GLITCH! Please fix Animation Glitch!

[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?

[Chaos]

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.

[Francois]

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

[Blue66]

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

No problem, you have a new PM

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:

[Pixelthief]

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

[Blue66]

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

[Skyhunter]

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?

[Pixelthief]

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

[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.

[Nifflas]

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.