GLITCH! Please fix Animation Glitch!

Hi everyone!

I'm having quite a struggle with random crashes of my game. I have been talking with Clickteam about this for a time and they already mentioned to fix an issue that might cause this in a patch. But even with the current version, it is not fixed!

What happens is this (I hope I understand Clickteam correctly with that, if not, you're welcome to correct me)

When you spawn lots of objects in rapid speed and give randomized animation sequences (to change appearence), sometimes the animation numbers of these sequences screw up somehow and become impossible values (for example animation sequence -14, which is not possible to do with regular events). Because Multimedia Fusion 2 cannot handle these animation sequences, it appears to use animations from entirely different objects (look at screenshot) or simply crashes the game. Clickteam already tried to do a work-around by catching "impossible values" and simply replacing it with 0, but this seems not to work in this case.

This screenshot shows a random "sparkle" object appear at the enemy. But instead of showing the "sparkly object" animation, it takes the animation from a different object and appears as a solid block!
http://www.paulinthenet.de/panda/ani_glitch.jpg

I only have basic programming knowledge, but this glitch might possibly lead to memory leaks and other issues like crashing, so it's not a "visual glitch" that I'm addressing here. I'm posting this, because maybe this might be the reason for other people's issues with their games as well and they might report this glitch under different circumstances.

Please Clickteam have another look at this issue, it might fix the random crashes of my game for good!
Very much appreciated

Could you create an example file recreating this issue? There might be an easy workaround...

What other event except "change animation sequence to..." is there to do a workaround?

Well "change animation sequence to..." doesnt make any crashes or use animations from different objects in any of my games whatsoever, and i tend to use it a lot. So it's rather obvius your game is using it in a certain manner that causes these issues to occur...

thats why an example of the mfa that does crash would be helpfull to see why and to see what fixes there might be to it...

Yeah... an example that can recreate the issue would be super helpful... I'm not sure Clickteam has much to go on with "please fix animation glitch!"

I actually have this happen all the time with CoI and found that using 'Load on Call' was the culprit in most cases. I haven't been able to forcefully re-create it, but again I've fixed pretty much all occurrences I've had of this by un-checking that pesky Load on Call box. I haven't been able to get a solid answer to what causes this from my talks with CT staff or other developers, but try what I mentioned above and let me know if it works for you.

Quote:

---

Originally Posted by Skyhunter

Well "change animation sequence to..." doesnt make any crashes or use animations from different objects in any of my games whatsoever, and i tend to use it a lot. So it's rather obvius your game is using it in a certain manner that causes these issues to occur...

thats why an example of the mfa that does crash would be helpfull to see why and to see what fixes there might be to it...

---

I can guarantee you that I'm not doing anything "out of the ordinary" with my events. Basically I have an alterable value that gets set to a random(#) and then there are events for each #, that change the animation of the object to one that I have selected by hand from the animation selection menu of that object (no expressions are used with the animation sequences!). So with this method, it should be impossible to screw the animations up that bad that they get taken from entirely different objects. Maybe I'm using way more different animations within the same object as others, or maybe it's caused by the use of Global Events - but these two things are ordinary features of MMF2, so I suppose I'm allowed to use them as much as I want, right? I've already sent an MFA to Clickteam and they have acknowledged that it's a problem on their part, they know this issue exists and it was causing the game to crash for them too. They said they make a work-around for it in the current version, but like the screenshot shows, it is not working.

Quote:

---

Originally Posted by Chaos

I actually have this happen all the time with CoI and found that using 'Load on Call' was the culprit in most cases. I haven't been able to forcefully re-create it, but again I've fixed pretty much all occurrences I've had of this by un-checking that pesky Load on Call box. I haven't been able to get a solid answer to what causes this from my talks with CT staff or other developers, but try what I mentioned above and let me know if it works for you.

---

Thanks for your help - I've checked my MFA again and all the "load on call" have always been unchecked. I'm pretty sure I've never even touched that checkbox before, so in this case, I'm afraid it's not the problem/solution.

