[PolarisGang]

I'm making a tool to make more condemning evidence(linear aiming, accuracy, extreme difference in angles between 2 ticks and more) during manual review of player-demos in CS:GO but I lack imagination as for more features I can add.

As of right now it has:
+Safe way of approaching the game's memory to ensure noone gets falsely banned
+Automatically making & loading ingame-config to make the process faster/more fluent
+Boring but functional GUI - gyazo (dot) com/fd8411db1892255a323a2e62e30d2cc0.png
+Collects player's view-angles for each tick
+Can find nearest enemy and calculate distance to head
+Can use first collected view-angles as baseline and subtracts the rest with it (could be useful for drawing etc.)
+Can convert all the above values from angles to pixels
+Can add aim-time
+Make reports that looks like this (made up values used):

Code

AimAnalyze Report
---------------------------------
Tick: 1/5 | ViewAngles = X: 2,30912  Y: -101,6177 | Distance to HS = X: 2,218454  Y: -360,6094 | Aim-time = No tickrate chosen!
Tick: 1/5 | ViewAngles(px) = X: -1408,227  Y: -32 | Distance to HS(px) = X: -4997,358  Y: -30,74354 | Aim-time = No tickrate chosen!

Tick: 2/5 | ViewAngles = X: 2,45344  Y: -100,2467 | Distance to HS = X: 2,379275  Y: -359,314 | Aim-time = No tickrate chosen!
Tick: 2/5 | ViewAngles(px) = X: -1389,228  Y: -34 | Distance to HS(px) = X: -4979,407  Y: -32,97222 | Aim-time = No tickrate chosen!

Tick: 3/5 | ViewAngles = X: 1,65968  Y: -100,1024 | Distance to HS = X: 1,562638  Y: -358,9358 | Aim-time = No tickrate chosen!
Tick: 3/5 | ViewAngles(px) = X: -1387,228  Y: -23 | Distance to HS(px) = X: -4974,166  Y: -21,65518 | Aim-time = No tickrate chosen!

Tick: 4/5 | ViewAngles = X: 0,3607998  Y: -100,8961 | Distance to HS = X: 0,2875579  Y: -359,7595 | Aim-time = No tickrate chosen!
Tick: 4/5 | ViewAngles(px) = X: -1398,228  Y: -4,999997 | Distance to HS(px) = X: -4985,581  Y: -3,985004 | Aim-time = No tickrate chosen!

Tick: 5/5 | ViewAngles = X: -0,5772802  Y: -100,9683 | Distance to HS = X: -0,6988829  Y: -359,8787 | Aim-time = No tickrate chosen!
Tick: 5/5 | ViewAngles(px) = X: -1399,228  Y: 8,000003 | Distance to HS(px) = X: -4987,232  Y: 9,685184 | Aim-time = No tickrate chosen!

---------------------------------
End of report

On todo list:
+GOTV support (only POV demos are supported atm)
+Adding recoil compensation for weapons such as rifles into "from crosshair to nearest head"-calculations

Any suggestions?

[SlowMehMo]

-snip-

What about some backtracking? i dont wanna be that guy, but its easy to implement, check for valid tics and abuse the games built in lag compensation.
You may not need a complete compensation, but maybe draw a area for predicting spread patterns.
Not that i would know anything about game hacking using C#, but the menu looks fine, its a really good start. 😉 Best Of Luck!

[Dalvik]

If you want to make accurate predictions of cheaters using the Overwatch system, why not opt for machine learning (just like Valve is doing right now)? I know SharpNEAT is a good library for the NEAT algorithm in C#, though I am not sure if that is the best algorithm in this case.

[PolarisGang]

What about some backtracking? i dont wanna be that guy, but its easy to implement, check for valid tics and abuse the games built in lag compensation.
You may not need a complete compensation, but maybe draw a area for predicting spread patterns.
Not that i would know anything about game hacking using C#, but the menu looks fine, its a really good start.  😉 Best Of Luck!

I mean I could add something for backtracking however I thought it was too obvious for the naked eye. I've participated in 200+ cups and tournaments in CS:GO alone and never seen anyone dumb enough to use backtracking. About the spread, the application launches csgo with a configuration that among other things shows the "spread crosshair". It's really not an issue since the spread is serversided.

As for game hacking, whether you're using C++, C# or VB.NET is irrelevant. Internal cheats are a bit more hassle in managed languages and you're always recommended to use the same language + compiler settings as the game you're making 3rd party code for but it's entirely subjective, the basics work the same.

---

If you want to make accurate predictions of cheaters using the Overwatch system, why not opt for machine learning (just like Valve is doing right now)? I know SharpNEAT is a good library for the NEAT algorithm in C#, though I am not sure if that is the best algorithm in this case.

This is mostly for cup/tournament admins that sit carefully analyzing PoV demos. Feeding overwatch cases wouldn't be sufficient at all.

UPDATE:
Added warning for linear aiming and at how many occurances
Added warning for most likely inhuman amount of angle/pixel travel between ticks.