What almost all aim bots are missing

Posts 1624 of 24 · Page 2 of 2
Quote Originally Posted by FlutterM4rk View Post
anomalies as in your anti-virus popping up, your pet sitting on your hand, anything that breaks the pattern and does happen irl

when you play for hours, you gradually become less 'sharp', if you don't; sure there's your anomaly, but it's not really an anomaly when it goes on for weeks or months. again, easy to get a pattern from
But the aimbot relies on someone actually playing the game! Who has the power/time/electricity/internet connection to play for a week? At the end you're still the controller, but you drop the task of aiming for the computer instead! You leave it, you'll die eventually as your character can't dodge balls itself or atleast you'd need a quantum computer to analyze every single attack and its relative position, speed and risk on your character and the rest of the nasty image recognition stuff.

Speaking of unexpected real life events, sure let it happen, you'll fuck up dodging in a very intensive task and you'll lose health, auto nexus/backup fast. Seems legit also xD

If I'm not aware of something that can automate the whole game then tell me, I only use the client because I get disconnected while playing the official game, and some other informational tools but I don't really bother with 'actual hacks' if you let me name it.
This is something I've been thinking about too. The way I see it, the primary issue is that Aimbots change their target very quickly and jump the aim direction directly to it. In normal aiming, the aim direction would transition from one position to another - since they actually have to move their cursor - cursors can't just teleport directly to where they are needed. If Aimbots implemented a short, smooth transition aiming between different targets, that would be one big step on making them look more humanlike.

The problem with this is that it would need quite a big change in code workflow. You'd basically need either a multithreaded approach or an event based approach. Because the transitioning would take some time and if you have functions which take a lot of time in code normally - the code will hang.

Event-based code would look kinda like this:

Event-Based: (probably better)
Code:
(other bot code...)
...
(other aimbot code...)
doTransition(); //call transition method

doTransition(){
 (transition blah blah)
 onTransitionDone(); //call this method
}

onTransitionDone(){
(do the rest of the aimbot code...)
}
Point is, it's way easier to just have the aimbot how it already is in terms. I'm not sure how the clients like 059's client is made - if they're made by patching the ROTMG client with Jpex or similar - this may be kinda a lot of effort to do. If you want to make an aimbot KRelay plugin or similar, however, this isn't too hard.

- - - Updated - - -

Quote Originally Posted by MI550 View Post
you'd need a quantum computer to analyze every single attack and its relative position, speed and risk on your character and the rest of the nasty image recognition stuff
I think you are overestimating how difficult that task is for a computer. If you had a really bad implementation of getting data like image recognition, of course, you would need a kinda good computer. What people do instead is use proxy servers and access gamedata directly e.g: KRelay https://www.mpgh.net/forum/showthread.php?t=1371239. Keeping track of loads of projectiles is not intrinsically hard - the game client does it for example. All that's left to do is process this data and make decisions. This processing part can be as intensive as you want to make it, but it doesn't have to be - there are many options.

Quote Originally Posted by MI550 View Post
I'm not aware of something that can automate the whole game
Just because you are not aware of one doesn't mean that it requires a quantum computer to run. People just don't really make stuff like this because it's a lot of effort to code I guess.
Quote Originally Posted by makurell View Post

I think you are overestimating how difficult that task is for a computer. If you had a really bad implementation of getting data like image recognition, of course, you would need a kinda good computer. What people do instead is use proxy servers and access gamedata directly e.g: KRelay https://www.mpgh.net/forum/showthread.php?t=1371239. Keeping track of loads of projectiles is not intrinsically hard - the game client does it for example. All that's left to do is process this data and make decisions. This processing part can be as intensive as you want to make it, but it doesn't have to be - there are many options.


Just because you are not aware of one doesn't mean that it requires a quantum computer to run. People just don't really make stuff like this because it's a lot of effort to code I guess.
Thanks for the insight about projectiles, I actually don't know anything about the client except it's probably written in ActionScript3 or something like that. But I honestly know absolutely nothing about K-Relay and there's no documentation/information in its post or whatsoever to let me know what is it and what it does.

Overall, I wasn't saying these words to relate to my actual purpose of the thread, enhancing aimbots I was just replying to the dude that talked about anomalies and stuff, but still thank you for sharing the same opinion and suggesting the two approaches.

Speaking of risk of single-threadedness, a small Google showed that there's a 'Worker' API to handle multi-threading, but still don't you think that separating the UI thread from a calculatory aiming one might introduce false aim, or late aim?

Why would the code hang? Shouldn't there be, like any game, a main loop that calls various functions, so the aim process is only a part of a loop and won't hang anything? And subsequently the transitioning itself doesn't necessarily have to be executed in a self-containing loop, it can just be a variable updating every second frame or something and this way we can ensure less overhead.
Quote Originally Posted by MI550 View Post
Thanks for the insight about projectiles, I actually don't know anything about the client except it's probably written in ActionScript3 or something like that. But I honestly know absolutely nothing about K-Relay and there's no documentation/information in its post or whatsoever to let me know what is it and what it does.

Overall, I wasn't saying these words to relate to my actual purpose of the thread, enhancing aimbots I was just replying to the dude that talked about anomalies and stuff, but still thank you for sharing the same opinion and suggesting the two approaches.

