53 replies to this topic

[Abrahms]

Checked in for my 6 month follow up to see where the game is at. Game is shaping up a lot, but boy is that lag shield still alive and well.

See light mechs doing 1k dmg in a game... as they run around eating AC20 and large lasers like nothing. Sparks, hot spots, fire, smoke... but no damage.

Big bummer for sure.

[FupDup]

If you step into the cockpit of one, they won't seem so "invincible" as they do when you're on the shooting end.

[SpeedingBus]

FupDup, on 11 December 2014 - 11:38 PM, said:

If you step into the cockpit of one, they won't seem so "invincible" as they do when you're on the shooting end.

Tell that to NS who drop all in lights and manage to trade kill for kill against much heavier mechs.

[Abrahms]

SpeedingBus, on 11 December 2014 - 11:46 PM, said:

Tell that to NS who drop all in lights and manage to trade kill for kill against much heavier mechs.

Speed and lag shield >>> armor. Not only will damage naturally spread around the mech, but most of the damage wont even register.

Its much easier to kill an atlas. Hits far more likely to register, and much easier to focus on one component. Add it all up, and a light mech can often take a much heavier beating.

[FupDup]

SpeedingBus, on 11 December 2014 - 11:46 PM, said:

Tell that to NS who drop all in lights and manage to trade kill for kill against much heavier mechs.

That requires either a very good light pilot, a very bad heavier mech pilot, or a mix of both. Assuming all other factors are equal (such as pilot skill), the light is going to be at a disadvantage in a straight fight, barring cases like the heavier mech boating LRMs or something.

[SpeedingBus]

Abrahms, on 11 December 2014 - 11:47 PM, said:

Speed and lag shield >>> armor. Not only will damage naturally spread around the mech, but most of the damage wont even register.

Its much easier to kill an atlas. Hits far more likely to register, and much easier to focus on one component. Add it all up, and a light mech can often take a much heavier beating.

Also they focused your legs the lightest armored thing on a mech usually. It wasn't fun to say the least between lag shield and random FPS chugging.

[YueFei]

Abrahms, on 11 December 2014 - 11:32 PM, said:

Checked in for my 6 month follow up to see where the game is at. Game is shaping up a lot, but boy is that lag shield still alive and well.

See light mechs doing 1k dmg in a game... as they run around eating AC20 and large lasers like nothing. Sparks, hot spots, fire, smoke... but no damage.

Big bummer for sure.

You realize that 35-ton Light mechs are supposed to be able to take an AC20 to anywhere on the front and not even have armor penetrated, right? It can take 2 AC20 hits to the same leg and not lose the leg.

The top players and top teams have no issue hitting and killing Light mechs.

Your issue is either your internet connection, or your aim.

[Karamarka]

Lag shield is massive in CW right now

everyone has 5-30 fps

and the hit reg is horrid for lights

[EvilCow]

This may surprise some players but LagShield != Skill

Right now the game is hugely skewed if favour of light mechs by the bad hit detection and weapons designed to spread damage as much as possible. Heavy weaponry simple does not have enough stopping power vs fast moving targets.

In CW soon you will see just firestarters because there are no classes restrictions (and the problem is not the lack of classes restrictions)..

[Mcgral18]

EvilCow, on 12 December 2014 - 01:08 AM, said:

This may surprise some players but LagShield != Skill

Right now the game is hugely skewed if favour of light mechs by the bad hit detection and weapons designed to spread damage as much as possible. Heavy weaponry simple does not have enough stopping power vs fast moving targets.

In CW soon you will see just firestarters because there are no classes restrictions (and the problem is not the lack of classes restrictions)..

I have my doubts...what's the light queue at again?

[The Boz]

FupDup, on 11 December 2014 - 11:38 PM, said:

If you step into the cockpit of one, they won't seem so "invincible" as they do when you're on the shooting end.

I have a few lights, all are going 150+, and I definitely do have moments that I survive, and I probably should not have.

[Karamarka]

12 IS lights can win community warfare in less than 3 minutes

[Kneipenboxer]

Very bad lags and low fps in CW in my opinion.
I pilot assaults and lights alike.
CW is fun but this is defintly a problem and at same time an advantage for light mechs, being much smaller and faster.
nobody should doubt that.

hopefully pgi can fix this.

have fun y'all

[AlexEss]

Yes hitreg get a bit wonky every now and then... Especially now when the servers are under heavy load. But at the same time it is worth noting that it is still better then it used to be and lights are not even close to being as immortal as they used to be.

But i am also sure they will take another stab at it once CW is running stable Until then i guess you will have to buld a mech or two specificity for hunting lights with a lot of crit seekers and SSRM.

[Mike Oakenwall - the khadoran]

Not sure,
I one shot a locust with dual ERPPC yesterday, dont think he was really damaged before.

[Cyborne Elemental]

Hitreg is very inconsistent from player to player just in general, it probably has alot to do with connection quality in that respect.

I've yet to see a multiplayer game that actually has flawless hitregistry, BF4 comes close IMO and at least is better than the majority of modern FPS shooters that I have played.

