PDA

View Full Version : [00276][R#178] Mob damage rebound



Dribble Joy
29-01-13, 18:07
This is something that has been around for years, and never seemed to happen back in the day and I'm not sure exactly when it began.

As many of you know, mob dmg nearly always bounces back. You only take dmg at what seems like random, without any action by the mob. This also make melee mobs almost harmless, only ranged mobs seem to actually do anything unless you stand near a melee mob for a good period of time.

This makes many low level mobs no threat to a fresh runner. I remember when a small sewer rat as actually something to be wary of as a 0/2. Not you can stand next to one (or several for that matter) and swing away.

Is this something to do with the client and server not being aligned? It's as if the client does one thing, which is then rejected, and the server says another, which isn't acted on by the client except for the dmg. If the server says that a mob makes an attack, shouldn't that make the client perform the animation/effect as well as the dmg and the client not to do anything by itself?

Netphreak
29-01-13, 18:25
This is something that has been around for years, and never seemed to happen back in the day and I'm not sure exactly when it began.

As many of you know, mob dmg nearly always bounces back. You only take dmg at what seems like random, without any action by the mob. This also make melee mobs almost harmless, only ranged mobs seem to actually do anything unless you stand near a melee mob for a good period of time.

This makes many low level mobs no threat to a fresh runner. I remember when a small sewer rat as actually something to be wary of as a 0/2. Not you can stand next to one (or several for that matter) and swing away.

Is this something to do with the client and server not being aligned? It's as if the client does one thing, which is then rejected, and the server says another, which isn't acted on by the client except for the dmg. If the server says that a mob makes an attack, shouldn't that make the client perform the animation/effect as well as the dmg and the client not to do anything by itself?

I've mentioned something similar to this before, were I will be hit by a mob and my health drops say 600 to ~50 for a second, and then bounces back up to the level it should be after the damage has been calculated (lets say it worked out the mob did 150 damage to me, so my health bounces up to 450).
However if the mob does 2 attacks in quick succession I will die. Even though each attack should only be doing 150 damage (which should leave me at 300 health), the second attack hits me when my health is sitting at 50 before bouncing back and I die.

It definitely seems to happen much more than it used to, and I agree that melee mobs seem to be affected bay attacks seemingly not registering/hit you as if the server thinks that your character is not beside the mob or it hasn't caught up position wise yet.

This is something I will be trying to test on the test server when I get on later tonight.

PsiCorps
29-01-13, 19:20
I don't believe this was in NC1 untill the anti-oneshot mechanics were in place.
Ever since then, it's been screwy.

nabbl
29-01-13, 20:25
Can confirm that. Problem exists with Small Spiderbots also. Their attack does 200 damage. but 180 damage gets neglected. But if you are at 200 HP or less their shots are insta kills.

hatmankh
29-01-13, 20:49
This has always annoyed me since I started playing Neocron, I never knew it didn't happen in NC1, I thought this was always here and always will be. Needless to say, I hope it gets fixed.

Dribble Joy
29-01-13, 22:17
Sometimes there's no dmg at all after the bounce. Warbot beams, doomreaper/chaser beams, mauler/perci blasts, even spiderbots. All take dmg off but bounce back with nothing lost at all.

William Antrim
30-01-13, 00:36
This problem was in nc1. At the time it was explained as resists kicking in by people who I thought had mire knowledge than me. I have therefore never honestly questioned it but I have to agree with you guys. If it got fixed I'd be well happy.

Dropout
30-01-13, 01:22
This problem was in nc1. At the time it was explained as resists kicking in by people who I thought had mire knowledge than me. I have therefore never honestly questioned it but I have to agree with you guys. If it got fixed I'd be well happy.

In the current state of the game, I wouldnt..
I hope that they balance mobs (any everything else, but I start to sound like a broken record here hehe) before fixing something like this.

L0KI
30-01-13, 13:40
Regeant's is a prime example of this. The parasites can hit you constantly and the HP just bounces back up. It's only when you get a couple of contamination stacks on you that it actually causes issues.

Also, Grim Persecutors. They hit really hard with what seems to be burst force damage. My HP can go from 800 -> 200 -> 780. If I happen to be on 500 when they hit, it's insta-death.

Netphreak
30-01-13, 14:30
Regeant's is a prime example of this. The parasites can hit you constantly and the HP just bounces back up. It's only when you get a couple of contamination stacks on you that it actually causes issues.

Also, Grim Persecutors. They hit really hard with what seems to be burst force damage. My HP can go from 800 -> 200 -> 780. If I happen to be on 500 when they hit, it's insta-death.

This. ^

Makes farming on my spy somewhat tricky when I die from a mob hit that shouldn't have killed me.

Kame
31-01-13, 16:59
This bug actually makes PVE manageable. If not from that, PVE would be A LOT harder. Like 4x harder.

Deus Ex Machina
31-01-13, 21:30
This bug actually makes PVE manageable. If not from that, PVE would be A LOT harder. Like 4x harder.I don't think the bug is not that it bounces back, but that it first goes down so far. And sometimes kills you in the process. So it would actually be easier without the bug.

demonssword
03-02-13, 01:35
Endlessly see this happening, espesially in hacknet. The Phoenix CPU doing 300-400 dmg a hit sometimes then it bounces and turns into about 20. This isnt to bad as his attack timer is aweful and as a 127/127 he cant even kill someone with the 40-50int hacknet stuff.
However the mobs like the "Ping of Death" mobs that are 124/124 they hit for about 300 or so and normally sticks about 100ish after the bounce and can shoot 2-3 times in a second some times for no apparent reason.

pottburter
06-02-13, 13:55
Hi,

i just was outside and tried my re-specced Spy on fire mobs (just increased RC and WPL) and noticed the "health lag" to be worse than in R178.
I'm talking about the "lag" where you get damage and shortly thereafter your health bar jumps back to full/more health.

