Devs: Please turn on character collision in PvP

Ser Lobo

There is a known tactic, where many players will move on top of each other, casting spells which effectively boost their own stats and overlap each other, while also limiting the effect of enemy area-of-effect attacks due to the inherent cap on the number of players that this AOE can effect. This tactic is known as 'turtling'.

There is another known tactic as old as gaming itself, where players with abilities capable of doing AOE damage, will run into the middle of a mob of enemies, spamming said attacks and killing far more players (without actually having to worry about targeting, aiming, or maneuvering). This is obviously known as 'AOE spam'.

There are multiple calls on these here forums (and the forums of many great games that have large-scale warfare, as ESO is not alone in experiencing AOE spam or turtling), for developers to 'fix' these issues.

One very common and semi-effective fix to AOE spam, is limiting the number of enemy or ally targets any AOE can possibly affect. This, however, reinforces the offensive power of turtling.

How do you solve both?

Don't allow players to collide with each other, which in turn limits the number of players that can stand in a specific area, while also limiting the number of players which can be affected by AOE or cone-based attacks.

A quicker resolution is, instead of the hard-collision rules normal in the PVE area of the game, lets take on a form of soft-collision: players are gently 'pushed' to the side or forward or backward of a player. Players can still pass 'through' each other, but motion remains constant and is not allowed to be ceased, even if the player stops moving his character.

At this point, even though two characters can share the same spot for a split second, they won't be able to stand on top of each other, and will constantly be pushed away.

Would this resolve the abuses of griefing and character lock seen in other games which did not allow character collision? Would this be a larger or smaller hit on server assets than the other fixes to turtling?

TheGrandAlliance

INdeed... "smart collision detection" is the solution. Also would make PvP more realistic: You couldn't just run through players up the walls and oil the defenders from above. You would actualy have to "Fight" your way into a keep just like any real battle.

"Collision detection" exists in real life. WHy not in video games is a most fundemental question yet to be answered...

ONly logical conclusion would be due to server lag. However this is more likely an excuse not to upgrade server capacity. It is difficult to determine the truth without experimental evidence however nonetheless...

ruzlb16_ESO

TheGrandAlliance wrote: »

ONly logical conclusion would be due to server lag. However this is more likely an excuse not to upgrade server capacity. It is difficult to determine the truth without experimental evidence however nonetheless...

It's also incredibly hard to code efficiently, and even harder to make it work properly when latency is taken into account - and while upgrading the server can help with the former, you'd need to upgrade the broadband infrastructure of a fairly large portion of the planet to deal with the latter.

Ser Lobo

ruzlb16_ESO wrote: »

TheGrandAlliance wrote: »

ONly logical conclusion would be due to server lag. However this is more likely an excuse not to upgrade server capacity. It is difficult to determine the truth without experimental evidence however nonetheless...

It's also incredibly hard to code efficiently, and even harder to make it work properly when latency is taken into account - and while upgrading the server can help with the former, you'd need to upgrade the broadband infrastructure of a fairly large portion of the planet to deal with the latter.

This is my argument against 'hard' collision detection, whereas no two characters can every occupy the same spot at the same time due to the game having to essentially predict the players movement, stop the movement into the other characters environment, and then prevent further movement that might interfere with that collision. And all this on a three-dimensional hitbox.

Whereas the proposed 'soft' collision detection can take place after the movement, and simply shifts the characters constantly in order to prevent prolonged overlap.

If constant 'rubber-banding' is happening by the player (where the character is being pushed into the same occupied location over and over again in order to achieve the same 'turtling' effect), a secondary movement freeze could occur for a second or two, reducing the server lag considerably while enforcing the motion.

And since the direction chose for the random move is not always an instant reversal, players would have to actively attempt to maintain that collision.

rich_nicholsonb16_ESO

Reducing group numbers and changing how healing works will also stop this turtling tactic.

Let's say make it 10 ppl per group and heals only effects group members unless everyone is full health then roaming with 2 groups ( 20 ppl ) would be less effective as the 2nd group won't get free heals.

Also change the way xp/ap rewards work. Makes it if the first group hits a player before the 2nd group they would get a bigger percent of the reward, tagging some ppl might call it.

