Guild Wars 2

I was playing on my elementalist, but this has happened with my ranger’s pet before.

Basically, my teammate (in pstory mission) targeted an enemy and ran after it. Unfortunately, the enemy did the same. Unfortunately, they were both melee. UNFORTUNATELY, there was a corner they had to go around to reach eachother.
UNFORTUNATELY, the ai is too stupid to detect movements.

They just kept running around the corner “chasing eachother”.

Green spray = where we came from
Orange spray = where they came from
Green/Orange lines = movements
Box with red lines = go figure…. (obstacle for the stupid ones)

When this happened with the pet, there wasn’t even an obstacle. They were just running back and forth trying to catch eachother.

the AI sucks because they dont have someone who can actually program a good one I think…
The Petsystem/AI in this game is the worst one I have played for quite some time.
It is beyond me how that could happen.
There is no bond to the ranger pet whatsoever (they dont even “remember” their name) – but it actually does not matter because you just switch them around anyways … and the AI… wow… whoever did it should consider changing his/her profession…

/rantoff

So sad, but so true…

Mardermann.7468:

the AI sucks because they dont have someone who can actually program a good one I think…
The Petsystem/AI in this game is the worst one I have played for quite some time.
It is beyond me how that could happen.
There is no bond to the ranger pet whatsoever (they dont even “remember” their name) – but it actually does not matter because you just switch them around anyways … and the AI… wow… whoever did it should consider changing his/her profession…

/rantoff

Maybe I was going insane for a second, but I think my pets started remembering names I gave them after the last patch. Try it out and see. I can’t play much for a while so I can’t test it more.

Indoles.1467:

Maybe I was going insane for a second, but I think my pets started remembering names I gave them after the last patch. Try it out and see. I can’t play much for a while so I can’t test it more.

If you have the same pet in both ground and underwater slots, it will remember if you switch it off and on again in only one of the slots.

example: I have Tails in both slots, I switch Tails for a panther in the ground slot, and then switch the panther for Tails again. It will remember the name.

Monster AI and pet Ai are two different things. Both monster and pets have a problem detect different elevations when there is no detectable paths (paths that don’t require jumping on or off.)

Pet A.I is meant to be imperfect. Having a set it and forget it pet isn’t what being a guildwars 2 ranger is about. Ranger are suppose to work in tandem with their pet. Not one or the other doing all the work. Just as you are able to lose an attacker you should be able to lose the pet. If pet hit you with every strike or auto attack it would be unfair.

I just played an arcanist in FF reborn, I have to say pets can even detect the right time to use a skill, example : use a push when an enemie is near your master.
Why we can have a decent AI??

Serraphin Storm.2369:

Monster AI and pet Ai are two different things. Both monster and pets have a problem detect different elevations when there is no detectable paths (paths that don’t require jumping on or off.)

Pet A.I is meant to be imperfect. Having a set it and forget it pet isn’t what being a guildwars 2 ranger is about. Ranger are suppose to work in tandem with their pet. Not one or the other doing all the work. Just as you are able to lose an attacker you should be able to lose the pet. If pet hit you with every strike or auto attack it would be unfair.

It’s a different story when the ranger is very dependant on the pet. The pet can’t “work in tadem” with the ranger because it’s pathfinding is poor. I agree you should be able to lose the pet but the method shouldn’t be running around in circles.

The problem is with the pathfinding in this game. Object A checks where Object B is and is instructed to move to Object B. Object B is moving away while Object A is still moving to Object B’s old position. Object A checks where Object B is and is instructed…. and so on.

The time between checks is too long and it results in a jaggy path.

Ai moving around a corner is not a smooth action. They simply put instantly turn 90 degrees.