Quote:

---

Originally Posted by Konidias

Yeah... an example that can recreate the issue would be super helpful... I'm not sure Clickteam has much to go on with "please fix animation glitch!"

---

I'd love to do that, but my time is very limited. The issue is extremely hard to recreate and appears at random, so it might take days or weeks to create a "example MFA" with the same conditions to have this issue appear. As I mentioned in my first post, I've sent an early version of my game as MFA file to Clickteam before, but I will not make this public.

Yes, if you coul disolate the problem a small application with just the object showing the problem that would be good.
Do I already have your MFA?

Yes, you should already have the MFA. We talked about this over PM in june last year. The MFA should be called something like Pandagame_frametest. If you don't have it anymore, I can upload it again for you.

I'm not sure if the problem still exists if I just try to isolate the single object, since it uses animations from all other objects as well. It would take a lot of time to find the exact mix of objects that causes this issue again, because it appears at random and is not recreated that easily. Maybe it's better for this "investigation" to use the whole MFA that I've sent you back then.

Many thanks!

I've noticed that MMF2's set animation number, frame or direction logic can easily go out of array bounds and take images from subsequent memory, particularly in directly setting it from xLua where there is no sanity check afaik. Now normally, MMF2's bounding should prohibit this and preclude the possibility, and I use the set anim value functions all the time with no problem. But you mention that your problem is tied to generating tons of objects of different kinds

Without seeing your code it is difficult to test, but I'd hypothesize one thing to test for is if this is related to the "create object scoping" glitch. Is this only occurring when the number of objects in your frame reaches the maximum #, particularly if you didn't raise the frame from the default 500 up to the maximum 20000? In MMF2, if you have a line of code that creates an object and then sets its values, that line will malfunction if the create event fails, scoping all objects of that class and applying the actions to them without creating anything new. Its a particularly common cause of screwing up peoples maths and references in code, and if you had any particularly complicated math going on in terms of setting the animation values, maybe that math is getting screwed up from what *seems* sound and stable, and setting it to something out of bounds, which in turn yanks animations from other objects in memory, as after all they are all stored next to each other


If anything, can you post a screenshot of the event editor / event list editor snippets of code where you set the animation values for objects directly?

When it has occurred for me it was never in reference to forcing animation frames or anything. Only changing animation directions or what animation is playing. It is entirely possible its due to hitting a cap of objects per frame that I forgot to set. I'll look into it for my game.

Sorry, cant find it. Can you send it to me again?

Quote:

---

Originally Posted by Francois

Sorry, cant find it. Can you send it to me again?

---

No problem, you have a new PM ;)

Quote:

---

Originally Posted by Pixelthief

I've noticed that MMF2's set animation number, frame or direction logic can easily go out of array bounds and take images from subsequent memory, particularly in directly setting it from xLua where there is no sanity check afaik. Now normally, MMF2's bounding should prohibit this and preclude the possibility, and I use the set anim value functions all the time with no problem. But you mention that your problem is tied to generating tons of objects of different kinds

Without seeing your code it is difficult to test, but I'd hypothesize one thing to test for is if this is related to the "create object scoping" glitch. Is this only occurring when the number of objects in your frame reaches the maximum #, particularly if you didn't raise the frame from the default 500 up to the maximum 20000? In MMF2, if you have a line of code that creates an object and then sets its values, that line will malfunction if the create event fails, scoping all objects of that class and applying the actions to them without creating anything new. Its a particularly common cause of screwing up peoples maths and references in code, and if you had any particularly complicated math going on in terms of setting the animation values, maybe that math is getting screwed up from what *seems* sound and stable, and setting it to something out of bounds, which in turn yanks animations from other objects in memory, as after all they are all stored next to each other


If anything, can you post a screenshot of the event editor / event list editor snippets of code where you set the animation values for objects directly?

---

The object maximum value has been risen to 5000 for each frame and currently in the most busy frames I'm at 1500-2000, but not higher, even with all kinds of objects spawning.

