointained some surfaces and tyres (probaby he refered to the tyre stacks).

Cars i tested: tm-modding Clio R.S. cup, Legion's Formula vee, Kunos Alpha 33 Stradale, Kunos Huracan GT3 and Guerrilla 570S GT4. It flooded the ram at all times.

For weahter SOL 2.2, wather - clear, 5 kph wind. CSP 1.7.7,

 

[NightEye87]

He removed some of the .kn5 and had a drop on usage of 1 or 2 GB, apparently the one causing issues was the one that cointained some surfaces and tyres (probaby he refered to the tyre stacks).

I can only imagine the main kn5 having so much impact on RAM. But that wasn't it or else he would have driven on an invisible track Any idea what kn5 he means?

The two kn5's that have these mentioned objects (in current version already, so no update needed):
ne_ttassen_NORTH.kn5 - cones only
ne_ttassen_AUTO.kn5 - tyre stacks, some cones and the steel roundels

They are not essential for the track and can be disabled in the models_XXX.ini if you want.
Let me know if that helps him or if it's a different kn5 that gives this RAM usage.

 

[Leonardo Ratafiá]

this circuit is so nice! is there any update planned for later?

 

[NightEye87]

this circuit is so nice! is there any update planned for later?

Yes, definately I really want to update this to full 2021+ spec, but due to covid restrictions I couldn't visit the track. Now I finally can again but having long covid and switching jobs soon, I'm afraid I'll have to wait a bit for some time to free up

 

[Vintagedriver01]

If doing the update do not forget to reconstruct the flagpoles in the pitlane, as they shimmer like hell in VR.
And equipped with flags is also great.

 

[NightEye87]

If doing the update do not forget to reconstruct the flagpoles in the pitlane, as they shimmer like hell in VR.
And equipped with flags is also great.

What could I do to reduce this shimmering? They are thin objects, they require hefty AA to render properly. I'm afraid I can't do anything about that.

I added the flags on the todo list; they're long overdue indeed

 

[Vintagedriver01]

What could I do to reduce this shimmering? They are thin objects, they require hefty AA to render properly. I'm afraid I can't do anything about that.

I added the flags on the todo list; they're long overdue indeed

There is another Assen from Easy, his version has flagpoles that do not shimmer that much. Check what his secret is.

 

[Leonardo Ratafiá]

sorry to bump again but just wanted to share this vid which shows couple of the best mods for AC1 but IRL

 

[NightEye87]

There is another Assen from Easy, his version has flagpoles that do not shimmer that much. Check what his secret is.

I just checked and his flag poles are even thinner.
It might be a material property that's causing it. It uses ksPerPixelMultiMap now; I'll see if I can split the flag poles and make them a new/other material, simple ksPerPixel.

His version uses ksPerPixel. It might be that the ksPPMM shader does something funky with alpha channels, detail maps, etc. that become visible for VR users or when no AA is used. It's worth a shot for sure.

 

[AccAkut]

I just checked and his flag poles are even thinner.
It might be a material property that's causing it. It uses ksPerPixelMultiMap now; I'll see if I can split the flag poles and make them a new/other material, simple ksPerPixel.

His version uses ksPerPixel. It might be that the ksPPMM shader does something funky with alpha channels, detail maps, etc. that become visible for VR users or when no AA is used. It's worth a shot for sure.

I think your poles are hexagonal, how are his? Stuff with too many edges can cause strong moirre effects, learned that on my guardrails Maybe smooth shaded cuboids work better.

AC could really use a simple camera facing shader for things like this

 

[NightEye87]

His are 10-sided or so, mine are 6 or 8 sided, I think. Maybe I can test this with no AA and/or low-res, if so I'll figure something out. If not I'll let one of the VR guys test it

 

[Vintagedriver01]

You can always count on me. I did test lots of 5star tracks, but believe me 90% of all these famous ranking 5star track makers do have problems in making a track without shimmer. Maybe somebody could make an manual in how to reduce shimmer objects as some track makers seems to have no problems.

 

[Vintagedriver01]

Ter info: @ninezeroniner on GTplanet did create an VAOfile for Assen.

 

[NightEye87]

Ter info: @ninezeroniner on GTplanet did create an VAOfile for Assen.

I still plan to update the track to 2022 spec (whenever that will happen....), then it will include a new VAO file again and older ones will be broken.

 

[kevinwalter]

Is this still working in the last version? I am unable to get it to work.
AC crashed with a couple of errors.

Redownloaded it multiple times, this is the only mod installed.

Error might have something to do with:

00007FFAABD8FB0C (KERNELBASE): (filename not available): RaiseException
AC\inireader.cpp (488): INIReader::errorReport
AC\inireader.cpp (190): INIReader::getString
AC\trackavatar.cpp (369): TrackAvatar::init3D
AC\trackavatar.cpp (59): TrackAvatar::TrackAvatar
AC\sim.cpp (1066): Sim::loadTrack
AC\racemanager.cpp (351): RaceManager::initOffline
AC\racemanager.cpp (174): RaceManager::loadSessions
AC\sim.cpp (320): Sim::Sim
AC\splashscreen.cpp (319): SplashScreen:ostRender
AC\game.cpp (181): Game::onIdle
AC\game.cpp (210): Game::run
AC\acs.cpp (477): wWinMain
f:\dd\vctools\crt\crtw32\dllstuff\crtexe.c (618): __tmainCRTStartup
00007FFAAD18244D (KERNEL32): (filename not available): BaseThreadInitThunk
00007FFAAE5ADF78 (ntdll): (filename not available): RtlUserThreadStart

When installing the shaders mod I receive te following message:

Error might have something to do with:

AC\inireader.cpp (488): INIReader::errorReport
AC\inireader.cpp (190): INIReader::getString
AC\trackavatar.cpp (369): TrackAvatar::init3D
AC\trackavatar.cpp (59): TrackAvatar::TrackAvatar
AC\sim.cpp (1066): Sim::loadTrack
AC\racemanager.cpp (351): RaceManager::initOffline
AC\racemanager.cpp (174): RaceManager::loadSessions
AC\sim.cpp (320): Sim::Sim
AC\splashscreen.cpp (319): SplashScreen:ostRender
AC\game.cpp (181): Game::onIdle
AC\game.cpp (210): Game::run
AC\acs.cpp (477): wWinMain
f:\dd\vctools\crt\crtw32\dllstuff\crtexe.c (618): __tmainCRTStartup

 

[NightEye87]

Is this still working in the last version? I am unable to get it to work.
AC crashed with a couple of errors.

Redownloaded it multiple times, this is the only mod installed.


When installing the shaders mod I receive te following message:

Haven't tried for a while, so I wouldn't know. I'll do my best to try it soon. Could you post your entire log.txt after a crash?

 

[NightEye87]

Is this still working in the last version? I am unable to get it to work.
AC crashed with a couple of errors.

Redownloaded it multiple times, this is the only mod installed.


When installing the shaders mod I receive te following message:

Works fine for me on the latest CSP.

 

[Fanapryde]

Works fine for me on the latest CSP.

I really wish I could say the same, but just tested with latest CSP and I get this now (don't mind the splits, that's the bezels of my triple setup):

 

[Fanapryde]

@NightEye87
The above must have been a bug in that version of CSP, because I just drove the track again without issues... (0.1.80 prev)