Porting to keeps should have a timer when they first get taken. So if a keep gets taken a guild must claim it and wait let's say 10 mins before the port stone appears allowing ppl to port there. This will help spread the action out and in turn stop this zerging about. It will also give a reason to defend the area once taken instead of groups just running off to the next keep.

shimmyatwa_ESO

Used to love this in warhammer. Parking my iron breaker at the top of the stairs in a shield wall defending the keep for hours until the enemy could get enough numbers or give up

shimmyatwa_ESO

Would be great if they could implement this and move the flags upstairs. A smaller defense team would be on more even ground if they could bottleneck a huge zerg in the stairway with only so many people allowed on it so 50 people can't stack on each other and spam aoe.

Would also be cool if they could add enemy npc siege groups that would attack keeps and forts if a population owned the majority of the map. Might help keep a faction from rolling over a map and camping the gates if they had to worry about losing emporship bonuses or shrines to port back if they die.

Imagine being a lookout at the top of your keep twiddling your thumbs, when all of a sudden some khajit assassins start scaling your walls.

arnaldomoraleseb17_ESO

Is impossible to add collision detection right now. The servers will explode and desync issues will make the game unplayable.

Sureshawt

Yes....lets add even more overhead to the server cluster. After all its performing so well these days! /sarcasm off

ruzlb16_ESO

It would also permit viable tanking in PvP. But yes, it's basically undoable. Remember the lag in WAR when you had like 20 people on each side in a keep? The collision detection was a big part of that.

TheGrandAlliance

ruzlb16_ESO wrote: »

It would also permit viable tanking in PvP. But yes, it's basically undoable. Remember the lag in WAR when you had like 20 people on each side in a keep? The collision detection was a big part of that.

Yes but WAR is also a MMORPG that is over 5 years old... made by EA Games of course. HOpefully server technology has improved over the years...

frwinters_ESO

Collision detection will turn the game into a bunch of griefers. 50 people push 20 people against a keep wall trapped and beat them down. Thats all it will cause.

TheGrandAlliance

frwinters_ESO wrote: »

Collision detection will turn the game into a bunch of griefers. 50 people push 20 people against a keep wall trapped and beat them down. Thats all it will cause.

Thats how REal lifes works btw...

ruzlb16_ESO

TheGrandAlliance wrote: »

Yes but WAR is also a MMORPG that is over 5 years old... made by EA Games of course. HOpefully server technology has improved over the years...

It has, but collision detection remains a hugely processor-intensive operation. It's why many many modern engines (like Unity, say) don't permit mesh colliders over a certain number of surfaces, won't allow mesh-on-mesh by default, and why box/sphere/capsule colliders remain the norm even in single player games. On a server trying to deal with 2000 colliders at the same time... server tech is better now, but not THAT much better

Akhratos

ruzlb16_ESO wrote: »

It would also permit viable tanking in PvP. But yes, it's basically undoable. Remember the lag in WAR when you had like 20 people on each side in a keep? The collision detection was a big part of that.

I might remember different.

I remember they hired the very same hosting company that led DaoC european players to the US servers just because their inability.

With such a smart move I would not be so quick to blame server performance issues to collision.

joshisanonymous

TheGrandAlliance wrote: »

ruzlb16_ESO wrote: »

It would also permit viable tanking in PvP. But yes, it's basically undoable. Remember the lag in WAR when you had like 20 people on each side in a keep? The collision detection was a big part of that.

Yes but WAR is also a MMORPG that is over 5 years old... made by EA Games of course. HOpefully server technology has improved over the years...

WAR also had graphics from 5 years ago. It's not as if the game is literally WAR played on newer technology.

I'm gonna trust the judgment of the people who actually know the code in-and-out and have worked on building the server technology for the game for the last several years. I'm sure they would've come up with a solution by now if it were possible. When they added collision detection to PvE, they explained that that the only way to make it work between players would be to place greater limitations on how many people can be in PvP at once. I'm glad they didn't choose to go that route.

Ser Lobo

I believe some type of loose collision detection should be put into place. Obviously, hard collision detection (read: YOU SHALL NOT PASS) is a resource hog. But a soft collision system that lowers movement speed, pushes players apart, and generally attempts to keep time travelers from being in the same place at the same type might work and be less intensive.

