[shavit]

https://github.com/ValveSoftware/sou...md.h#L164-L165

Those variables are exposed in the OnPlayerRunCmd forward.
I need to figure out how to properly predict the value for them.

What I've tried, and solutions I've been suggested to try - none have worked: (mousex)
1. Delta between the last viewangle's and current's Y axis, divided by m_yaw. < not even close, and I get a float. Results aren't correct even by rounding to floor, ceiling or nearest.
2. The same as 1, but m_yaw is divided by sensitivity.
3. Took the player's m_filter value into account and divided by 2.
4. Solutions found in a cheating forum, won't link for obvious reason. But they don't seem to properly predict this either.

I feel like I'm doing something very wrongly, any help please?

[backwards]

I don't believe it's possible to predict accurately. If you look at the source code for where it's placed into the usercmd it's converted to an int from a float and rounded.

https://github.com/ValveSoftware/sou...mouse.cpp#L555

This differs depending on the mouse samples accumulated over the update interval which may be tied to fps. The most accurate thing you can do that i can think of is get an average of large view angle jumps where they should be above 0.5 rounded and compare that to the short value to see if it has any sort of consistency. I'm assuming you're wanting this for an anti cheat detection method. The only time I've seen a public attempt at using this is on TortoiseOrHare's github for nocheatz.

https://github.com/TortoiseOrHare/No...ouseTester.cpp

[shavit]

Originally Posted by 1337norway I don't believe it's possible to predict accurately. If you look at the source code for where it's placed into the usercmd it's converted to an int from a float and rounded. https://github.com/ValveSoftware/sou...mouse.cpp#L555 This differs depending on the mouse samples accumulated over the update interval which may be tied to fps. The most accurate thing you can do that i can think of is get an average of large view angle jumps where they should be above 0.5 rounded and compare that to the short value to see if it has any sort of consistency. I'm assuming you're wanting this for an anti cheat detection method. The only time I've seen a public attempt at using this is on TortoiseOrHare's github for nocheatz. https://github.com/TortoiseOrHare/No...ouseTester.cpp

NoCheatZ-4's detection method seems promising. Thanks for linking!