[Crash] Crashing after loading screen, before entering the game

[Barbafire]

Hello hivers once more. I'm here to ask for help about the issue related to the name of the post: Crashing when finishing loading, before game start.

About one month more less, my project Outerworld Arena is having issues with a crash that looks like happen randomly. The crash happens right after the game finishes loading (Loading Screen), and right before entering the game (First instance of the game).

There are two major things that changed on the project, when this started to happen:

• Warcraft update to 2.0.3.23175
• Optimizing Process of my triggers

So, at a first moment, I thought: "Probably I made a mistake when optimizing the initialization. That's why it is causing the crash". So, I've moved everything to a secondary initiation, that happens only after the game succesfully load, and guess what? It keeps crashing randomly. Map Initialization logic has been discarded.

Making some tests, the characteristics of the crash are:

• It happens after finishing loading, but before entering the game;
• It doesn't crash everyone. Always crash a random number of players, on lobbies with 3 or more players;
• Being alt+tabbed increases the chance of crash;
• Clicking like hell into the game screen increases the chance of crash;
• Once inside the game, no one crashes anymore;

Things I've already tried for now:

• Check corrupted custom assets; All of them have size above 1;
• Removing any pre-placed custom asset on the game's scenario;
• Clearing Map Initialization trigger, leaving it empty;
• Removing custom loading screen;
• Using "call SetRandomSeed(0)" in the Map Initialization; Some players said that probably players were randoming different numbers, and when the game starts and server synchronize, caused the crash;
• Crucify myself;

If it was a warcraft update issue, other maps should suffer from this, but I didn't find anything related. If it was my optimizing process, the removal of the triggers from map initialization should be enough to solve this.

So here I am, asking for help for those who have any idea about what is happening here. Any idea is welcome.

Thanks in advance.

 

[A]mun]

Those crah-threads are always unpleasent to read.

Anyway, have you tried earlier versions?

It seems like rigorous testing is required. For starters: going through all the player slot permutations.
I can lend a hand, just dm me.

 

[nana7mi]

Your issue is exactly the same as what is happening on my side right now. I would like to add some observations based on my current situation:

1. At the same location, with the same player slot, same race, and same host, some players crash sometimes, and other times they don’t, under seemingly identical conditions.
2. Players who are “prone to crashing” seem to crash more easily and more frequently than others.
3. My map uses the following trigger during initialization. I am not sure whether it is dangerous or related to the crash:
4. According to my friends, the increased crash rate occurs only on this specific map.

Melee Initialization all
Event
Map initialization
Conditions
Actions
Melee Game - Create starting units (for all players)
Melee Game - Use melee time of day (for all players)
Melee Game - Limit Heroes to 1 per Hero-type (for all players)
Melee Game - Give trained Heroes a Scroll of Town Portal (for all players)
Melee Game - Run melee AI scripts (for computer players)


My friends sent me their crash logs, and I consulted ChatGPT-5.
Its conclusion is that the crash is caused by a DirectX 11 initialization / GPU driver compatibility issue, which results in a Null Pointer Access Violation.


ChatGPT-5 reached this conclusion not because of the ACCESS_VIOLATION message itself (which is a common and non-specific error), but because the crash point in the logs shows that the game fails during the DirectX 11 initialization stage on the main thread—right after the graphics device and display mode setup, but before entering the main render loop. This strongly indicates that the failure happens in the DX11 initialization pipeline, rather than in map scripts, triggers, or model loading.

Since certain players crash much more easily than others, this theory does make sense — but I’m still trying to understand the underlying cause. My friend also report that only my map shows a noticeably higher crash frequency, and this matches my own experience across many games. Empirically, the overall crash rate looks roughly random at ~10%, while for certain players it rises to around ~30%.

So, if it is a render issue, how can we solve this xD?

As I said upthere, I've made all kinds of changes on my system, even turning it back as it was before, and the crash continues, now with a great lower rate, but it still exists.

I’m working on a likely suspect.

In your map, do you have any particle-based models located at the players’ starting positions?

In my own case, I discovered two things:

1. There are some native Warcraft III fire models placed at the initial camera location — and I strongly suspect this may be related.
2. There are also imported models at the starting area that might contain particle effects, which could be contributing to the issue.

Oct 24, 2025:

• I removed the particle-related model at the starting point, as it seemed to have no effect.
• My map originally had 30 Map Initialization triggers used for setting up various configurations. I have now reduced them to about 5, and it appears this has lowered the disconnect rate.

 

[Chaosium]

I sometimes get crashes when testing a campaign map and that seems to happen mostly when I press a key too quickly as soon as the loading is finished as opposed to waiting for 3-4 seconds, not sure if there's any veracity to it but that's what I can say on the matter

 

[Barbafire]

Those crah-threads are always unpleasent to read.

Anyway, have you tried earlier versions?

It seems like rigorous testing is required. For starters: going through all the player slot permutations.
I can lend a hand, just dm me.

Thank you so much A]mun. I'll not spend more energy into this for a while, because I've emptied all my ideas and possibilities about this problem.

Since I've made every kind of thing, and didn't reach a 100% solution, the only thing remains is the game update. I saw that they changed some natives to fix issues related to the editor, so maybe, older maps aren't affected by this crash issue, but new ones are.

I've returned to my grotesque initialization, and the crash issue almost got solved. There are just two things that are crashing players now:

• Spam click at the loading screen;
• Alt tabbing when game is "Waiting for other Players";

Its conclusion is that the crash is caused by a DirectX 11 initialization / GPU driver compatibility issue, which results in a Null Pointer Access Violation.

So, if it is a render issue, how can we solve this xD?

As I said upthere, I've made all kinds of changes on my system, even turning it back as it was before, and the crash continues, now with a great lower rate, but it still exists.

I sometimes get crashes when testing a campaign map and that seems to happen mostly when I press a key too quickly as soon as the loading is finished as opposed to waiting for 3-4 seconds, not sure if there's any veracity to it but that's what I can say on the matter

It is not isolated. Everyone here noticed the same thing, and on the crash tests, the same crash by spamming click the loading screen.

---

EDIT: For me, instead of crash, I get logged out from battlenet. Some players also experience the log off issue, instead of the crash.