Leading for ping is ridiculous, Mk.2: An Example

[Krator]

I've been writing about this issue over and over again BEFORE the game was released. Of course "competitive playerbase" told me I need to L2P and if I want realistic bullet travel time I can G..O, because "Death 100ms after getting to cover would ruin this game, because IRL you totally know you've been hit the same milisecond it happens!" - ok, not really, IRL people were known to run for a few meters before realising that they've just received a shot to the heart, so the whole "lag compensation would result in unrealistic gameplay" goes out of the window.


Unfortunately, Red Orchestra 2 isn't realistic, so there is no realism to ruin anyways... PTRD penetrating T34 turret front HEEEELLL YEEEEAAAAAAH.

 

[TheMachine]

Krator, you're missing the point. Rather than there being a delay in the time it takes for your enemy to receive a shot, the enemy doesn't take the hit at all. You have to predict your enemies movements and shoot well ahead of them. (the more ping, the further) That's why shooting bots is so damn hard sometimes, because they make erratic movements. Not only do you have to compensate for bullet time-of-flight, but you have to be a psychic. The closer the enemy is to you, the quicker they move across your screen, and the quicker and more difficult it is to hit them if they are changing direction of movement often.

I'd rather deal with getting shot after going behind cover than be expected to predict the future. To alleviate this, in other games we have ...

ping filtering.

I already uninstalled the game in protest of unfinished game launches.

 

[RJ_MacReady]

The closer the enemy is to you, the quicker they move across your screen, and the quicker and more difficult it is to hit them if they are changing direction of movement often.

And it should be the opposite. It should be harder to hit someone from long distance and easier when he's close.

This, along with players being able to sprint up and down the stairs and slopes with their speed completely unaffected made me stop playing the game until some good soul tweaks it.

When guarding a staircase, leaned to the side while aiming down the iron sights, you have very little chance of surviving a single enemy rifleman attack that suddenly sprints from behind the corner, blitzes up the stairs and nails you with a bayonet. That is, unless you have reaction time of 0.1s, because that's the time it will take a player to pass your screen width. He'll reach you in 0.25s if not in less. I'm using one of Butovo staircases as an example, in a building to the south (with one wall very close to neighbouring building), in vicinity to initial Russian spawn point.

Ridiculous.

 

[u-s-e-r]

I believe the hit detection problems may not be entirely the fault of having no lag compensation, sometimes I feel like 50% of the bullets my trusty PPSH is spewing just vanish into thin air, they dont hit a wall or anything.

 

[=GG= Mr Moe]

I believe the hit detection problems may not be entirely the fault of having no lag compensation, sometimes I feel like 50% of the bullets my trusty PPSH is spewing just vanish into thin air, they dont hit a wall or anything.

Not saying its related, but isn't that similar to ROOST, where they had to slow the tank shells down to maybe half speed because they were going too fast to register hitting other tanks on some occasions? (the shell was already past the tank by the time the game checked for a hit)

 

[Xendance]

Not saying its related, but isn't that similar to ROOST, where they had to slow the tank shells down to maybe half speed because they were going too fast to register hitting other tanks on some occasions? (the shell was already past the tank by the time the game checked for a hit)

That's a problem with fast moving objects when it comes to collision detection in general. Just fyi

 

[leguizamo2040]

This is the only reason I did not buy RO2, we don't have dedicated servers in Argentina, I love this game but sadly I can't play it the way it should be played.

 

[dazman76]

Not saying its related, but isn't that similar to ROOST, where they had to slow the tank shells down to maybe half speed because they were going too fast to register hitting other tanks on some occasions? (the shell was already past the tank by the time the game checked for a hit)

Aye, as nestor mentioned - this is a general problem with high velocity objects in most physics systems, especially those in games. It's because, in real life - when a projectile travels from point A to point B, it occupies every tiny bit of space in-between those points along the way. In computer systems, projectiles move "n" units per tick of the engine - and since ticks happen a finite number of times, and projectiles move at a given velocity - they warp through space, even if the warp is tiny, relatively speaking So there are potentially many points between A and B that the projectile won't occupy - and if object C is smaller than (not as deep/wide/long as) one of these points, no collision will occur. It'll pass straight through it.

Using "delta time" for updates reduces this, but cannot ever remove it. The only way to get closer is to run a much higher tick rate for an engine, or simply run the simulation out of real-time - neither of which works with a game, and is only applicable to full-on physical sim engines

I guess there could be algorithms I'm unfamiliar with to lessen this by estimating or predicting - but those will still have errors given the right scale of projectile/collider, and still mean more processing power requirement.

 

[mattlach]

seriously?


Sorry for the sarcasm. Its not so much directed at anyone other than myself. I vowed to bow out of this (imo) often ludicrous discussion, yet here I am...Shame on me.

There are some obviously intelligent and articulate contributors. I don't know what baffles me more....The willingness to continue to espouse that distance induced latency, network protocol, the current cpu instruction set and game code issues can all be lumped into a quick fix by simply introducing (either client side hit detection) or latency compensation. Or the willingess of people to believe that.


In the meantime for those that can't compensate , I'd suggest (not entering any shooting matches and) finding a well endowed server not running over 52 players. Your play will be fine.

