109 replies to this topic

[Telmasa]

Bulletsponge0, on 27 June 2015 - 01:54 PM, said:

will these changes fix the bug for Clan ERPPCs where if a second one hits while the paper doll from the first ERPPC is still flashing, then second one will not register any damage?

I'm fairly certain that's just a HUD bug...not totally certain but, fairly so.

[Bulletsponge0]

Telmasa, on 27 June 2015 - 06:05 PM, said:

I'm fairly certain that's just a HUD bug...not totally certain but, fairly so.

doesn't seem to actually do damage though..

[Kjudoon]

The potential end of Assault Lights????

That might coax me back a little, except I already see the unintended consequence of even more instagimps on the horizon.

Once this fix proves effective PGI needs to make sure they don't buff lights back to abnormal supremacy.

Edited by Kjudoon, 27 June 2015 - 11:47 PM.

[FatYak]

Bah....wont believe it until it actually works

And you just know they will screw something else up given their track record

Edited by FatYak, 27 June 2015 - 09:56 PM.

[Neema Teymory]

Sarlic, on 27 June 2015 - 07:52 AM, said:

Does this include something / affect SRMs? I have the most troubles with SRMs. Some reg. others dont.
Ping average is around 100-120.

Yes, the changes will affect SRMs. More generally, the changes will affect all client predicted weapons. This means anything other than SSRMs and LRMs. Keep in mind that each person's experience will still be dependent on their connection quality. However, we still expect that everyone should see a noticeable improvement of some kind.

Jman5, on 27 June 2015 - 09:11 AM, said:

Hey Neema, will these improvements to hit registration have any impact on lock-on weapons like Streak SRMs and LRMs?

No. Lock on weapons are all 100% server authoritative. You might notice this when shooting SSRMs and LRMs with a high ping--there is a delay before the missiles actually launch on your client. HSR does not apply in this case.

Igor Kozyrev, on 27 June 2015 - 09:57 AM, said:

Hello! I have a question about zoom functionality. Many people say that zoom is host-authoritative, and that means that after pressing button client have to get host response to actually zoom in or out. This behaviour affects people with higher ping very noticeably, will this be fixed?

Yes, this is correct. Zooming and vision modes are server authoritative. There is no plan to change this behaviour at this time. But this is something we can take into considerations for the future.

Bulletsponge0, on 27 June 2015 - 01:54 PM, said:

will these changes fix the bug for Clan ERPPCs where if a second one hits while the paper doll from the first ERPPC is still flashing, then second one will not register any damage?

I am not certain about the cause of this bug, so I cannot say for certain whether the changes will fix or improve this problem. I haven't investigated this bug personally, but it's something I can look into.

[Sarlic]

Thanks Neema!

Your time and quick answers is much appreciated.

[Weeny Machine]

Don't get me wrong, I am glad when the hitreg improves and works properly. However, I am already shivering how low the TTK will be if all (read:most) weapon damage suddenly registers.


I suppose that the game will revolve even more around huge alphas on long ranges. Which will make the gameplay pretty bland



@Neema:
I think it great that you share those information with us. It is also very interesting. So big thumbs up from me!

Edited by Bush Hopper, 28 June 2015 - 02:01 AM.

[Appogee]

Thanks Neema.

Hitreg is my single biggest bugbear with MWO.

Between these changes, and Oceanic/Euro servers, I am hopeful I will get onto a level playing field with other Mechwarriors.

[TyphonCh]

Oh my god. The cyan writing? It is upon us. They ARE real!

Great answers thanks for that

[Kjudoon]

Bush Hopper, on 28 June 2015 - 01:23 AM, said:

Don't get me wrong, I am glad when the hitreg improves and works properly. However, I am already shivering how low the TTK will be if all (read:most) weapon damage suddenly registers.

Yep! If all weapons register, those 50-70 point Alphas walking around will insta-kill most lights in the game. Perhaps this will finally force the end of perfect convergence, my #1 pet peeve.

[latinisator]

Kjudoon, on 28 June 2015 - 05:40 AM, said:

If all weapons register, those 50-70 point Alphas walking around will insta-kill most lights in the game.

Well, aff! This will seperate the true light pilots from those wannabes. I would have more preferred the reimplementation of knockdown.