I've always had issues with Unreal engine games, whether it was Borderlands 1&2 (which has no interpolation/lag compensation code at all) or Killing Floor, or any of the Unreal Tournament series.

Halflife1&2, 2 being on the source engine which runs games like Left-4-Dead 1&2, Counterstrike:S & CS:GO, or Team Fortress2, is hands down one of the sloppiest examples of multiplay code I've ever seen.

Now me personally, I get 60-65 ping to the MWO servers, and I've tested my connection and get very little packet loss usually.

I always shutdown background programs and have nothing else running or occupying bandwidth while I'm playing, and for the most part, it doesn't matter what speed class of mech I'm playing, people seem to consistently register damage on me with high precision.

Now at the same time once I pass that 120 kph threshold in game, I do seem to notice that damage hitting me starts to do less and less, I hear AC/Gauss/PPC and especially Laser hits impacts but either see no damage come in, or very little at all.

There is definitely something being broken with hitreg once you hit a certain speed, but its so inconsistent from player to player that for the most part its just confusing as hell and is very frustrating.

How can one player be taken out quickly and reliably as long as your aim is true, but another player under similar circumstances has this magical lag shield that can absorb nothing less than a nuclear explosion and walk off scott free?

Seems like as long as your target is moving, under those circumstances, hitreg just goes bonkers.

But if he sits still or has to slow down alot for a sharp turn, suddenly hitreg is amazing.

Shrug..

[Triordinant]

Mister D, on 12 December 2014 - 03:21 AM, said:

Seems like as long as your target is moving, under those circumstances, hitreg just goes bonkers.

But if he sits still or has to slow down alot for a sharp turn, suddenly hitreg is amazing.

Shrug..

This sounds like a good reason for never introducing MASC in the game unless they change the game engine or do something else to fix hit reg.

[The Boz]

Brief explanation: I think server refresh rate is back at 30. This doesn't have to be the case, though...

Here's why lag shield is a thing: a 150 kph Firestarter is doing 41.7 m/s. In a 30 FPS (not the same F as in your personal FPS, but called the same) server, that's ~1.4 meters per frame. Problem exists because of the shooter's lag (the target's lag is almost entirely irrelevant in a server-authoritative system like MWO's). You are always at least one frame behind the server, and often more (divide one second by the number of server frames, then divide your ping in seconds by that number, and then round up; that's how many frames you are behind the server, minimum, sometimes +1).

A player with a ping of 100 to a server running at 30 FPS is at least four frames behind. The Firestarter mentioned before can move ~5.6 meters in those four frames, which is considerably more than that mech's width. He is literally not there where you see him.

This is where HSR comes in. HSR works by sending your current ping to the server with each command you send. The server then goes a number of frames back, depending on your ping, and sees if there was anything you were supposed to hit in that frame, attempting to compensate for your ping.
However, there can often be a discrepancy, especially if your ping fluctuates. First, the server goes back by twice the number of frames that your ping corresponds to, because let's not forget, ping works both ways. If what you see is four or five frames behind of what is actually going on, and you act on it, your action will reach the server at a time when that is between seven and ten frames behind.
And there is no easy way to determine if the server is supposed to go back seven, eight, nine, or ten frames behind. If it errs on the side of caution (and, although I don't know the details of PGI's implementation, it is supposed to), it will take the statistically most secure value, which is eight frames in this case. If the chosen value matches the "experienced" amount of frames of delay, you hit if your aim was good enough. Some shots that could have hit by a hair's breadth would sometimes miss, but that's life.
Lag shield happens if you experienced either seven or nine frames of difference, and aimed "good enough to hit", but not in the center of the target, or if you were aiming good enough for the dead center, but the actual difference was ten. Remember, that Firestarter moves almost three meters in two frames, fast enough to entirely clear it's own hitbox from two frames ago. And this is before ping fluctiation comes into play.

When the server is running at a higher refresh rate, the lag shield problem is much less pronounced, because the difference between each individual frame is smaller. Edge cases still do occur, and ping fluctuation is still a problem, but the experience is much better in something like 80% of the cases.

So, the components of lag shield are server-side refresh rate, target's speed, shooter's ping, and target's profile from the shooter's perspective. This is why Ravens are much easier to hit if they're going 150 than Spiders are; their bodies are longer and fatter.

There, hope that clears it out somewhat.

EDIT: Paragraphs!

Edited by The Boz, 12 December 2014 - 03:57 AM.

[EvilCow]

The only way to have a reliable hit detection is to move it on the client side and use HSR on the server side to validate the information sent by the client using a time window rather a definite point in time for verification, the time window should be dimensioned to account for server FPS jitter and connection jitter too (within reasonable limits, like 10% fluctuation).

Using this model, the client always makes the damage as seen but it is the server to be authoritative and decide if the client is sending faked info.

The model could be applied to lasers only, weapons with travel time work already decently well in the current implementation.

[Cyborne Elemental]

I think you're right there Boz, I'm having to lead targets again to even get hit confirmation.

Just like when the servers were running at low FPS.

I wonder why that keeps getting reset on their end..?