38 replies to this topic

[Roadbuster]

Paigan, on 22 September 2017 - 01:03 PM, said:

Interesting.
I'd say:
- Implement a replay function.
- Gather game replays with the occurance.
- Debug the replay reproduction at the interesting point.
(work smarter, not harder)

There's tons of reasons for a replay function anyway (players sharing games they're proud of, reporting, e-sports, solaris, even MC spending) and now there's one more: debugging the countless bugs in the game.

As much as I'd like to have a replay function, I doubt we'll ever get one.
Well, I guess it's to be expected, considering that they don't even have plans to review hitreg, one of the most important mechanics in MWO.

I've had a funny experience a few weeks ago. I shot a laser ~20-30m in front of an afk assault (afk, not disco)...and hit it.
It was standing there and didn't move and I shot the ground WAY in front of it, and my shot hit the mech...

Hitreg seems to vary alot in general. There are matches where enemy mechs just die in seconds without dealing much damage to them, and the next match the enemy mechs won't die, no matter how much you pump into them.
It's really strange sometimes.

Edited by Roadbuster, 23 September 2017 - 01:01 AM.

[Exilyth]

Being an e-sport and having hitreg issues are pretty much mutually exclusive.

[Asym]

Bobby Jubraj, on 22 September 2017 - 03:14 PM, said:

Yeah my answer here would be the same conclusion Neema provided, which is if you guys run into this behavior in a match, send over a short video of this to support so we can log this with the right folks to look into.

The more videos we get, the more information can be collected and the greater chance that the priority can be bumped up

Well now, I haven't seen or read the "get GuD" response from you'all.....since I've been writing about this issues for months and some of you have said "you're a potato and get GuD" Un Huh, this is a potato issue........

No, it's a precision issue in a fluid battlespace that is complicated by hundreds of nerf, balance and quirk errors that have and are compounding mathematically..... Go out on polar and take a 2x Gauss and try headshots at different ranges on the various sizes of mechs......what do you think you will find? As crticial hit space gets smaller and the amount of movement increases, the accuracy of a non-line-of-sight weapons degrades due to error.... To the point of going through the target without registration. We see this same effect in aviation avionic software all of the time....... As precision increases in a 3D model, aspect change becomes problematic if the stationary calculations are flawed from the beginning.... As little as .001 degree of pitot static probe error can rended a autopilot system so flawed that the AC can't operate in a RVSM corridor..... Compounding precision errors in gameplay are the same thing.... If the Gauss rifle has been altered from it's day one condition and the target has been altered in survivability let's say, the translation errors will compound........you'll miss, you'll pass through, you'll lag and the hits will register somewhere else.....etc....

Precision isn't easy and many game systems simply can't keep up and the list of compounded errors will never be detangled: there isn't enough money or time to ever get to the root cause even identified, let alone fixed...

So, when 5 larger mechs are shooting at an Urban mech and you are wondering "how in the heck is that light" even slightly alive? Because a lot of you ammunition isn't actually hitting the mech from the calculations being performed. Even if you are "GuD' and you know the gun sight is where it belongs.......

[Exilyth]

It's always a tradeoff between precision and performance.

There are programming libraries like gmplib (https://gmplib.org/) which provide near arbitrary precision, but the additional calculations required make it unsuitable for real time applications.

[Asym]

Exilyth, on 23 September 2017 - 08:30 AM, said:

It's always a tradeoff between precision and performance.

There are programming libraries like gmplib (https://gmplib.org/) which provide near arbitrary precision, but the additional calculations required make it unsuitable for real time applications.

True, especially in games. Unfortunately, precision is a requirement for MWO...... If not, why have any weapons that fire beyond 500 meters?
Having actually worked with this type of data for real stuff: no, because if you screw something up and train without precision, well, the results are not a good thing, eh? Aim small, miss small actually means somthing......

The optimal question for MWO is: is it value added to even attempt to fix the precision issue? Which, IMO is not ever going to happen because of where MWO is moving to: 2D brawling arena shooting at mid to close ranges...
If PGI wants to fix Faction Play or/and even QP, Oh yes, that would be a neat fix and make the games a lot more relevant and exciting.... No more Urban mechs leg hugging in the open water as 5 other heavier vehicles pound away at that one light mech and can't, no matter what they do, kill it....... That would change because they are actually hitting that Urban mech and it's not being calculated towards damage because of the errors in the precision calcutaions are so great, a majority of that ordance is not even counted as hitting the mech...... Ghost shots with PPC's that pass through. Gauss rifle shots that seem to hit but nothing happens.... Effective weapons are the LBX series weapons because of the number of sub-munitions that some how change the area damage calculations? I've seen a MCII watch a light mech get clobbered by his team mates and the MCII pilot waited till the light cleared the leg hugging and he hit it with a -20LBX and significanlty damage the same mech the others have been shooting for 30 seconds or so with no color changes????

What kind of game do you want? A game that says "what can be seen can be hit but, well, we're not sure if that hit meant anything" or, a game that is "what can be seen can be hit; and, what can be hit can be killed?"........

[Dr Cara Carcass]

Regarding shots registered in the back while actually beeing fired from the front...

I always had a feeling that it had to do with the refresh rate of the server, since i play gauss extensivlely and get these shots against reds more often than other from my unit, while reciveing them equally often.

The gauss makes 2000m/s, which means when polled with 30Hz points along the trajectory of the bullet are (2000m/s)/(30/s)=66,6m. Which means further apart than a mech size. so the last point in front of a mech and behind a mech can be used to draw a line trough them and find out where the hit must have happened.