Not to mention the pet pathfinding AI which usually overshoots the target

Serraphin Storm.2369:

M
Pet A.I is meant to be imperfect. Having a set it and forget it pet isn’t what being a guildwars 2 ranger is about. Ranger are suppose to work in tandem with their pet. Not one or the other doing all the work. Just as you are able to lose an attacker you should be able to lose the pet. If pet hit you with every strike or auto attack it would be unfair.

that is the most bullkitten excuse of a bad written code I have ever read (and I read it quite a lot of the “all is good” crowd)

dont get me wrong – I play the animal pet class in almost any mmo I have ever played (and I played a lot of them in my time) – but I have never witnessed a worse implemented petsystem then the one in GW2… (not to mention the AI which is substandard even compared to 10 y/o games (all imho of course)

I am not grumpy, I am not jaded – but neither do I look at everything with the rose tinted goggles… we where told over a half year ago that we would get fixes and a better AI etc etc… all we got was a not well thought out function (pet getting behind the mob when no aggro) and more nerfs (well not just nerfs – we are pretty good under water now)

I love GW2 … but I really hate what they did with the ranger…

But then I have 8 level 80s now and I rather enjoy my warrior and guardian…

So I will keep watching the ranger forums (smiling when reading from the “all is good” crowd) cause my ranger is still my main character…

Soronthar.7236:

Indoles.1467:

Maybe I was going insane for a second, but I think my pets started remembering names I gave them after the last patch. Try it out and see. I can’t play much for a while so I can’t test it more.

If you have the same pet in both ground and underwater slots, it will remember if you switch it off and on again in only one of the slots.

example: I have Tails in both slots, I switch Tails for a panther in the ground slot, and then switch the panther for Tails again. It will remember the name.

Yeah, that must have been it. Lashtail devourer underwater op.

A.I. code is actually really hard to write. Its the most complex code that can be written. A.I. is also extremely CPU heavy with many condition statements. A.I. in games like Final Fantasy Reborn are not ordinary and have been developed over years (not talking 1 or 2 years I am talking 10-20 years of development). Also have to consider game mechanics such as stealth and porting that would drastically effect A.I. performance since you would have to find all possible conditions and solutions to each.

I can honestly say I could make an A.I. for GW2 that would be hard to beat but at the same time hardly any computer would be able to run it due to the complexity. A good A.I. would have to know literally every skill in the game and know when you used it last so that it can estimate the time necessary. GW2 game engine can not handle such aspects. Very few game engines can.

Witcher 2 REDengine is so far the only engine I know of that allows advanced A.I. that is capable of learning and countering. REDengine 3 engine will be a vast improvement, but it still isn’t beyond a zerg mentality.

In all games I have played each A.I. has weaknesses that are extremely easy to counter. GW2 A.I. is actually decent for all the mechanics needed to be countered by the A.I.

If you guys want to know something pathing is an NP-Hard to NP-Complete problem in computer science. If you don’t know what those stands for look it up lol. In this case the A.I. pathing you guys are wanting are NP-complete problem. On a game of this scale it would be extremely hard.

I am not saying their algorithms for pathing are perfect for the problem (they could be optimized more I bet but i can’t see the pathing code so i don’t know)

Thank god: someone who actually knows what they’re talking about. Thanks for that post +1

Also, rangers keep in mind: you are the “AI”. You can pull your pet back from aoe attacks, you can swap, you can recognize the danger.

Chopps.5047:

Thank god: someone who actually knows what they’re talking about. Thanks for that post +1

Also, rangers keep in mind: you are the “AI”. You can pull your pet back from aoe attacks, you can swap, you can recognize the danger.

Sure, so when the guy I thought was standing near me retreats into a bunch of AoE fields I didn’t see, tell me again how I can recall my pet? Or when the person I THINK is targeted isn’t the one I have targetted, or I mysteriously DROP target (happens a lot) and my pet is chasing someone out there… Or, I loose track of my pet because all of a sudden my health bar is dropping and I’m trying to figure out why, what thief has unstealthed on me. When it is time for me to start dodging around what’kittenting me and my poor pet is 1200 units away and my swap is down. Honestly, there are times when I just flat out die before my pet does.

I rarely find myself running solo, I’m usually with the zerg or in a guild group. As such, Zerg battles pets die or sometimes I die so fast it isn’t funny. In guild groups, pet AI work is easier as long as I’m not being focused. If I am, it is busy time, with dodging and rolling and hitting proper sword skills to worry about the pet. And, If I’m dodging, rolling and moving about chances are the pet isn’t hitting anything anyway, cause their chasing.

Suddenflame.2601:

A.I. code is actually really hard to write. Its the most complex code that can be written. A.I. is also extremely CPU heavy with many condition statements. A.I. in games like Final Fantasy Reborn are not ordinary and have been developed over years (not talking 1 or 2 years I am talking 10-20 years of development). Also have to consider game mechanics such as stealth and porting that would drastically effect A.I. performance since you would have to find all possible conditions and solutions to each.