In R178 I took a hit from a Persicutor and my health jumped back after about 1 second.
Now I battled with one of these beasts just now and noticed it takes about 2-3 seconds for my health to re-adjust.

Anyone else noticed this?

Cheers,

William Antrim
22-01-14, 16:34
Had this issue in the last two days, fighting against mad copbots at battledome and they take small amounts of health off me whilst I heal back up. It seems that gatlin bursts from their guns only do a tiny amount of damage per hit but the rof is enough to give the impression of constant damage.

Warbots and spiderbots however (with a totally different attack and different ROF) have killed me. Health starts to go down, get into the red and die as normal or go down to 0 from a particularly crunching hit only to go back up to 14 or so (sometimes more this is just an example) a few seconds later.


Strangley hasnt happened in pvp though - at least not that I noticed in the last few days.

Divide
22-01-14, 17:57
Health bouncing has existed in some form since beta. It is my understanding that the client and server both perform resist checks & damage calculations. The problem being that the two don't align, even though they both have the ability to deeply impact your situation. Broken legs that aren't broken, amirite?!?!

If you do happen to die due to a calculation issue where a mismatch has taken place, if the server believes you are still alive, you can relog and your char will indeed be alive.

This is legacy behavior from the days when compute, latency, and bandwidth demands were insurmountable for everyone but industry leaders like IBM & MS. You had to give the client some amount of responsibility when your pipe was 10megs at absolute best and your entire server farm was built on Willamette or Northwood-based Xeons that were great at heating up a room but left a bit to be desired otherwise.

Netphreak
23-01-14, 03:03
For me, this issue is currently worse than I ever remember it being and really impacts every aspect of the game play/experience.
As the poster above pointed out one of the annoying things that happens due to this is your legs will break from a hit that shouldn't have broken then thus making you not able to run to cover. If you combine this with taking a hit which takes 500+ of your health (which after the bounce will have only taken
less than 100), you get crippled can't get out of the way for the next volley which hits before the client has corrected itself (the bounce) which once again hits for 500+ damage (even though it should be less than 100) you die. No chance of escape, nothing to do with bad resists or anything, purely due to the delay/bounce.

While I appreciate that neocron is running an ancient engine, surely there is a way to drastically reduce the delay /client updates from the server.

Outside of weapon and mob damage balancing this is something that really holds back the game potential.

Doc Holliday
23-01-14, 04:31
Has happened to me twice in regants legacy. On a tank wearing full kevlar armour and Camo PA and also a PE wearing the same. The genotoxic crawler comes over and lays a hit on me. My health plummets to 30hp or so (enough so 2 sections of the body are fully damaged and health is in red) and stays there for a few seconds before bouncing back up to a full amount or reasonably close.

Both characters have a minimum of 40 in poison specced in natural resists and then all the armour calculation on top.

William Antrim
23-01-14, 13:07
Would it help if more mob damage was over time and less burst damage? to give the server chance to correct it I mean?

Netphreak
23-01-14, 14:55
I think the client update rate really just needs to be somehow increased.

William Antrim
23-01-14, 20:56
I am sure our pcs could handle it.

Divide
23-01-14, 21:55
I am sure our pcs could handle it.

The sustained data rate for the NC client is well below the real-world speeds that a 56k modem could deliver. We could definitely handle it.

Netphreak
24-01-14, 01:03
The sustained data rate for the NC client is well below the real-world speeds that a 56k modem could deliver. We could definitely handle it.

There are commands/values in the net.ini file
//maximum bandwidth for each client
set net_maxclientbw 2000

From a quick google search it appears that this related to 2000 bytes bandwidth.
I'm not sure is this command even does anything or if it is all controlled server-side now but 2Kb is less than 56K could provide.
I'm rather curious if this value has been increased to be more inline with the bandwidth of internet connections we now have.


There's also this other value/command
//maximum time between player full updates(in secs)
set net_playerupdtime 2.0
Which I hope isn't used any longer as that allows up to 2 seconds between player full updates which could be one of the reasons for the health bounce issue.

Alduin
24-01-14, 12:47
The net.ini file is mostly obsolete, because the settings are enforced code-wise. Increasing the update rate does not solve the problem itself it just makes it less visible in the best case. On top it increases the load on the server side by a non negligible amount. The relation between between update rate and cpu load is not linear.
Interpolation and prediction are still done in every networked game. In fact: most of the information has still to be predicted by the client, even in the age of mbit lines. The issue with bouncing help is that client and server predict something different.
We already started to adress the issue of prediction, but fixing it will take quite some time, there are simply far too many interdependencies.

Netphreak
24-01-14, 15:45
The net.ini file is mostly obsolete, because the settings are enforced code-wise. Increasing the update rate does not solve the problem itself it just makes it less visible in the best case. On top it increases the load on the server side by a non negligible amount. The relation between between update rate and cpu load is not linear.
Interpolation and prediction are still done in every networked game. In fact: most of the information has still to be predicted by the client, even in the age of mbit lines. The issue with bouncing help is that client and server predict something different.
We already started to adress the issue of prediction, but fixing it will take quite some time, there are simply far too many interdependencies.

As always thanks for taking the time and commenting on this Alduin.
It's good to have to clarification and to know that you are well aware of the issue and have already been looking into fixing it.

William Antrim
25-01-14, 11:56
This problem can be reproduced every time you exit an unsuccessful hack.

Dropout
25-01-14, 18:43
This problem can be reproduced every time you exit an unsuccessful hack.

Uh yeah, thats true!
Man.. There's so many bugs in this game I dont even notice anymore...