Here's a screenshot of the events:
http://www.paulinthenet.de/panda/ani_glitch_events.jpg

Well theres certainly nothing in that event that sets the animation number or direction or frame to a value, but yeah that code would set *all* particles to animation health, b = 3+r(5), etc, if the create object event failed. Still, none of that should cause memory overflow, just a nasty grouping of your objects. So no red flags

Oh, I just noticed an error with the red arrows that I drew on there, which suggest all animations get set to health as shown in the box. Actually the stuff in the right box is different for each event to give everything different looks and properties. Each number has a different animation, just as the comment above them all shows (1=health, 2=mana etc.). But good to know you don't see any red flags :)

Thats quite some number of objects. Are you using a loop to destroy them aswell? Some problems may occur when destroying objects when theres a lot, rather than creating them... try destroying them in a loop, also make sure to also disable their alterable values before destroying them so no events can fire on those objects while being destroyed.
And are you using appear dissapear or any movement preset animations on that object?

Make sure you sanitize your random(n) calls too, I can't see how you set value C there, but if you take random() of a negative number, bad things ensue!
If you have a line of code that sets the animations to a random() value tucked in there somewhere that could potentially be the problem, I'm just taking a wild stab

I know this isn't the solution... but is there any reason why you have so many near identical foreach loops? Is it just because each condition is setting to a different animation? Couldn't you just sequence your animations so you can call them with an expression? You'd reduce loop checks down to a single condition/set of actions instead of one for each different type of particle.

Blue66, every time I've needed something fixed, I've narrowed the problem down to an .mfa demonstrating the problem alone, and posted it here in the community, not just to the CT staff. If you need help, be helpful. I've created lots of games and have never had an invalid frame number return. I can not figure out how to reproduce it from screenshots of MMF2 events, but with an .mfa it's much more likely that people here can figure out what's going on. It eases the burden for CT if we users can isolate the problem. Then they just need to fix it rather than having to figure out where the problem is.

Without an example, users can only come up with theories to what's wrong.

Yeah what Nifflas said... Even if it takes a few days to narrow down the problem and isolate it into an MFA, it's worth it to get the game fixed. This can happen with any programming language or game dev tool, so don't feel like it's just an MMF thing. You just have to really try to find a way to cause the problem on command and then narrow it down to just that. It can't be entirely random.... there's gotta be at least some sort of method to the madness.

Quote:

---

Originally Posted by Konidias

I know this isn't the solution... but is there any reason why you have so many near identical foreach loops? Is it just because each condition is setting to a different animation? Couldn't you just sequence your animations so you can call them with an expression? You'd reduce loop checks down to a single condition/set of actions instead of one for each different type of particle.

---

I'm not only setting the animation sequence in each event differently, but other properties as well (like random positioning, directions, if it rotates or not, if size increases or not etc.), so I can't see how I could handle this within one event. Maybe I'm just not good enough to be that efficient :(

To Nifflas and in general:
As I've said before, I'd love to be more helpful and provide examples, but in this case the issue is so random (it happens about once every 10-15 min of regular gameplay), that I'm literally fishing in the dark here. I could just copy&paste random stuff from my events into a new MFA, but it would take many many many hours, days or even weeks to find out what constellation of things is necessary to exactly recreate this problem. I'm in no way professional or experienced enough to narrow this issue down so easily like some of you guys probably could. I'm also on a very tight schedule for my game currently, which makes it absolutely impossible for me to spend another day or even weeks on this issue like I've done before. Trust me, I've spent A LONG TIME on this problem before, and in the end not even Clickteam could find a solution to this issue. BTW they were the guys who actually found out that the problem is animation related! So I was not even able to figure that one out by myself (my best guess was it has to do with movement types of different objects).

If its different in each events, its likely that only some of the events have the offending code.
If you *do* find a way to actually see if this glitch occurs or not in testing, try isolating it by disabling some of the particles and not others, and figure out which ones do it
Or post us the related code for each one- it doesn't necessarily have to be assembling a blank MFA with copy/pasted code like you say, a collage would work