32 replies to this topic

[Nesutizale]

Just tried out a build with 3 IS-ER-PPPCs and noticed something strange in the testing grounds. Even while standing still and not moveing the crosshair I got different damages for chainfire and alpha.

I shot the arms of an Atlas on tourmalin. Counting how long it would take to get through the armor and while I needed the same amount of shots for both arms they where different shades of yellow between chain and alpha fire.

Also during matches I have the feeling that sometimes alpha shots of PPCs that hit the same location don't do the damage they are supposed to do? It feels weeker then chainfire and try to hit the same location again...also that is harder to do.

Anyone else noticed some strange behavior like that? Note that it only seam to happen when the PPCs are build very closely together while on builds where they are, for example left arm, right arm it dosn't seam to matter?

Never had that problem with Laser or other weapons.

PS: Anyone else also bothered by the PPCs giang "hitbox"? Shooting over terrain with any other weapon is fine but PPCs constantly hit terrain where other weapons don't.

Edited by Nesutizale, 28 April 2019 - 03:39 AM.

[Bulletsponge0]

PPCs, of all varieties, have had hit registration issues for years and years. unfortunately, PGI has shown no interest, or simply lacks the ability, to fix those issues.

[Bloodwitch]

Goes for all weapons. You don't notive it as much witch weapons which tick like lasers. More so with mult-hit weapons such as missiles or UACs.

With flpd it's noticable. especially bigger calibers / cooldowns. PPC, AC10/20, Gauss.

Sadly, it's just the way it is. On the bright side, everyone deals with it.

[Steel Raven]

I've been back shot in my Griffin while facing a enemy, I can only figure a round hit me in the *** while torso twisting. The Cryengine particle effects and MW hit boxes and hit reg are not the best mix, Toss in that some PPCs have splash damage because god forbid we have a PPC that does more damage than a AC/10 and you get some WTH? results in game.

[Ssamout]

Also for testing, the testing grounds aint the best spot. Better to do stuff like that in live environment using a private lobby.

[Moebius Pi]

Steel Raven, on 28 April 2019 - 07:20 AM, said:

I've been back shot in my Griffin while facing a enemy, I can only figure a round hit me in the *** while torso twisting. The Cryengine particle effects and MW hit boxes and hit reg are not the best mix, Toss in that some PPCs have splash damage because god forbid we have a PPC that does more damage than a AC/10 and you get some WTH? results in game.

I actually stopped using my Orion IICs outright because I'd get back cored from the front repeatedly on release; something to do with the ballistic nipple it seemed. Never did find out if they fixed that, if it was one variant bugging or what the hell but it was so frequent without anything firing from behind, I just gave up and considered it "PGI'ed"; screwy, but unlikely to see a fix like numerous other persistent hit reg and other issues.

Though, I was gone for a good while... did they ever mention it in the patch notes?

[Nesutizale]

I din't encounter that problem with my Orion IICs or I just never noticed.
The PPC bug I noticed earlier butt never bothered testing or posting until now. Good to know that I don't imagne things.

[KodiakGW]

Bulletsponge0, on 28 April 2019 - 06:53 AM, said:

PPCs, of all varieties, have had hit registration issues for years and years. unfortunately, PGI has shown no interest, or simply lacks the ability, to fix those issues.

Yep, that and the fact that PPC shots get absorbed by tree leaves is why I never bother with them anymore.

[panzer1b]

Virtually every weapon in the game has hit reg issues, albeit its less bad with rapid fire ones like lasers or clam UACs that spit out alot of pellets (if a few dont connect big deal). Still, ivde had entire 60 point alfa strikes from laser vomit mechs do no damage so its not inherent with PPC.

That said, ppcs are so hot and hard to use on most mechs that it stands out way more as a PPC shot going into thin air is alot more problematic then anything else which you can just fire again 2 seconds later. I think PGI should make all PPCs slightly cooler running as PPFLD is just not worth such a heat penalty right now...

[Nesutizale]

I would allready be happy if the PPC could be shot everywhere I could shoot an AC or Gauss without hitting terrain or trees.
Yesterday I had the situation again with my Nighstar. The arms are freaking far appart and I stood near a hill. The arms that looked like its covered by the hill could fire its Gauss but the PPC directly under my cockpit couldn't.

Thats..... why is that even happening? Is it because the graphic of the PPC is bigger then the one used for the Gauss? I would gladly drop the PPC graphic if it would stop hitting stuff.

