Has anybody solved the "shaking stairs"

[R1CH]

This should hopefully, finally be fixed now. The current release contains backported 3.20 code for input and frame handling. Please try it out and let me know!

[Digital]

Thanks R1CH, it's partially fixed. I've used Q2's default cl_maxfps (90 ) for years and for me it's perfect now. But a mate who uses cl_maxfps 99 tested and told me the issues I mentioned in this thread are still there. I tested and noticed he is right. With maxfps 90 or anything under 90 things are running smoothly. But with maxfps 91 or above the movement/gravity/physics issue remains. Hope you find the source of the problem...

[R1CH]

I think I may have found what's causing it, but I suck at double jumps and such so let me know any feedback from this version: http://www.r1ch.net/stuff/r1q2/pre/

[Digital]

Nope, the problem is still there with b4246. With maxfps 90 it works but not with 91 ->

[R1CH]

Do you think you could get some demos of how some cl_maxfps 120 cl_async 0 jumps react in NoCheat and R1Q2?

Also, is there any difference if you try m_directinput 1?

[Digital]

It seems fixed. The maxfps 91 and above issue seems to be there only when launching R1Q2 locally = run R1Q2 -> map q2dm1 -> cl_maxfps 91 and above results in glue-effect when doublejumping. But when doing the same on a server the glue-issue is gone so hopefully it's fixed