168 replies to this topic

[SQW]

I'd like to see a frame by frame analysis of hit-reg of LRM-on-light.

Seriously, normal lasers are TERRIBLE at hitting a light travelling at 130+km/hr.

Okay, correction: hitting is easy but actually staying on one spot long enough to do decent dmg? That Atlas pilot's aim is actually above average for pub game - his situation awareness is lousy tho for being caught out without support.

PPC, AC, Pulse lasers, LBX-10s, SRMS, SSRMs - these are the weapons you trash lights with. You don't try to wash your dog with a Super Soaker while its running for its life.

[Wolfways]

GreyNovember, on 07 March 2016 - 04:43 PM, said:

I don't know about you, but if I run straight towards an enemy mech's front, I get blown to pieces because I was stupid enough to expose my torso directly towards all the guns. Of course this is hearsay as much as your experience is.

Any mech running straight towards an enemy will get blown to pieces.

Quote

But I mean if your statement is true, then there would be more pilots using easymode, now wouldn't there, and not heavies?

Not really. Think how many players choose mechs based on meta, or what other players/friends suggest.
Also, the game is about big mechs shooting the crap out of each other. It just makes sense that even if lights were the equal to heavier mechs many players would still pick the bigger mechs. Imo if assaults weren't so slow they would always be the most played, but heavies are because while they have a lot of tonnage in weapons they also have decent speed.
I think the only way to make the light queue bigger would be to make light mechs very op.

Edited by Wolfways, 08 March 2016 - 11:53 AM.

[mailin]

I can't believe I read through all 7 pages. I can't say how many times I've given this advice, if you want to kill a light, chain fire your weapons and go for the legs.

To the argument that speed makes it hard to hit, this is only talking about the speed of the target, and talking as if the firing mech were standing still. If the argument is truly valid, there is no way I should be able to hit an ACH from my 5D if he's going 138ish kph and I'm moving at 148ish. And yet, I can and do regularly leg ACHs.

It's not the hit reg, nor the hit boxes. It's the speed of the target. Well, here's a news flash, fast moving targets are hard to hit.

You silly assault drivers keep trying to alpha us and we'll keep making you pay for it. Set your weapons to chain fire and it will make the competent light drivers look for greener pastures.

Totally from an anecdotal note, I have been playing since August 2012 and I remember knockdown and I remember people claiming lag shield. Which I could never figure out because my ping has always been right around 40. When PGI came out with HSR there were a LOT of people on the forums applauding because they would finally be able to hit lights and the light queue tanked. And lights did become easier to hit . . . a LOT easier. But apparently they weren't easy enough because PGI lowered speed tweak and lights became even easier to hit. Now some members are QQing that it isn't enough.

My advice to those players is switch weapons and chain fire them. Aim for the legs and watch the lights get legged very, very quickly. Once legged all of your other aiming problems will seem to go away.

Edited by mailin, 07 March 2016 - 06:41 PM.

[Sigilum Sanctum]

mailin, on 07 March 2016 - 06:40 PM, said:

My advice to those players is switch weapons and chain fire them. Aim for the legs and watch the lights get legged very, very quickly. Once legged all of your other aiming problems will seem to go away.

The key point is that people should be aiming for the THIGH. The thigh on a bipedal light mech has the least dimension of movement compared to the shin or ankles that I see people shooting all the goddamn time.

From the axis where it connects to the hip down to the knee there is relatively little motion of movement compared to the ankle or shin that has the long arcs where the mech makes a step. This is easier to discern on the Cheetah because by default its running animation seems slower than that of the Firestarter.

[Kaisha]

mailin, on 07 March 2016 - 06:40 PM, said:

To the argument that speed makes it hard to hit, this is only talking about the speed of the target, and talking as if the firing mech were standing still.

If you're in an assault you might as well be standing still. In fact most of the time you're slowly moving backwards simply trying to keep the light in front of you.

Quote

If the argument is truly valid, there is no way I should be able to hit an ACH from my 5D if he's going 138ish kph and I'm moving at 148ish. And yet, I can and do regularly leg ACHs.