[LordNothing]

to maximize ppc hit detection i find its better to chain fire them. just think of them as uacs and double tap.

[LordNothing]

Nesutizale, on 28 April 2019 - 11:42 PM, said:

I would allready be happy if the PPC could be shot everywhere I could shoot an AC or Gauss without hitting terrain or trees.
Yesterday I had the situation again with my Nighstar. The arms are freaking far appart and I stood near a hill. The arms that looked like its covered by the hill could fire its Gauss but the PPC directly under my cockpit couldn't.

Thats..... why is that even happening? Is it because the graphic of the PPC is bigger then the one used for the Gauss? I would gladly drop the PPC graphic if it would stop hitting stuff.

graphics usually dont dictate collisions. hit boxes do. with the polycounts used in modern games its hard to do polygon accurate hit detection. i know it was a thing in the 90s, i know descent 3 had it. but hit detection being a big n^2 problem the last thing you want is to add complexity. i have a hunch that the collision model for the ppc is either too large or too small. to large and it hits things it shouldnt, to small and it can be hard to detect it going through a thing. hit detection is ******* hard and i feel sorry for any programmer who has to write that code. in pgi's case i think they just use the existing system built into cryengine. and you can misuse a system if you don't understand it well.

Edited by LordNothing, 29 April 2019 - 12:09 AM.

[Nesutizale]

I saw somewhere some extracted files that showed whats used for the different weapons. They seam to use geometrie and while laser and ballistics are pretty small the ppc was huge. My guess would be that the "hitbox" was made 1:1 to the geometrie used for the different weapons. Thats why I set graphics = hitbox in my mind.
I agree that is not correct.

Still I find it strange that the PPC is the only weapon suffering that much from bad hit detection for such a long time. Was that meant to be so in the first place and not a bug? When it is a bug, why wasn't it never changed?
That the hit detection is overall bad is one side but when you gavea weapon a hitbox in the size of an Atlas then something is either wrong or its intentional.

----------

As for hit detection beeing difficulte and a poly accurate one even more. Its not only hard at the programming side. I remeber when we worked on Wing Commander Saga and had to work with poly accurate hit detection and some ships where fu... hard to hit like the pancake from the front where in older games they had big hitboxes you would hit from miles away.
How do you fix that or do you leave it be and let people addapt in playstyle?

I wonder if that is whats going on with the PPC...do they want people just to adapt?

[GuardDogg]

Yes, PPCs of any kind is really messed up in MWO. Even getting hit by them is not doing its job. Static up the cockpit visually, mess up the computer system and more. Imagine if PGI implemented that from Battletech, their would be a lot of complainers. A lot of weapons in MWO not doing its job.

Edited by GuardDogg, 29 April 2019 - 04:23 AM.

[LordNothing]

Nesutizale, on 29 April 2019 - 04:02 AM, said:

I saw somewhere some extracted files that showed whats used for the different weapons. They seam to use geometrie and while laser and ballistics are pretty small the ppc was huge. My guess would be that the "hitbox" was made 1:1 to the geometrie used for the different weapons. Thats why I set graphics = hitbox in my mind.
I agree that is not correct.

Still I find it strange that the PPC is the only weapon suffering that much from bad hit detection for such a long time. Was that meant to be so in the first place and not a bug? When it is a bug, why wasn't it never changed?
That the hit detection is overall bad is one side but when you gavea weapon a hitbox in the size of an Atlas then something is either wrong or its intentional.

----------

As for hit detection beeing difficulte and a poly accurate one even more. Its not only hard at the programming side. I remeber when we worked on Wing Commander Saga and had to work with poly accurate hit detection and some ships where fu... hard to hit like the pancake from the front where in older games they had big hitboxes you would hit from miles away.
How do you fix that or do you leave it be and let people addapt in playstyle?

I wonder if that is whats going on with the PPC...do they want people just to adapt?

wcs? sounds familiar, was that the freespace mod?

my theory is that pgi simply doesnt understand the pre existing collision system build into cryengine well enough for any kind of serious debugging. and that code may very well be in a closed source blob. nor do they have the skill to create a new system from scratch or leverage effective workarounds.

what i find is ppcs are hit and miss, i use them well say on a ppc warhawk or the quad lppc grinner i used in fp earlier. and on those mechs i got them split into 2 weapon groups each in chain fire mode and alternate the button presses. this usually lands better than pairs or triplets for some reason, sometimes just unlocking a velocity node can make a noticeable change.

its not just the ppcs srms are kind of like that too. i always wonder why srms do 2.15 damage/missile and i can only assume that was some kind of work around. they have the same problem where by firing too many at one time results in some hitting dead air and you get less damage. theres a limit on how much data you can shove down the network interface, so a lot of hits arent making it to the server.