Please note the issues here:
- The ability of small groups to engage and possibly triumph over large groups (many feel that lack of AOE caps is the solution, but this also is a resource hog as it allows so many more damage checks, which exponentially scale with the size of the battle).
- The ability of defensive and offensive AOE to affect a logical amount of friendlies or enemies (AOE caps are put in place simply to lower server load; meanwhile, players who 'stack' confuse the system either on purpose via turtling or on accident, since the red zone is obviously not effecting everyone in it logically).
- The ability for terrain and map control to have import (without any form of collision, the only way to hold an objective is to simply kill the enemy, with no further strategy capable of being implemented).
- The ability for melee to have import (wide open terrain is suitable for range and mobility, while close-quarters combat is suitable to melee; but if all close-quarters combat in the game is essentially a wide open hallway, melee have few options in place for them to set the stage in their favor, whereas ranged can always back up or simply walk past).
- Tanking and formations would have value (keeping casters in the back and the guys with shields in the front is a rather traditional element of combat, but with the current free movement, enemies simply walk through you to get to your casters, bypassing the tanks altogether).

Tarwin

Being a non programmer, could there be a way to just have CD on points of the body? Say head, elbows, knees and feet for example? Not the entire body saving resources?

Right now NPC CD is weird. If you ever try to scoot through two NPC's talking, you can't. It's almost as if they have a bubble around each NPC and even though they are (game) 3 feet apart, it is still like a wall

TheGrandAlliance

gamebouncerb16_ESO wrote: »

Being a non programmer, could there be a way to just have CD on points of the body? Say head, elbows, knees and feet for example? Not the entire body saving resources?

Right now NPC CD is weird. If you ever try to scoot through two NPC's talking, you can't. It's almost as if they have a bubble around each NPC and even though they are (game) 3 feet apart, it is still like a wall

Yes... a "hitbox" is a 3D field... like a sphere or whatever around a target. ANyhing engages that "Sphere" and motion is stopped.

Draconiuos

I know they should add a passive into the Alliance War skill lines that allows you to get an NPC collision bubble. This would be a double edge sword for the player because they could stop players from running through a gap in a wall, but would not be able to stack with others on their team.

frwinters_ESO

TheGrandAlliance wrote: »

frwinters_ESO wrote: »

Collision detection will turn the game into a bunch of griefers. 50 people push 20 people against a keep wall trapped and beat them down. Thats all it will cause.

Thats how REal lifes works btw...

Well yeah i guess in real life people can be jerks rough you up against a wall and slice you with magical abilities and swords, yeah your right.

But i play a game to get away from real life where i can walk through people and stabby them with daggers.

Lowbei

its too late, the change would require a population cap drop, which would not be worth it to the company, despite it being a good feature.

chrisub17_ESO104

ruzlb16_ESO wrote: »

TheGrandAlliance wrote: »

Yes but WAR is also a MMORPG that is over 5 years old... made by EA Games of course. HOpefully server technology has improved over the years...

It has, but collision detection remains a hugely processor-intensive operation. It's why many many modern engines (like Unity, say) don't permit mesh colliders over a certain number of surfaces, won't allow mesh-on-mesh by default, and why box/sphere/capsule colliders remain the norm even in single player games. On a server trying to deal with 2000 colliders at the same time... server tech is better now, but not THAT much better

The main issue I ran into while working on a project where we were doing collision detection on thousands of game objects, was that it really only works on a lan with deterministic latency. Otherwise you are constantly having to correct clients, and they get the rubber-banding effect. With a lot of objects in a small area it could get to the point where you either have to let objects pass through each other, or they would bounce around forever before being able to get out of the 'zergball' of objects. It would be a very unpleasant experience as a player, getting tossed around like a rag doll basically in worst case scenarios.

We were using a distributed computing framework that's open source along with bullet for physics, no commercial software. Latency issues aside we scaled it to 40,000 game objects in the same grid using half a dozen commodity intel servers. We had no server performance issues. We ended up ruling it out as practical because there was just no good way to work around the client latency.

You need pretty low latency also, around 30ms guaranteed before it just works without being able to notice rubber-banding (which still happens, it's just that the corrections are tiny and you have far less of them).

popatiberiuoneb18_ESO

ruze84b14_ESO wrote: »

There is a known tactic, where many players will move on top of each other, casting spells which effectively boost their own stats and overlap each other, while also limiting the effect of enemy area-of-effect attacks due to the inherent cap on the number of players that this AOE can effect. This tactic is known as 'turtling'.

There is another known tactic as old as gaming itself, where players with abilities capable of doing AOE damage, will run into the middle of a mob of enemies, spamming said attacks and killing far more players (without actually having to worry about targeting, aiming, or maneuvering). This is obviously known as 'AOE spam'.