I can honestly say I could make an A.I. for GW2 that would be hard to beat but at the same time hardly any computer would be able to run it due to the complexity. A good A.I. would have to know literally every skill in the game and know when you used it last so that it can estimate the time necessary. GW2 game engine can not handle such aspects. Very few game engines can.

Witcher 2 REDengine is so far the only engine I know of that allows advanced A.I. that is capable of learning and countering. REDengine 3 engine will be a vast improvement, but it still isn’t beyond a zerg mentality.

In all games I have played each A.I. has weaknesses that are extremely easy to counter. GW2 A.I. is actually decent for all the mechanics needed to be countered by the A.I.

If you guys want to know something pathing is an NP-Hard to NP-Complete problem in computer science. If you don’t know what those stands for look it up lol. In this case the A.I. pathing you guys are wanting are NP-complete problem. On a game of this scale it would be extremely hard.

I am not saying their algorithms for pathing are perfect for the problem (they could be optimized more I bet but i can’t see the pathing code so i don’t know)

That “lol” in your reply made me see your post less seriously…

However, what said is totally irrelevant because the game obviously had AI that works. The problem is that, as said before, pathfinding is not frequent enough.

Once Object A finds a path to Object B, it takes X time until Object A can “see” Object B again. In that X time, Object B has enough time to alter it’s position. In GW2, X is too big which gives us a noticable impact on gameplay.

Serraphin Storm.2369:

Pet A.I is meant to be imperfect. Having a set it and forget it pet isn’t what being a guildwars 2 ranger is about. Ranger are suppose to work in tandem with their pet. Not one or the other doing all the work.

I wouldn’t have a problem with that if they gave us more ways to control our pet. But right now it’s very limited control combined with a poor AI. The complete opposite of your “set it and forget it” scenario you’re warning against. Either extreme is bad.

Suddenflame.2601:

A.I. code is actually really hard to write. Its the most complex code that can be written. A.I. is also extremely CPU heavy with many condition statements.

And the solution to this is to have the AI run on the client computer. Just like your character is commanded by the player’s computer, your pet should be commanded by the player’s computer. It’ll just be commanded by a (more sophisticated) AI instead of keyboard inputs.

But that gets into the whole control issue. They don’t want us to control the pet, preferring it remain a mostly autonomous entity, yet at the same time they won’t give it decent AI. So they don’t want to put the AI on the client.

I’ve had reasonable luck in pve with pets connecting with targets, but more and more in wvw and pvp, a great deal of disappointment with pets. Early on before beastmaster was the thing, people ignored pets and pet attacks connected. Now beyond the pet nerf, players have wised-up to realize the pet is nothing if all they do is run around using normal combat speed. My pet is either traited for speed, under the effects of swiftness, or signet of the hunt and it rarely connects with players who are simply moving. I can occasionally get a hit with stealth on jaguar or if the player is immobilized, stunned, knock-downed, or backing up. If the player stealths as thieves, mesmers, and engineers do, the pet is totally lost and resets its position. I’ve traited down the beastmaster line for quickness, which again is lost on pets due to half of the 3 seconds the pet resets position before attacking.

Too much damage is divided with the pet, when the pet rarely connects. I’ve mentioned and I believe others have as well, but for the reason of programming, why not speed up pets. In GW, pets naturally moved 25% faster (not saying ai was better but connecting wasn’t an issue because most skills required stopping to activate), why not give them risen speed, and instant activation of F-skills. That would satisfy me. Like most every other class when I hit F2 the skill happens. For example, if I hit F3 the pet shadowsteps back rather than turning and running through all aoe fields. If that were the case, I could be the “active” player many people say is the root of the pet problems. If I jump down a ledge, I F3 to recall then F1 to attack again, instead of waiting the for the pet to find a path down a hill or slope. People might mention the necro has to put up with the same ai problems, but the necro and others aren’t tied to a pet as their class attribute.

Found this video on youtube, which shows the moving target issue.
http://www.youtube.com/watch?v=Qk5DiFJAtWQ

Solandri.9640:

Serraphin Storm.2369:

Pet A.I is meant to be imperfect. Having a set it and forget it pet isn’t what being a guildwars 2 ranger is about. Ranger are suppose to work in tandem with their pet. Not one or the other doing all the work.