Completely untrue. The problem with BUGS is that they aren't consistent. Sometimes you hit them fine, other times if goes right through them. I've had times where I've 1-shot a light, other times where I've unloaded everything for 60s straight and maybe did 5 dmg total.

Quote

It's not the hit reg, nor the hit boxes. It's the speed of the target. Well, here's a news flash, fast moving targets are hard to hit.

News flash, that's a strawman arugment. You (and others) keep implying that everyone who can't hit a light are bad pilots and saying so because they just hate lights. Funny thing is, people complaining about hit regs play lights themselves, and find the whole thing ridiculous. When just picking a light mech makes me a 10x better player... something is wrong.

Quote

You silly assault drivers keep trying to alpha us and we'll keep making you pay for it. Set your weapons to chain fire and it will make the competent light drivers look for greener pastures.

Makes little difference. And again, this whole notion that everyone who is complaining only play assaults. Everyone plays every mech size. No one want to see lights gutted because then our lights would be useless as well. We just get sick of watching yet another locust face tank a DWF and come out without a scratch.

Quote

My advice to those players is switch weapons and chain fire them. Aim for the legs and watch the lights get legged very, very quickly. Once legged all of your other aiming problems will seem to go away.

... That's what everyone's been saying. The server/net code can't keep up with the speed the fast mechs can move at, causing issues with hit reg at high speeds... If people could hit the legs they would...

At this moment in the game, lights have the highest survivability in the game by a long shot, which does not make sense from a gameplay perspective, a TT perspective, or a lore perspective. Whether its by adjusting hit boxes, server side/net code changes, weapon convergence changes, or something else, something has to change because as it stands things are broken.

[MrMadguy]

1) You should understand, that I don't tell you, that this problem affects all Lights - and this is the root of problem. If I would have bad aim, then I wouldn't be able to kill any Light 'Mech, but it isn't the case. Some Light 'Mechs die properly, despite of having exactly the same hit boxes and speed. And some are offenders. And during the thousands of matches of collecting statistics I found out, what 'Mechs are offenders.

2) I have 40ms ping, I play at 75fps, I have a good mouse, I specially trained my anti-Light skills, so usually I'm better against Lights, than any other player in my team. And you should understand, that game should be built around average human capabilities - around average reaction time, average aim time, average aim accuracy. It's called ergonomics. When something has a size of just few pixels and moves with extreme speed, so it's just physically impossible to aim at it - then it's bad game design, not bad players.

3) You are talking like Paladins back in WotLK. You tell us, that Lights aren't OP - they are fine, you just need L2P to play against them properly. Do you understand, that if something requires higher level of skill to play against it - then it's the exact definition of imbalance? I tried to L2P. I know about leading, I know about convergence, I found amounts of leading, required by every weapon, I learned tactics - it isn't helping at all. Same with Paladins: you have to deal with plate wearer, 2H equipping player with greatest crits in a game, 300% survivability (bubble+LOH without restrictions), 8 seconds of God Mode, passive healing, DOT on other player, spellpower, enough for getting back to full HP via just one flash cast, debuff to other player's stats, best CC in a game - stun, has ability to remove any slowing/rooting effects every few seconds, remove defuffs from himself, most spells are ranged and instant, etc. And you are the player, who have only one "life" and no CC. One mistake - and you're instantly dead. And he can faceroll or even push buttons randomly to win (Can't find that glorious video, where Paladin used castrandom macros to beat other competitive player).

And when one player is required to learn a lot of things and another - to just jump in a push two buttons - that's, what is called imbalance. And don't tell me about "Light skill" - I tried playing Lights by myself - piece of cake. I don't play them, just because I don't like them visually and it's boring to play on easy mode.

Conclusion? We need normalization!

Yeah! Found it! Proceed to 17:00.

Edited by MrMadguy, 08 March 2016 - 01:55 AM.

[Tarogato]

mailin, on 07 March 2016 - 06:40 PM, said:

I can't believe I read through all 7 pages. I can't say how many times I've given this advice, if you want to kill a light, chain fire your weapons and go for the legs.