There are multiple calls on these here forums (and the forums of many great games that have large-scale warfare, as ESO is not alone in experiencing AOE spam or turtling), for developers to 'fix' these issues.

One very common and semi-effective fix to AOE spam, is limiting the number of enemy or ally targets any AOE can possibly affect. This, however, reinforces the offensive power of turtling.

How do you solve both?

Don't allow players to collide with each other, which in turn limits the number of players that can stand in a specific area, while also limiting the number of players which can be affected by AOE or cone-based attacks.

A quicker resolution is, instead of the hard-collision rules normal in the PVE area of the game, lets take on a form of soft-collision: players are gently 'pushed' to the side or forward or backward of a player. Players can still pass 'through' each other, but motion remains constant and is not allowed to be ceased, even if the player stops moving his character.

At this point, even though two characters can share the same spot for a split second, they won't be able to stand on top of each other, and will constantly be pushed away.

Would this resolve the abuses of griefing and character lock seen in other games which did not allow character collision? Would this be a larger or smaller hit on server assets than the other fixes to turtling?

As much as i would love to have colision detection in PvP (and remove or revamp it in PvE) im olmoast certain that curent game engine and prediction alghoritms used to compensate for the latency will not be able to handle this in an enjoyable way.

Would be awsome if ZoS could pull this off though.

Nazon_Katts

@chrisub17_ESO104 (man, how many accounts have you got?! )

Since you seem to be in the known, if we tied CD to an ability that only works when activated/triggered and when not moving, wouldn't this significantly lower the strain on the server and work with higher latency?

So in terms of ESO, if you block and stand still for a certain amouint of time (suppose that's where you could counter latency issues), you gain CD, if you move, you lose it.

Or would the toggling of CD cause even more problems, even though there'll be less objects(players)?

Oblongship

ruze84b14_ESO wrote: »

There is a known tactic, where many players will move on top of each other, casting spells which effectively boost their own stats and overlap each other, while also limiting the effect of enemy area-of-effect attacks due to the inherent cap on the number of players that this AOE can effect. This tactic is known as 'turtling'.

There is another known tactic as old as gaming itself, where players with abilities capable of doing AOE damage, will run into the middle of a mob of enemies, spamming said attacks and killing far more players (without actually having to worry about targeting, aiming, or maneuvering). This is obviously known as 'AOE spam'.

There are multiple calls on these here forums (and the forums of many great games that have large-scale warfare, as ESO is not alone in experiencing AOE spam or turtling), for developers to 'fix' these issues.

One very common and semi-effective fix to AOE spam, is limiting the number of enemy or ally targets any AOE can possibly affect. This, however, reinforces the offensive power of turtling.

How do you solve both?

Don't allow players to collide with each other, which in turn limits the number of players that can stand in a specific area, while also limiting the number of players which can be affected by AOE or cone-based attacks.

A quicker resolution is, instead of the hard-collision rules normal in the PVE area of the game, lets take on a form of soft-collision: players are gently 'pushed' to the side or forward or backward of a player. Players can still pass 'through' each other, but motion remains constant and is not allowed to be ceased, even if the player stops moving his character.

At this point, even though two characters can share the same spot for a split second, they won't be able to stand on top of each other, and will constantly be pushed away.

Would this resolve the abuses of griefing and character lock seen in other games which did not allow character collision? Would this be a larger or smaller hit on server assets than the other fixes to turtling?

No.

2 reasons..

1 Griefing/exploiting this and you know it would happen.

2 It is performance reasons...the lag would be unbarrable with all the additional physics calculations.

Talcyndl

They've already said they can't do this with the present system because of performance issues. Maybe a nice idea, but not technically feasible.

In fact, they initially didn't even want to have collision in PvE environments because of the performance hit. No way that Cyrodiil can handle it.

Nooblet

I absolutely hate collision detection in mmos.

aeroch

Could the server handle it if collision was tied to a short duration skill, like Immovable? The idea of limited collision is appealing but it isn't something I'd want to deal with all the time. I could see it being fun in some choke point scenarios, though.

Lowbei

the argument that it would cause griefing is invalid, since theres easy workarounds like moving slowly thru friendlies or only while sneaked.

the only thing stopping CD in ESO is the performance hit.

babanovac

Nooblet wrote: »

I absolutely hate collision detection in mmos.

I think the only MMO that had collision detection in PvP was Warhammer Online. And no, it did not cause any griefing at all, because it was only enabled while in combat. So out of combat there was no CD.