I wouldn’t have a problem with that if they gave us more ways to control our pet. But right now it’s very limited control combined with a poor AI. The complete opposite of your “set it and forget it” scenario you’re warning against. Either extreme is bad.

Suddenflame.2601:

A.I. code is actually really hard to write. Its the most complex code that can be written. A.I. is also extremely CPU heavy with many condition statements.

And the solution to this is to have the AI run on the client computer. Just like your character is commanded by the player’s computer, your pet should be commanded by the player’s computer. It’ll just be commanded by a (more sophisticated) AI instead of keyboard inputs.

But that gets into the whole control issue. They don’t want us to control the pet, preferring it remain a mostly autonomous entity, yet at the same time they won’t give it decent AI. So they don’t want to put the AI on the client.

Your missing the point all together. The A.I. your wanting is by far too sophisticated and difficult to make. Your asking for Anet to make a program that is capable of doing path finding with multiple constantly changing variables with blockades (AOE) that can appear out of no where and disappear and able to calculate all that information and transmit it in a matter of seconds. Just making a static path finding is NP-Complete problem what your asking for is by far at our current technology level is fundamentally impossible.

You would need several super computers working together just to figure out the dynamic path finding along with predicting movements of humans (cause one of the targets is a human player) just to compute the information in time let along transmit several hundred GBs (possibly TB) of information.

They give control to us already. One button to attack your target and one to retreat. Use the one to retreat to tell your pet to get out of AOEs. Then tell him to reengage afterwards its not complicated.

Anet has stated they don’t want to give the players too much control since it would over-complicate the Ranger making it a lot harder to play. How many players you think are able to control 2 characters at the exact same time?

Edit: Your computer is not capable of handling such a CPU heavy task. What your asking for is a multithread program that would use 100% of your CPU for the entire time your game is running.

solrik.6028:

Suddenflame.2601:

A.I. code is actually really hard to write. Its the most complex code that can be written. A.I. is also extremely CPU heavy with many condition statements. A.I. in games like Final Fantasy Reborn are not ordinary and have been developed over years (not talking 1 or 2 years I am talking 10-20 years of development). Also have to consider game mechanics such as stealth and porting that would drastically effect A.I. performance since you would have to find all possible conditions and solutions to each.

I can honestly say I could make an A.I. for GW2 that would be hard to beat but at the same time hardly any computer would be able to run it due to the complexity. A good A.I. would have to know literally every skill in the game and know when you used it last so that it can estimate the time necessary. GW2 game engine can not handle such aspects. Very few game engines can.

Witcher 2 REDengine is so far the only engine I know of that allows advanced A.I. that is capable of learning and countering. REDengine 3 engine will be a vast improvement, but it still isn’t beyond a zerg mentality.

In all games I have played each A.I. has weaknesses that are extremely easy to counter. GW2 A.I. is actually decent for all the mechanics needed to be countered by the A.I.

If you guys want to know something pathing is an NP-Hard to NP-Complete problem in computer science. If you don’t know what those stands for look it up lol. In this case the A.I. pathing you guys are wanting are NP-complete problem. On a game of this scale it would be extremely hard.

I am not saying their algorithms for pathing are perfect for the problem (they could be optimized more I bet but i can’t see the pathing code so i don’t know)

That “lol” in your reply made me see your post less seriously…

However, what said is totally irrelevant because the game obviously had AI that works. The problem is that, as said before, pathfinding is not frequent enough.

Once Object A finds a path to Object B, it takes X time until Object A can “see” Object B again. In that X time, Object B has enough time to alter it’s position. In GW2, X is too big which gives us a noticable impact on gameplay.

Path finding is a CPU heavy program. If you have read the other posts then you would know that GW2 is already a CPU heavy game. By making path finding perform more it takes a lot more bandwidth and CPU power. Which increases stress on the servers and on your own computer. Path Finding from what i seen is done very frequently, but pet misses not do to path finding but because of animation times which is meant to give the enemy player enough time to move out of it.

My pet does not miss his target unless the animation is already going and the enemy moved. At which point pathfinding is not the issue.