To the argument that speed makes it hard to hit, this is only talking about the speed of the target, and talking as if the firing mech were standing still. If the argument is truly valid, there is no way I should be able to hit an ACH from my 5D if he's going 138ish kph and I'm moving at 148ish. And yet, I can and do regularly leg ACHs.

It's not the hit reg, nor the hit boxes. It's the speed of the target. Well, here's a news flash, fast moving targets are hard to hit.

You silly assault drivers keep trying to alpha us and we'll keep making you pay for it. Set your weapons to chain fire and it will make the competent light drivers look for greener pastures.

Totally from an anecdotal note, I have been playing since August 2012 and I remember knockdown and I remember people claiming lag shield. Which I could never figure out because my ping has always been right around 40. When PGI came out with HSR there were a LOT of people on the forums applauding because they would finally be able to hit lights and the light queue tanked. And lights did become easier to hit . . . a LOT easier. But apparently they weren't easy enough because PGI lowered speed tweak and lights became even easier to hit. Now some members are QQing that it isn't enough.

My advice to those players is switch weapons and chain fire them. Aim for the legs and watch the lights get legged very, very quickly. Once legged all of your other aiming problems will seem to go away.

Not sure why you would advocate chainfire when dealing with lights when you already know that the goal is to concentrate damage. It's best to light up one good shot and let 'er rip. One good burn is usually enough to open a light's leg if you can hold it steady, and once a light has an open leg they're not so brave anymore. Or CT in the case of those blasted Oxides. xD
But if you're using chainfire you're essentially adding a pre-built damage-spreading mechanism into the mix.

MrMadguy, on 08 March 2016 - 12:31 AM, said:

1) You should understand, that I don't tell you, that this problem affects all Lights - and this is the root of problem. If I would have bad aim, then I wouldn't be able to kill any Light 'Mech, but it isn't the case. Some Light 'Mechs die properly, despite of having exactly the same hit boxes and speed. And some are offenders. And during the thousands of matches of collecting statistics I found out, what 'Mechs are offenders.

2) I have 40ms ping, I play at 75fps, I have a good mouse, I specially trained my anti-Light skills, so usually I'm better against Lights, than any other player in my team. And you should understand, that game should be built around average human capabilities - around average reaction time, average aim time, average aim accuracy. It's called ergonomics. When something has a size of just few pixels and moves with extreme speed, so it's just physically impossible to aim at it - then it's just wrong.

...

And when one player is required to learn a lot of things and another - to just jump in a push two buttons - that's, what is called imbalance. And don't tell me about "Light skill" - I tried playing Lights by myself - piece of cake. I don't play them, just because I don't like them visually and it's boring to play on easy mode.

Conclusion? We need normalization!

If lights are so powerful and so easy to play, then why aren't more people playing them? Why isn't the light queue constantly at 40% or higher? It would make sense to me that if lights are as strong as you think they are, there would be a lot more people playing them.

Edited by Tarogato, 08 March 2016 - 12:33 AM.

[GreyNovember]

Kaisha, on 07 March 2016 - 11:25 PM, said:

... That's what everyone's been saying. The server/net code can't keep up with the speed the fast mechs can move at, causing issues with hit reg at high speeds... If people could hit the legs they would...

Hold on. Let's back up on the logic train a bit here.

If speed is an issue, then we should see a lot more Dual Gauss users being upset that their shots don't register.

Yet strangely that isn't the case.

Two colliders are still at play here. The Gauss Slug and the component being hit.

So servers not being able to keep up with MECH speed does not add up. Something is fishy here.

[Kaisha]

Tarogato, on 08 March 2016 - 12:31 AM, said:

If lights are so powerful and so easy to play, then why aren't more people playing them? Why isn't the light queue constantly at 40% or higher? It would make sense to me that if lights are as strong as you think they are, there would be a lot more people playing them.