Speaking of risk of single-threadedness, a small Google showed that there's a 'Worker' API to handle multi-threading, but still don't you think that separating the UI thread from a calculatory aiming one might introduce false aim, or late aim?

Why would the code hang? Shouldn't there be, like any game, a main loop that calls various functions, so the aim process is only a part of a loop and won't hang anything? And subsequently the transitioning itself doesn't necessarily have to be executed in a self-containing loop, it can just be a variable updating every second frame or something and this way we can ensure less overhead.
More information on KRelay can be found here: github(dot) com/TheKronks/K_Relay_Plugin_Documentation/blob/master/README (dot)md.

There are various APIs to help with multithreading. Multithreading isn't intrinsically hard either - it'll just need a big change in workflow. People make aimbots by patching the game client - which is difficult to edit easily as opposed to just making a program.

Of course, having a state where the aimbot is transitioning between two directions will result in reduced accuracy, however, that's kind of the point - to make it more humanlike. It probably will still be pretty pro aim.

Yes, games have main loops which call various functions, however, the functions they call will complete in a very small amount of time. If a transitioning function takes, e.g 200ms to complete - that is a considerable amount - hanging for 200ms every frame (that is, if you had a transition loop on the main thread). (a transition function would need to take a bit of time, to provide a smooth transition - the transition may take 200ms to complete).

Your last point about the transition is valid. It's a synchronous version of the event-based code. It would look like:
Code:
mainLoop(){ //is called once every frame
(other bot code...)
...
(other aimbot code...)
doTransition(); //call transition method
}

doTransition(){
 (transition blah blah)
 if(transitionDone){
 onTransitionDone(); //call this method
 }
}

onTransitionDone(){
(do the rest of the aimbot code...)
}
It could of course also look like (variable polling instead of event driven approach):
Code:
transitioning=false; //global variable
mainLoop(){ //is called once every frame
(other bot code...)
...
(other aimbot code...)
 if(transitioning){
   (transitioning code)
   (if finished transitioning, set transitioning to false)
 }else{
  (other aimbot code...)
  //if need to transition, set transitioning to true
 }
}
Anyway I'm not saying that these aren't possible, I'm just suggesting why people aren't.
Quote Originally Posted by makurell View Post
This is something I've been thinking about too. The way I see it, the primary issue is that Aimbots change their target very quickly and jump the aim direction directly to it. In normal aiming, the aim direction would transition from one position to another - since they actually have to move their cursor - cursors can't just teleport directly to where they are needed. If Aimbots implemented a short, smooth transition aiming between different targets, that would be one big step on making them look more humanlike.

The problem with this is that it would need quite a big change in code workflow. You'd basically need either a multithreaded approach or an event based approach. Because the transitioning would take some time and if you have functions which take a lot of time in code normally - the code will hang.

Event-based code would look kinda like this:

Event-Based: (probably better)
Code:
(other bot code...)
...
(other aimbot code...)
doTransition(); //call transition method

doTransition(){
 (transition blah blah)
 onTransitionDone(); //call this method
}

onTransitionDone(){
(do the rest of the aimbot code...)
}
Point is, it's way easier to just have the aimbot how it already is in terms. I'm not sure how the clients like 059's client is made - if they're made by patching the ROTMG client with Jpex or similar - this may be kinda a lot of effort to do. If you want to make an aimbot KRelay plugin or similar, however, this isn't too hard.

- - - Updated - - -



I think you are overestimating how difficult that task is for a computer. If you had a really bad implementation of getting data like image recognition, of course, you would need a kinda good computer. What people do instead is use proxy servers and access gamedata directly e.g: KRelay https://www.mpgh.net/forum/showthread.php?t=1371239. Keeping track of loads of projectiles is not intrinsically hard - the game client does it for example. All that's left to do is process this data and make decisions. This processing part can be as intensive as you want to make it, but it doesn't have to be - there are many options.


Just because you are not aware of one doesn't mean that it requires a quantum computer to run. People just don't really make stuff like this because it's a lot of effort to code I guess.
You don't have to multithread or do something overtly complex. Since the normal autoaim code already chooses a gameobject as a target, you could have a simple system like this:
-keep track of which object is being targetted
-if the target is different from the last, set a flag to denote doing a smooth aim change
-each time the autoaim check is run, if the flag is set then keep doing the change until it is within x% of the angle to the enemy
-once arrived then turn the flag off
Could make the angle changed per shot attempt partially random and based off time so it will be the same with slow vs fast weapon. If you want to go more advanced and more realistic, change the angle based on where your mouse would have had to have moved to reach the other enemy.

Not really looking to implement this because as far as I know you cannot get banned for autoaim alone, however it does make you more "sus".
Quote Originally Posted by 059 View Post
Not really looking to implement this because as far as I know you cannot get banned for autoaim alone, however it does make you more "sus".
Well there's still more into the thread other than averaging out the precision, the instant fire/stop could be discussed also. I believe it could be the easiest of them all.
Brillant post, thanks to have taken the time to share this though
Posts 1624 of 24 · Page 2 of 2

Post a Reply

Similar Threads

Tags for this Thread

None

Need help?