For my sanity, I am officially done....lol.
Enjoy....

Overloaded server induced latencies are certainly not the same as long distance network latencies.

However they do have an additive effect. If the server CPU load issue could be addressed (sortof like I mentioned earlier in this thread with a runtime engine benchmark on the server, limiting player count) a much larger percentage of people would have a server with less atrocious in-game latencies available for them to play on.

It wouldn't solve the issues of how the game behaves for people with bad network latencies, but it should be considered a parallel task that would help with the issue either way.

Two approaches.

1.) Fix server side issues causing more latencies than warranted

2.) Add some sort of hit detection fix for when network latencies actually are high.

2 isn't required for 1, but 1 is probably required for 2, as 2 would probably introduce more server load...

Either way, both are needed and both would improve the game. I would start with 1 and then move on to 2.

 

[Mekhazzio]

There are a few places where they can make gains in server performance. The scripting for artillery, for instance, is a huge mess, and explains why servers near their performance limit tend to choke when artillery is going. Beating down those spikes will be nice and all, but it's not going to impact typical everyday playability. Servers that aren't underpowered for their set player caps would see no gain from increasing server performance, since it's incredibly unlikely that enough efficiency can be gained to increase the tick rate.

Even doubling the tick rate wouldn't be huge, anyway. We're currently looking at 25ms average delay to next tick. Halving that to 13ms would only be a dramatic impact on LAN gaming; not many places in the world are fortunate enough to get internet connections such that 13ms is a significant portion of total traffic time. 10% less leading for ping is an improvement, sure, but it's still ridiculous

 

[Barleyman]

Servers that aren't underpowered for their set player caps would see no gain from increasing server performance, since it's incredibly unlikely that enough efficiency can be gained to increase the tick rate.

AFAIK that excludes almost all 64x servers due to not breaking processing to multiple cores.. Since Ramm-Jaeger has apparently been deposed in a palace-coup, perhaps there is hope for lag compensated shooting yet. Simple and uncomfortable band-aid would be to delay firing animation by lag, apparently the engine supports this.

 

[storPoteten]

I wonder why we havn`t heard anything from Tripwire about the issue? This and the terrible broken sound is what brings the game down.

I had 11 friends buying the game,all of them stopped playing because of:high ping+no lag compensation= unintuitive and sluggish gunplay.

The gunplay is the "meat and potato" in a fps and if bad cooked,no trimmings or side dishes are going to help,the meal is ruined.

So instead of heeding the call of a few Ro:O vets for a "realistic" game mode Tripwire HAS to put all their time and effort fixing the game for us who are actually playing the game

 

[mattlach]

Since Ramm-Jaeger has apparently been deposed in a palace-coup

Hmm?

 

[Barleyman]

According to yoshiro's post on the Jan 30th thread, Strago has replaced Ramm as the RO2 lead programmer. Perhaps he sees leading for ping is ridiculous..

 

[kenijaru]

According to yoshiro's post on the Jan 30th thread, Strago has replaced Ramm as the RO2 lead programmer. Perhaps he sees leading for ping is ridiculous..

Too bad lag compensation and the such are a design decision, and as such, are taken by the lead designer. It's implementation, on the other hand, would be the lead programmer's job.

 

[RageQuit]

I personally think the hit registration should be client sided with bullets traveling at their intended speed no matter what latency you have.

Sure it gives you the whole "I died behind cover" problems, but what's better? Dying a couple times when you shouldn't have in a match, or having to lead for an unrealistic bullet travel time for every single round you fire.

The consistency is nice for everyone in a certain ping bracket in the current system but every player has a different bullet travel time due to their latency. It's not equal and is unfair in every way. The guy with 30 ping will always have it easier when trying to hit targets than the guy with 150 ping and this is simply not equal or fair.

At least with client side hit detection everyone has the same bullet travel time and I personally think the "dying behind cover" issue is completely forgivable as the pros are just too overwhelming versus the cons.

And don't even come at me with the exploitable and hackable crap that can come with client side registration, a good admin will keep that filth out.

 

[Synesthesia]

well, i redownloaded, full of hope, heard most issues were solved, but this problem remains at RO2's core. We never got a single answer from developers. yay for optimism!

Hope some people who've bought since the sale read the thread and add their voice to it. Maybe something will move?

Not all of us have a 40ping connection, this IS a niche game after all.

Anyway, guess i shouldve known better.

 

[captain pain]

well, i redownloaded, full of hope, heard most issues were solved, but this problem remains at RO2's core. We never got a single answer from developers. yay for optimism!

Hope some people who've bought since the sale read the thread and add their voice to it. Maybe something will move?

Not all of us have a 40ping connection, this IS a niche game after all.

Anyway, guess i shouldve known better.

Good news, dude: client-side hit detection is in, and it's apparently the default setting for servers. It's SO much better now.

 

[melipone]

As was mentioned, it is client side by default. Not sure if there are any without client side now. You lead for bullet travel time now only, so you still aim a bit in front of the target but not at a random amount based on ping + distance/movement speed

I would lock the thread so newbies don't get confused and think this old issue is current

 

[Mekhazzio]

This -was- the core problem with RO2, but now, it's solved, as part of the base game, even.