Because this game doesn't have high level competitive play. If there was actually a ranked queue with pro gamers playing for money, you'd never see an assault again. This is a casual game with alot of older (ie. slower reaction time) players. Most people are playing for fun, not simply 'to the meta'. That's still no excuse for game imbalance.

[MrMadguy]

Tarogato, on 08 March 2016 - 12:31 AM, said:

If lights are so powerful and so easy to play, then why aren't more people playing them? Why isn't the light queue constantly at 40% or higher? It would make sense to me that if lights are as strong as you think they are, there would be a lot more people playing them.

Straw-man. Correct fallacy? Or not? Don't use fact, that isn't connected with problem, as argument, that confirms your statement. Unpopularity of something can't be used as justification for making it OP, as unpopularity can be caused via other factors - may be majority of players just want to play big stompy robots? Forcing players to play something, that doesn't suit their playstyle, just because you need to fill your game with this thing? That's, what is called "bring the class - not the player" bad game design. It's same, as telling, that laser vomit - is the reason, why Meta 'Mech for Meta Map selection is impossible. Laser vomit - is weapon balance problem, while 'Mech selection - is Map design problem. This two aren't connected.

Edited by MrMadguy, 08 March 2016 - 01:07 AM.

[MrMadguy]

GreyNovember, on 08 March 2016 - 12:44 AM, said:

Hold on. Let's back up on the logic train a bit here.

If speed is an issue, then we should see a lot more Dual Gauss users being upset that their shots don't register.

Yet strangely that isn't the case.

Two colliders are still at play here. The Gauss Slug and the component being hit.

So servers not being able to keep up with MECH speed does not add up. Something is fishy here.

Projectile speed - is also one of the factors, that affects quantization error problem. As I understand, server can't handle such high projectile speeds - so projectile is represented as vector between it's positions on two neighbouring frames. If this vector intersects hitboxes between frames - hit is detected. Higher speed - means bigger displacement and longer vector - lower chance of quantization error and higher chance of hit. That's why higher projectile speed weapons - are usually more reliable against Lights. For example PPC and AC/20 were good against Lights in the past. But after speed nerf they turned into garbage.

Edited by MrMadguy, 08 March 2016 - 01:26 AM.

[GreyNovember]

MrMadguy, on 08 March 2016 - 01:15 AM, said:

so projectile is represented as vector between it's positions on two neighbouring frames. If this vector intersects hitboxes between frames - hit is detected. Higher speed - means bigger displacement and longer vector - lower chance of quantization error and higher chance of hit

...Quantization? Sorry, what?

Source, please. At what point did a developer reveal how projectile collisions are handled?

Furthermore, if in fact projectiles are vectors, and longer is better, then lasers should be EVEN BETTER Than gauss by virtue of being essentially a massive raycast. But that doesn't seem to be the case. People are insisting lasers are terrible against lights.

So you bring up an inconsistency.

Edited by GreyNovember, 08 March 2016 - 01:55 AM.

[MangoBogadog]

MrMadguy, on 07 March 2016 - 10:59 AM, said:

Usually something like that happens:

Every large laser is 9dmg and Firestarter has about LA+RA+LT+RT+ST 100 front armor total (in case of more even distribution, Light pilots usually prefer) - about 20 armor per part. So one 3xLL Alpha can core every part of Firestarter. Yeah, some shots were definitely partial - those, that happened before Firestarter overheated for example. But others, that were accurate enough - definitely not enough dmg registered. For example a lots of 100% damage at the beginning of the fights (right after hugging) obviously haven't registered on Firestarter's STs. Same happened at the end of the fight - dmg registers of arms, but not STs.

Watch it on 0.25x speed, a lot of misses and very few good hits.

Also many light pilots run 40 CT, 30 RT/LT, 24 RA/LA armour. Thats quite a bit more than 100 total front facing armour.

The 27 alpha CT when it shut down did about as much damage as expected. As did the other partial misses before.

[MrMadguy]

GreyNovember, on 08 March 2016 - 01:54 AM, said:

...Quantization? Sorry, what?

Source, please. At what point did a developer reveal how projectile collisions are handled?

