Theoretically better unlag based on bose position restoring

[a1batross]

How it (should) work?

The main idea is to remember client's bone position on each frame. It's intended to help with situation where one player is updated in real-time, but lagging second player not yet got this updates. Unlag fixes it only for player position(i.e. working when player is running) and does not restores any other states.

Example:
Let's say we have a server with two players and they "see" each other at one frame lag.

• Frame[0]:

  • Player[0]:
    • Physics is running.

• Frame[1]:

  • Player[0]:
    • Physics is running.
  • Player[1]:
    • Player[1] is connected.
    • Physics is running.

• Frame[2]:

  • Player[0]:
    • Player is ducking, i.e. his head is a bit below.
    • Physics is running.
  • Player[1]:
    • Player[1] shoots at Player[0] right to his head. Due to lag, Player[1] sees Player[0] as NOT ducking.
    • Physics is running. Bullet is traced.
    • WITH bone position history:
      • Player[0] origin restored at Frame[1] AND head hull position is restored at Frame[1] position.
    • WITHOUT bone position history:
      • Player[0] origin restored at Frame[1], BUT his bones are set up in Frame[2] position, because previous sequence is unknown.

Yes, this demonstrates only sequence changing(i.e. ducking) but due to unlag imperfection, there can be more real life examples.

P.S. I did not tested this at all as I don't have any server and players. :)
P. P. S. Originally I thought to make this for Xash3D, but it would be faster to check this concept on bigger community.

[a1batross]

Also, don't forget to set up sv_bone_unlag 1 in game config.

Interpolation is done at a4f4e34

[WPMGPRoSToTeMa]

i -> edict->v.modelindex

[a1batross]

Oops. Thanks :)

[In-line]

No need to include this under REHLDS_FIXES

[a1batross]

@In-line ok, will remove in next commit.

[di57inct]

is this the same as how source handles it? if so, PLEASE DONT FUCKING TOUCH IT
i don't want that cancerous source unlag on my goldsource games

[a1batross]

@di57inct I don't know how Source doing it.

[di57inct]

Sounds to me like that's it.
https://youtube.com/watch?v=6EwaW2iz4iA

[a1batross]

@di57inct I hear only common words, without the details. Will the reasoning be better? Or just childish "it's cancerous"?

[di57inct]

I don't have the patience to argue or explain it to you.
Don't force it into the main ReHLDS or if you do, make it optional.
Don't ruin GoldSrc games.

[WPMGPRoSToTeMa]

Sounds to me like that's it.

That's how it currently works (without this commit).
This commit adds only correct interpolation of animations, not player position, it'll be the same.

[di57inct]

Okay I sure hope so. Thanks for the info.

[a1batross]

Broken CS1.6 hitboxes? Of course not. It only makes them correct for small time window when the animation switches. 2ReHLDS devs: I don't care about this patch anymore. Feel free to close it or rework it without attribution. пт, 5 февр. 2021 г., 19:26 Very Strange Karaulov <[email protected]>:

…

This is not fix for bad hitboxes ?)) — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#624 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAYQTALPOWKPP3QF4M3K2RDS5QLZ5ANCNFSM4FD3ARRA> .

[SNMetamorph]

Is it still actual? It'll be cool to implement it, because there are issues with hitreg in fast-paced mods

[a1batross]

@SNMetamorph nah, not really.

You can try reimplement it for Xash though.

[Nord1cWarr1or]

@a1batross hope this PR will be completed & merged

[UnrealKaraulov]

No video with how to it works? No possibility to merge it?)