Edited by LordNothing, 29 April 2019 - 04:30 AM.

[yrrot]

Nesutizale, on 28 April 2019 - 03:36 AM, said:

I shot the arms of an Atlas on tourmalin. Counting how long it would take to get through the armor and while I needed the same amount of shots for both arms...

If it took the same number of hits for both, doesn't that mean it's working?

Hit registration is entirely server authoritative. So if you have packet loss or a bad ping to the server, what you see on your screen and what the server simulates aren't always going to line up--even with host state rewind and lag compensation in play. It's difficult to say "it's a bug" when it is also a symptom of network connectivity issues. People notice it more with PPCs, which also makes sense. It's a big projectile that you can see hitting things on your screen, with a low enough rate of fire that you are watching the hits.

When you start chain firing, you're sending more packets to the server. So instead of potentially losing a packet with "I'm firing 3 PPCs", you're sending 3 different packets saying "I'm shooting 1 PPC". That reduces the impact of packet loss, potentially. Same thing if you are spamming UACs. Lasers and machine guns probably get tied to several updates since they have to update your aim point for each tick of damage, which makes them less problematic in general.

I'm not saying the bug doesn't exist, only that the bug may not be "PPC hitreg sucks". It might be "PGI doesn't have a magic solution for packet loss".

[Nesutizale]

I needed the same amout of shot to go through the armor but the internal structure took different amounts of damage and sometimes I needed one less shot to shoot the arm of.
So if a mech has weapons in the arm and its a difference of takeing them out in the first try or haveing to face the enemy again, potentialy getting shot too, then there is something wrong.

For the topic of servers and package loose.
Are trainingrounds not startet on the client? I mean reserving a spot on a server for the testinggrounds seams a bit of a waste in resources, dosn't it?
Anyway yes server connections are another issue. I doubt that any company has found the 100% solution to that and thats okay.

About hit registration I would seperate that from the server problem because we are talking about the PPC hitting terrain that should clearly not be in the way because every other weapon dosn't get blocked.
Its also not about invisible terrain. I know sometimes you think something should it but then dosn't because of that. In my example above my left arm should have been blocked by terrain, still I could shoot the Gauss but the PPC that is mounted even a bit higher would get stuck in the terrain. Thats messed up.

@LordNothing
Yes its the Wing Commander Mod for Freespace.

[Karl Streiger]

The training ground are not up to date as the life version, at least it was this way. So you had the "delayed convergence" on training grounds when it was long gone in life game.

For PPC yes the bolt is much bigger - i think it might have started as a special weapon feature (its lighting and as such should connect to the first protruding parts of a Mech) ...

Yep thinking back to the beginning... head shoting a Catapult was impossible with PPCs from the same level or lower.... you needed your PPC to hit it from above. Then the cockpit was the first thing that was in the "flight path" - were there not even some dancing sparks on impact?

[yrrot]

You also can't compare structure damage due to crits, so being one one PPC hit off on damage to structure could have been 1 crit bumping you over the limit to kill the arm before the other. Even empty components take crit damage = +15% weapon damage per crit, including some damage if you remove the last part of armor. You'd need a large sample set to weed that out, or debug info...

Maybe if you had a target with exactly 30 armor so your shots would only hit armor and remove all of it in one alpha, it would increase consistency.

I'm guessing you can't reproduce a PPC hitting static target and doing 0 damage in testing grounds. It is on your local machine, as you point out. The server side code and packet loss is what I think most people see when they bring up PPC hit registration, as I'm talking about above.

If the problem is "PPC projectiles are bigger than gauss rounds", is that a bug or as designed? I haven't paid that close attention to it since most of the mechs that I have PPCs on are higher mounts or poptarts anyway.

[Nesutizale]

Quote

You also can't compare structure damage due to crits, so being one one PPC hit off on damage to structure could have been 1 crit bumping you over the limit to kill the arm before the other. Even empty components take crit damage = +15% weapon damage per crit, including some damage if you remove the last part of armor. You'd need a large sample set to weed that out, or debug info...

Good point, forgot about that.