Furthermore, if in fact projectiles are vectors, and longer is better, then lasers should be EVEN BETTER Than gauss by virtue of being essentially a massive raycast. But that doesn't seem to be the case. People are insisting lasers are terrible against lights.

So you bring up an inconsistency.

Sorry, but with 30 frames per second projectile displacement between frames would be 2000/(3.6 * 30) = 18.5 meters.

You should know, that world in 3D games isn't material - it's just a bunch of triangles. Physics are usually organized as ray intersection tests: you know, that during one frame you will move from point A to point B. How can you test, whether it's possible? You create A->B ray and test all triangles on your way for intersection with this ray. Closest intersection point - is point, where collision happens.

This is even better:

Yeah, many misses, but there were clear hits, that weren't registered.

Edited by MrMadguy, 08 March 2016 - 02:48 AM.

[GreyNovember]

MrMadguy, on 08 March 2016 - 02:38 AM, said:

Sorry, but with 30 frames per second projectile displacement between frames would be 2000/(3.6 * 30) = 18.5 meters.

What is this addressing? You kind of just pulled numbers out of somewhere and stated them.

Quote

You should know, that world in 3D games isn't material - it's just a bunch of triangles.

Right. Except not every collider is exactly uniform to the mesh. A Firestarter's leg, for example, could be setup as 3 cube colliders, instead of having every vertex be included in the collider. Much less costly to check on. But this is beside the point in this instance.

Quote

Physics are usually organized as ray intersection tests: you know, that during one frame you will move from point A to point B. How can you test, whether it's possible? You create A->B ray and test all triangles on your way for intersection with this ray. Closest intersection point - is point, where collision happens.

I'm not disagreeing with you. But what exactly is this supposed to get at?

You also didn't really address this at all:

Quote

Source, please. At what point did a developer reveal how projectile collisions are handled?

Furthermore, if in fact projectiles are vectors, and longer is better, then lasers should be EVEN BETTER Than gauss by virtue of being essentially a massive raycast. But that doesn't seem to be the case. People are insisting lasers are terrible against lights.

So you bring up an inconsistency.

[MrMadguy]

GreyNovember, on 08 March 2016 - 02:51 AM, said:

What is this addressing? You kind of just pulled numbers out of somewhere and stated them.



Right. Except not every collider is exactly uniform to the mesh. A Firestarter's leg, for example, could be setup as 3 cube colliders, instead of having every vertex be included in the collider. Much less costly to check on. But this is beside the point in this instance.



I'm not disagreeing with you. But what exactly is this supposed to get at?

You also didn't really address this at all:

I'm not a teacher and you don't pay me for teaching you, how game development works. Google is your friend.

[GreyNovember]

MrMadguy, on 08 March 2016 - 03:01 AM, said:

I'm not a teacher and you don't pay me for teaching you, how game development works. Google is your friend.

You refuse to address my point on the grounds of "I don't want to" then. Okay.

If you don't want to actually have a discussion for whatever reason, be it a valid reason or a cop out, you can just say " I don't want to discuss this further" y'know?

[MrMadguy]

GreyNovember, on 08 March 2016 - 03:16 AM, said:

You refuse to address my point on the grounds of "I don't want to" then. Okay.

If you don't want to actually have a discussion for whatever reason, be it a valid reason or a cop out, you can just say " I don't want to discuss this further" y'know?

Too long to explain and I can't find good article or video, where it's explained. Just imagine, that you play with low FPS or see other player with high ping - he "teleports". Now imagine, that server works the same way - movement isn't continuous - it's series of discrete states - frames. Objects are moved and collision tests are performed only during this discrete frames. If you won't take care about it - objects will just "teleport" from one position to other without interacting with each other. If server has 30FPS and projectile has 2000kph (555mps) speed - then it's 18.5 meters per frame. So, if you won't take care of it - projectile will have following discrete positions with every fame: 0, 18.5, 37, etc. And if your target is between this positions - projectile won't hit it - it will simply "teleport" through it.