Kilrik.6320:

Found this video on youtube, which shows the moving target issue.
http://www.youtube.com/watch?v=Qk5DiFJAtWQ

Path finding is not the issue in this. Pet reached the target, but the target is moving during the animation and the attacked missed. Easy way to solve the problem is lower the animation times, but the animation is there to give time for the enemy to “dodge” the attack by moving out of it. Or another situation is to allow the pets to move while in animation like a player. Similar situation is GS warriors. You can stand there and take the 100 blades attack, or you just move out of it.

Pathfinding issues /= standing still animation issue.

I don’t believe I called it pathing here, but hitting a target is the problem I was mentioning. Perhaps animations could be sped up or skills could be made instant. I can swing a sword while moving, why can’t my pet. I understand the “100 Blades” concept, we have “Whirling Defense,” but not all skills are stationary, so why should it be any different for our pets. F2’s are stationary fine, but why are all the other skills performed by pets “stop and attack” style.

Additionally…
http://www.youtube.com/watch?v=77D8fI2Ow4E

Kilrik.6320:

I don’t believe I called it pathing here, but hitting a target is the problem I was mentioning. Perhaps animations could be sped up or skills could be made instant. I can swing a sword while moving, why can’t my pet. I understand the “100 Blades” concept, we have “Whirling Defense,” but not all skills are stationary, so why should it be any different for our pets. F2’s are stationary fine, but why are all the other skills performed by pets “stop and attack” style.

Additionally…
http://www.youtube.com/watch?v=77D8fI2Ow4E

Ok that video correctly indicates the issue that pets suffer from. It is also viewable through the use of transformation potions and use the Custum brawl

> but why are all the other skills performed by pets “stop and attack” style.

They’re not. Use dogs or spiders. Look at this graph:

https://dviw3bl0enbyw.cloudfront.net/uploads/forum_attachment/file/48266/Ranger_Pet_DPS_Comparison_NEW_1.jpg

Some pets (birds and cats for example) lose a TON of attack DPS from chasing moving targets.

This is intentional and something ANET gave us to build around. Think about that. Also, spider DPS is not strongly function of moving targets and looks pretty good against them (bottom right).

How do you know you’re going to be fighting a moving target? A stationary target? A hidden target? Chopps every answer can’t be… Easy use this pet in this situation… We only get 2 up. Glad you can anticipate what fight you’re going to have before you have it.

Suddenflame.2601:

Solandri.9640:

Serraphin Storm.2369:

Pet A.I is meant to be imperfect. Having a set it and forget it pet isn’t what being a guildwars 2 ranger is about. Ranger are suppose to work in tandem with their pet. Not one or the other doing all the work.

I wouldn’t have a problem with that if they gave us more ways to control our pet. But right now it’s very limited control combined with a poor AI. The complete opposite of your “set it and forget it” scenario you’re warning against. Either extreme is bad.

Suddenflame.2601:

A.I. code is actually really hard to write. Its the most complex code that can be written. A.I. is also extremely CPU heavy with many condition statements.

And the solution to this is to have the AI run on the client computer. Just like your character is commanded by the player’s computer, your pet should be commanded by the player’s computer. It’ll just be commanded by a (more sophisticated) AI instead of keyboard inputs.

But that gets into the whole control issue. They don’t want us to control the pet, preferring it remain a mostly autonomous entity, yet at the same time they won’t give it decent AI. So they don’t want to put the AI on the client.

Your missing the point all together. The A.I. your wanting is by far too sophisticated and difficult to make. Your asking for Anet to make a program that is capable of doing path finding with multiple constantly changing variables with blockades (AOE) that can appear out of no where and disappear and able to calculate all that information and transmit it in a matter of seconds. Just making a static path finding is NP-Complete problem what your asking for is by far at our current technology level is fundamentally impossible.

You would need several super computers working together just to figure out the dynamic path finding along with predicting movements of humans (cause one of the targets is a human player) just to compute the information in time let along transmit several hundred GBs (possibly TB) of information.

They give control to us already. One button to attack your target and one to retreat. Use the one to retreat to tell your pet to get out of AOEs. Then tell him to reengage afterwards its not complicated.

Anet has stated they don’t want to give the players too much control since it would over-complicate the Ranger making it a lot harder to play. How many players you think are able to control 2 characters at the exact same time?

Edit: Your computer is not capable of handling such a CPU heavy task. What your asking for is a multithread program that would use 100% of your CPU for the entire time your game is running.

Yet my PC is able to simulate AI for 500 soldiers in Warband, or thousands of units in the Total War games. Even half life 2 can have better AI with barely any cost.