[Sjorpha]

Kjudoon, on 28 June 2015 - 05:40 AM, said:

Yep! If all weapons register, those 50-70 point Alphas walking around will insta-kill most lights in the game.

Or maybe bad players will still miss all the time or wave their lasers around and keep pretending it's hitreg issues.

Even if being a light might get a bit harder, lights are already super easy to instakill now. The trick to good light piloting is already to not get shot in the first place, and for the people who are good at that it won't matter so much if hitreg improves.

[ShinVector]

Sjorpha, on 28 June 2015 - 07:11 AM, said:

Even if being a light might get a bit harder, lights are already super easy to instakill now. The trick to good light piloting is already to not get shot in the first place, and for the people who are good at that it won't matter so much if hitreg improves.

Hitreg improvement matters...

If risk of getting insta kill gets too high.. More lights will adapt and will switch to mid to long range builds.
Hard to hit a small fast light pecking at you from >400M...

Why ? If your still in a brawler build, you have to depend on the team not being trash.
We do not like waiting for the 'Team' to do some and wasting time.
Longer range weapon provide a further safety distance... Only thing to worry around be VERY good gauss marksmen.

Edited by ShinVector, 28 June 2015 - 07:31 AM.

[Sarlic]

... Or give lights good rolewarfare were piloting lights matters. Give them some good long range capabilities. Give them great rewards for information warfare. Give them something that makes it worthwhile to light and help the team.

But this take a lot of work in account and frankly i don't see this happening anytime soon.

Hitreg is just a part of a underlying problem.

Edited by Sarlic, 28 June 2015 - 07:33 AM.

[ShinVector]

Sarlic, on 28 June 2015 - 07:32 AM, said:

... Or give lights good rolewarfare were piloting lights matters. Give them some good long range capabilities. Give them great rewards for information warfare. Give them something that makes it worthwhile to light and help the team.

But this take a lot of work in account and frankly i don't see this happening anytime soon.

Hitreg is just a part of a underlying problem.

I hear the Turrets might be coming down in Assault.. Hehehe...
Just love pissing the hell out of people forcing them to come defend their base... Before Turrets ruined the tactic.

We will see....

Edited by ShinVector, 28 June 2015 - 07:37 AM.

[Richard Hazen]

Could someone explain in layman terms what this does? Thanks.

[DAYLEET]

This just in, insane Locust/Commando pilot will still kick your ass and you will still blame the game.

[Motroid]

After reading the thread I can't help getting the impression that finally some kind of breakthrough in netcoding has been achieved allowing to include "bone rewind" into HSR all of a sudden. Along with some other minor improvements.
While I really appreciate Neemas feedback on this I don't buy it. He wrote in his first post that they were well aware of "bone rewind" being important to HSR and netperformance but it was somewhat too expensive to realize at the time. Read: too much traffic for the server to handle. Now after optimizing the netcode it's no problem at all.
Now I think this was a bussiness driven decision in order to keep the costs down for PGI. Not because the code "was not ready yet". Some time ago they also increased the intervalls at which the server collects input data from the clients in order to accomodate more users without the need to upgrade the server. (read: invest more money) Netperformance went down the drain with this change. You clearly can't find infos on that so easily. Go figure.
Now with the steam release ahead and regional servers incoming the finally MUST invest some real bucks and suddenly "bone rewind" makes it way to HSR. Just right in time...
Neema, nothing personal here, I really appreciate the work and effort you put in the game and that you reply to us in a very professional manner even on small issues but I don't fall for this "finally we were able to optimize the HSR to include bone rewind" when instead according to my perception it all boils down to money. It's bussiness as usual after all.

And most important these horrible circumstances are about to change and MW:O is finally released in EU as well. Maybe after 19k drops I will finally get to experience a game thats outcome is decided by the players and not some wonky hitreg/lag, ping spikes/jitter and latency variances between 30ms - 300ms.

[Wintersdark]

Motroid, on 28 June 2015 - 08:34 AM, said:

After reading the thread I can't help getting the impression that finally some kind of breakthrough in netcoding has been achieved allowing to include "bone rewind" into HSR all of a sudden. Along with some other minor improvements.