[GreyNovember]

MrMadguy, on 08 March 2016 - 03:36 AM, said:

Too long to explain and I can't find good article or video, where it's explained. Just imagine, that you play with low FPS or see other player with high ping - he "teleports". Now imagine, that server works the same way - movement isn't continuous - it's series of discrete states - frames. Objects are moved and collision tests are performed only during this discrete frames. If you won't take care about it - objects will just "teleport" from one position to other without interacting with each other. If server has 30FPS and projectile has 2000kph (555mps) speed - then it's 18.5 meters per frame. So, if you won't take care of it - projectile will have following discrete positions with every fame: 0, 18.5, 37, etc. And if your target is between this positions - projectile won't hit it - it will simply "teleport" through it.

FPS Refers to frames being rendered. That is irrelevant to the actual update rate of the server. It doesn't need to render anything. Only recieve and validate client events.

Worry not. You've explained twice now how something moving at X Speed in a 3D Space, that checks collisions every Y Interval, might bypass a Z Dimensioned area defined as a hitbox.

Again however. You are not addressing this. I'm still waiting for your source on how you know projectiles are not colliders, and how your raycast analogy holds any water when you consider that a laser is a raycast that goes on for a CONSIDERABLE length of time.

Quote

Source, please. At what point did a developer reveal how projectile collisions are handled?

Furthermore, if in fact projectiles are vectors, and longer is better, then lasers should be EVEN BETTER Than gauss by virtue of being essentially a massive raycast. But that doesn't seem to be the case. People are insisting lasers are terrible against lights.

So you bring up an inconsistency.

[Kaisha]

GreyNovember, on 08 March 2016 - 03:42 AM, said:

FPS Refers to frames being rendered. That is irrelevant to the actual update rate of the server. It doesn't need to render anything. Only recieve and validate client events.

Worry not. You've explained twice now how something moving at X Speed in a 3D Space, that checks collisions every Y Interval, might bypass a Z Dimensioned area defined as a hitbox.

Again however. You are not addressing this. I'm still waiting for your source on how you know projectiles are not colliders, and how your raycast analogy holds any water when you consider that a laser is a raycast that goes on for a CONSIDERABLE length of time.

You clearly know enough about how this works to answer that question on your own. I'm wondering why you feel the need for him to spell it out? Its like you're playing games to try to appear clever, but you're just being an ***.

In the interest of others who might be reading this....

Most physics simulations (whether on a server or on a client in a 1 player game) work with an internal tick count, much like a frame rate. Yes you can do variable frame rate physics, but for a video game that can be incredibly hard both to program as well as to design for as it is often less than intuitive (emergent physics is interesting, but its not necessarily good for a pvp game). Internal tick based physics (where the server is updated at discreet intervals at X times a second, each update is usually called a tick) is by far the most common situation. So to call it FPS might be a bit of a misnomer, but not entirely incorrect (and if you want to play the semantics card, in computer science the word 'frame' has many meanings based on context, it doesn't necessarily refer to a 2D rendered image displayed to a monitor/screen, it often does, but not necessarily, so lets lay off the semantic games).

Also while projectile travel is most likely handled by a bounded ray, the mech bounding volumes being tested against are not. They would move at discreet intervals. If the tick rate is low (in comparison to the speed the mechs are moving at) a mech could run through a laser beam and receive no damage; as in the first 'frame'/'tick' the mech would be on one side of the beam, and by the second 'frame'/'tick' would have completely traversed the beam ray and be on the other side of it, registering no 'hit' in between. I believe that is the phenomena they are referring to when talking about quantanization (the movement of the mechs of course being quantized, NOT the projectiles). If that is the problem the solutions then are to:
A ) increase the tick rate - increases server load but is trivial to implement and is effective
B ) decrease mech speed
C ) increase bounding volume size
D ) implement swept bounding volumes - complex to implement

Course all this in complicated by the fact that network latency makes things more difficult. Perhaps its a server issue, perhaps a network issue, perhaps (and most likely) a bit of both. Irregardless of the underlying reason, the problem needs to be fixed.

Edited by Kaisha, 08 March 2016 - 04:29 AM.