What if the last point before the mech and the first behind the mech are shifted due to lag in a way that the last point before the mech ends up inside the mech and the first one behind it is still behind the mech. The line between both points along the trajectory cuts the back but not the front.

But yeah no idea how the code finds the hit zones, just guessing here.

Edited by Cara Carcass, 23 September 2017 - 02:09 PM.

[Exilyth]

Talking about capturing gameplay videos, can anyone recommend a good video capture program?

[justcallme A S H]

OBS or if you have NVidia, just use the inbuilt software

[Exilyth]

justcallme A S H, on 24 September 2017 - 05:30 AM, said:

OBS or if you have NVidia, just use the inbuilt software

Nvidia shadowplay, or what was the name of the nvidia software?
What is OBS? Could you please provide a link?

I'm on AMD btw.

[Bulletsponge0]

Wait, the devs are saying that hit reg isn't a priority? in a first person shooter?

[Shadowomega1]

Bulletsponge0, on 24 September 2017 - 09:57 AM, said:

Wait, the devs are saying that hit reg isn't a priority? in a first person shooter?

Because they are having issues reproducing it, if you can't track down the root cause how do you expect to work on it?

[Appogee]

Shadowomega1, on 24 September 2017 - 10:07 AM, said:

Because they are having issues reproducing it, if you can't track down the root cause how do you expect to work on it?

You play your own game often enough to experience what your players are experiencing.

I had a situation two nights ago where I put five 48 point Alphas into the CT of a Bushwhacker and didn't even turn his armor red, let alone kill him. He wasn't even moving much when this happened.

His ping was 20, mine was 250. But HSR is supposed to address that disparity.

This happens at least once a week when I'm playing. Based on this thread, I'm recording all matches now and will send examples to PGI next time I see the problem occur.

Edited by Appogee, 24 September 2017 - 01:06 PM.

[Shadowomega1]

Appogee, on 24 September 2017 - 01:05 PM, said:

You play your own game often enough to experience what your players are experiencing.

I had a situation two nights ago where I put five 48 point Alphas into the CT of a Bushwhacker and didn't even turn his armor red, let alone kill him. He wasn't even moving much when this happened.

His ping was 20, mine was 250. But HSR is supposed to address that disparity.

This happens at least once a week when I'm playing. Based on this thread, I'm recording all matches now and will send examples to PGI next time I see the problem occur.

Yet do you have the Match ID number? Video only shows things that are happening in the forefront, not what is going on under the hood. Seeing what is going on in the code is what truly valuable as you can see in what part or parts is causing the failure. From my experiences of hit registration failure points to a multi-point failure in the code, which means that it could be in the engine itself or in buried deep in PGi's tacked on lag comp system.

[Appogee]

Shadowomega1, on 24 September 2017 - 01:43 PM, said:

Yet do you have the Match ID number? Video only shows things that are happening in the forefront, not what is going on under the hood. Seeing what is going on in the code is what truly valuable as you can see in what part or parts is causing the failure. From my experiences of hit registration failure points to a multi-point failure in the code, which means that it could be in the engine itself or in buried deep in PGi's tacked on lag comp system.

Perhaps they could add the unique match id to the end score screen. (We see that number in our event match lists.)

[blaqBox]

Its good to see that this is a genuine issue on PGI's radar.

Playing faction warfare most days, with a ping of 230 - 250 has made me proficient at leading my shots against fast moving targets, I can hit them 95% of the time, no problem there. Slow(er) moving targets are far less of a problem as I can single out a torso for an XL check kill or CT core them after a few alphas.

I am also used to seeing damage being displayed on the enemy paper doll 200-300 ms after HSR has performed its calculations and sent my client the hit registration feedback. No problems there I understand how network and processing latency effects my in game experience and I can compensate.

What I have found really frustrating lately is the lack of damage transfer when hitting player mechs in a pre-made team. Its not every mech in the pre-made, some suffer damage in the usual way, while others seem to just absorb most or all of the damage received. It seems to be more noticeable when there are more than 6 players in the pre-made and the players have a mix of high and low pings.

At the moment its just a very frustrating observation, but I'll try and capture some games and send them off to PGI.

[justcallme A S H]

Pre-made or 12 solos makes no difference.

It cannot as those players are treated as solo's in the MWO server level. The fact they are in a group doesn't affect anything as it's Client - Server. Not Client - Server - Group - Client or any such.

[blaqBox]

It is making a difference to my gaming experience Ash.

I agree that each mech should be treated as an individual independent object in game, with their own hit registration and damage application algorithms inherited from the same template.

But the reality is, many players have observed this damage absorption behavior in different game modes. If this is can be repeated and observed using known parameters like a pre-made team of mixed pings, it may provide some clues.

Its not up to us to fix, just provide some reproducible evidence.

[justcallme A S H]

I'm not saying people are not experiencing it.

I'm just saying that a large group being treated differently by MWO servers, just isn't the case.

[Arnie1808]

Had a horrible time last night getting the IS gauss mission done. Had a King Crab with a light gauss in each arm and it took me a lot longer than it should have because my shells just travelled right through the mechs I was firing at. I had a Mauler standing still about 800 meters away and I could see the shells hit the ground behind him on the fourth salvo they hit and he realised he was getting shot and moved but that means six shells past right through him!
Iv had this happen a lot with ppc's as well you can actually watch them go right through and hit whatever is behind the mech your firing at .
Also I have the Supernova Boiler and I've got dual LBX 20's I fired 3 salvos into the left side of an Urbanmech and he just ran off he lost the arm but just merrily buggered off down the other side of the slope and ran away is this another problem of the hit register or something else because he was less than 100 meters away?