What you are saying is almost fully irrelevant because the AI our pets need is not as advanced as (say) Arma 3. Arma 3’s AI has all sort of things like reaction time, accuracy, ‘logic’ for when to change stances.

GW2’s pet AI is only an algorithm for finding the path and then a bunch of if statements that trigger animations and effects.

Basically,

findPath(target);
if(range(target) == 400) skills();
————————
skills…(..)
{
//determines which skills is used based on if statements
//e.g. use the skill that immobilizes if the target is moving around or
//use the skill that heals if “this” is on low health
}

findPath…(..)
{
//this one also has a bunch of if statements. This can be made in hundreds of different ways but if each object has it’s own pathfinder assigned to it then the pathfinder can keep track of changes.
}

I could write all kind of examples but that would take all day.
MY POINT IS THIS:
Gw2 already has AI implemented. The problem with it is that it’s not used frequently enough. There is at least 500ms delay before the pet realises that something is now blocking it’s would-be path. The solution to this can be and is not limited to doing parts of the calculations on the client while finishing them on the server.

Chopps.5047:

> but why are all the other skills performed by pets “stop and attack” style.

They’re not. Use dogs or spiders. Look at this graph:

https://dviw3bl0enbyw.cloudfront.net/uploads/forum_attachment/file/48266/Ranger_Pet_DPS_Comparison_NEW_1.jpg

Some pets (birds and cats for example) lose a TON of attack DPS from chasing moving targets.

This is intentional and something ANET gave us to build around. Think about that. Also, spider DPS is not strongly function of moving targets and looks pretty good against them (bottom right).

They lose a tonne but they still do the most according to those charts.

NargofWoV.4267:

How do you know you’re going to be fighting a moving target? A stationary target? A hidden target? Chopps every answer can’t be… Easy use this pet in this situation… We only get 2 up. Glad you can anticipate what fight you’re going to have before you have it.

You can actually. Your enemy will most likely be moving in WvW but it’s the opposite in PvE where the ai doesn’t even try to avoid your attacks. In tPvp, holding points is important and that creates a mix of the above two.

pets will suck in a lot of situations. in wvw they are useless. i hope they will give us pets without Ai. whisp like things whats follow us closely not dealing or receiving dmg just buffing us and/or the party.

So my question is this. If the AI required to hit a moving target is so complex and CPU intensive, how come my EQ and EQ2 Beastlords and Shaman, my Vanguard Shaman, my FFXI Beastlord and other pet classes I’ve played over the last 15 years never had such a problem?

Go back to EQ in 1999 and you could type /follow to put your character on cruise control and follow anyone anywhere as long as they didn’t get you stuck on a tree somewhere. You were always spaced directly a foot or so behind your /follow target and never got confused over which direction they may be heading, and barring any speed buff mismatches or terrain malfunctions that got you stuck you could stay that way for an indefinite amount of time.

Fast forward to 2013 and GW2, my pet can’t even follow me around out of combat without charging off in random directions as it tries to follow and perceive and react to my movement as I run across the world.
http://www.youtube.com/watch?feature=player_detailpage&v=8ROD17gjtsE
If it can’t even follow me properly, how does it ever have a hope of following a moving enemy? Compound that with the attack delays demonstrated in Kilirk’s video above, our pets don’t stand a chance.

In the games I mentioned above as well as others, pets also attacked as they moved. If you have interest in seeing them, go google pvp fight videos from them and watch how pets consistently hit moving targets while moving themselves.

Some games also gave you control of several pet skills as well as full pet UI’s that showed health, conditions, gave many pet commands (or allowed custom binding of commands using macros) and it was never a learning curve or control problem for any of the however many hundreds of thousands of players that played them, so again I call BS on the learning curve argument that ANet puts forth on the pet system here.

You can defend it all you like, but the fact of the matter is that the pet system they have implemented is not up to par with games developed a decade plus before it, is not able to keep pace with human opponents, and is not able to compete properly against the content that they themselves are introducing into their game. If it can’t be brought up to par – for ANY reason – it should be replaced with a completely separate mechanic that can compete on a level playing field. Making excuses about how it is hard to do what the players want, or how it works great in controlled circumstances is total crap.