A breakthrough in the HSR algorithm efficiency, not in netcode.

Quote

While I really appreciate Neemas feedback on this I don't buy it. He wrote in his first post that they were well aware of "bone rewind" being important to HSR and netperformance but it was somewhat too expensive to realize at the time. Read: too much traffic for the server to handle. Now after optimizing the netcode it's no problem at all.

No, not too much traffic.

Too computationally expensive. HSR doesn't severely impact network traffic, it's very computationally expensive however. The server "simply" has to remember *everything* that's happened in the last 400ms, then every time someone fires, it has to calculate their relative ping, figure out where that particular mech was that long ago (this is server side), and check to see if it was a hit, and where, and for how long/how much damage.

They could just ask the client, but then we'd have those awesome hacks where people headshot you from across the map - a non-starter. So, the server is basically mathematically rewinding game play a variable amount of time for each player on the map every time he fires. In a brawl, that's happening simultaneously, in real time, for 24 players.

That's very expensive in terms of processing power and memory usage.

Quote

Now I think this was a bussiness driven decision in order to keep the costs down for PGI. Not because the code "was not ready yet". Some time ago they also increased the intervalls at which the server collects input data from the clients in order to accomodate more users without the need to upgrade the server. (read: invest more money) Netperformance went down the drain with this change. You clearly can't find infos on that so easily. Go figure.

In a sense this was a business decision. I suppose they could have upgraded server processing power to provide more headroom to do this, and that would have cost a fair bit - and a fair bit as an ongoing cost. That's a much more difficult decision to make, though.

This isn't a cunning, evil plan to screw people, or to sacrifice gameplay for Teh Almighty Dollah.

Quote

Now with the steam release ahead and regional servers incoming the finally MUST invest some real bucks and suddenly "bone rewind" makes it way to HSR. Just right in time...

Regional servers, if anything, make this less necessary, not more. It'll cut ping times dramatically for many people, making HSR less important.

Steam isn't "right around the corner" either - it's still being talked about, but it's not that soon, so this isn't "just in time" for a Steam launch.

Quote

Neema, nothing personal here, I really appreciate the work and effort you put in the game and that you reply to us in a very professional manner even on small issues but I don't fall for this "finally we were able to optimize the HSR to include bone rewind" when instead according to my perception it all boils down to money. It's bussiness as usual after all.

According to your paranoid musings, at least.

Quote

And most important these horrible circumstances are about to change and MW:O is finally released in EU as well. Maybe after 19k drops I will finally get to experience a game thats outcome is decided by the players and not some wonky hitreg/lag, ping spikes/jitter and latency variances between 30ms - 300ms.

If you have ping spikes and jitter, these changes won't really help you very much, and that's a problem that isn't PGI's fault.

EU servers will probably help a lot for you, though.

[Wintersdark]

Deimos Alpha, on 28 June 2015 - 07:51 AM, said:

Could someone explain in layman terms what this does? Thanks.

You understand what HSR does, right? Currently, to determine if you hit, the server (where hit detection happens, not on your client) rewinds the location of your target mech to where it was when you hit fire, because due to the time it takes messages to reach your PC from the server, and your fire command to travel from your PC to the server, the mech isn't where it looked to you like it was. So, HSR rewinding time by your ping amount allows the server to know where the mech was when you fired at it (thus, where it looked like it was to you), and calculate whether or not you hit from that. This removes "lag shields" where you need to lead your target to compensate for ping (up to how much data the HSR system stores and can rewind).

Currently, this rewinds the location of the mech, but not the position of it's arms/legs/torso/facing. It just moves the mech back. If that mech was twisting, turning, running fast (read: A light being a light) it's legs are probably in an entirely different position, it may be facing a different location, etc. You may have shots into it's back miss entirely because FOR YOU it was looking directly away from you, but because the player was twisting at that time for the server it had already turned 90 degrees to the right. On your screen, this would result in lasers burning it's arm/side torso, for the server it would be a complete miss. Similar for legs, you could hit the wrong leg or none at all, if the mech was in a different phase of the walking animation for the server.

Bone rewind causes the HSR algorithm to rewind *everything*, including animation states, so the server knows not just where the mech itself was, but the relative positions and facing of it's arms/legs/torso/etc.