Also, some of the AI required to hit moving targets is actually present in the game already. For example, compare drakes to veteran ice worms. Drakes stop, charge an attack facing whatever direction you happened to be when they started and never think about changing until the attack is complete. Ice works begin channeling their skills facing wherever you were, turn to wherever you are when they are done, and then throw their rocks which then heat seek towards you and hit you on the run unless you actually evade or run out of range. ALL ranged NPC attacks with any kind of cast timer should do this by default game wide, not just our pets. But ESPECIALLY our pets that are expected to face other players should AT LEAST be given this advantage. Melee range attacks should happen without stopping the pets movement, unless they literally have zero activation time.

NargofWoV.4267:

How do you know you’re going to be fighting a moving target? A stationary target? A hidden target? Chopps every answer can’t be… Easy use this pet in this situation… We only get 2 up. Glad you can anticipate what fight you’re going to have before you have it.

Easy. You create the conditions for it. That’s part of the reason we get abilities like Muddy Terrain or Frost Trap. Or a cripple on sword autoattack or shortbow. Or a trait like agility training. These things help cats and birds land hits and creates interplay between the ranger and the pet. So clearly we’re in beastmaster/regen territory here.

If you’re running rabid bleeds, you probably want a drakehound and a spider to immobilize which gives you a means to control stun warriors (also muddy terrain). If you’re running traps with rabid, hyenas are great for multiple knockdown in a trap. The paralyze spider’s immobilize is very good.

What about zerging? Drakes really aren’t too bad, neither are moas or bears. Hopefully that situation gets a buff.

What about dungeons? It depends. Need more fury? Red moa. (Gain in group dps from fury boom is much greater than the loss in dps from switching to moa). I like moas in general because of the healing skill in dungeons, that’s nice for melee players. But a lot of times, especially in early dungeon designs, bosses are stationary so for practiced groups birds are a no-brainer because DPS is king in that situation.

What if you’re a new ranger and can’t micro pets well? Devourer. Devourer is great for the experienced ranger too: you can stand behind it and dps while it body blocks projectiles. I do this all the time.

It’s very clear: choose the right pet for the job.

Khayoss.2019:

So my question is this. If the AI required to hit a moving target is so complex and CPU intensive, how come my EQ and EQ2 Beastlords and Shaman, my Vanguard Shaman, my FFXI Beastlord and other pet classes I’ve played over the last 15 years never had such a problem?

Go back to EQ in 1999 and you could type /follow to put your character on cruise control and follow anyone anywhere as long as they didn’t get you stuck on a tree somewhere. You were always spaced directly a foot or so behind your /follow target and never got confused over which direction they may be heading, and barring any speed buff mismatches or terrain malfunctions that got you stuck you could stay that way for an indefinite amount of time.

Fast forward to 2013 and GW2, my pet can’t even follow me around out of combat without charging off in random directions as it tries to follow and perceive and react to my movement as I run across the world.
http://www.youtube.com/watch?feature=player_detailpage&v=8ROD17gjtsE
If it can’t even follow me properly, how does it ever have a hope of following a moving enemy? Compound that with the attack delays demonstrated in Kilirk’s video above, our pets don’t stand a chance.

In the games I mentioned above as well as others, pets also attacked as they moved. If you have interest in seeing them, go google pvp fight videos from them and watch how pets consistently hit moving targets while moving themselves.

Some games also gave you control of several pet skills as well as full pet UI’s that showed health, conditions, gave many pet commands (or allowed custom binding of commands using macros) and it was never a learning curve or control problem for any of the however many hundreds of thousands of players that played them, so again I call BS on the learning curve argument that ANet puts forth on the pet system here.

You can defend it all you like, but the fact of the matter is that the pet system they have implemented is not up to par with games developed a decade plus before it, is not able to keep pace with human opponents, and is not able to compete properly against the content that they themselves are introducing into their game. If it can’t be brought up to par – for ANY reason – it should be replaced with a completely separate mechanic that can compete on a level playing field. Making excuses about how it is hard to do what the players want, or how it works great in controlled circumstances is total crap.

Also, some of the AI required to hit moving targets is actually present in the game already. For example, compare drakes to veteran ice worms. Drakes stop, charge an attack facing whatever direction you happened to be when they started and never think about changing until the attack is complete. Ice works begin channeling their skills facing wherever you were, turn to wherever you are when they are done, and then throw their rocks which then heat seek towards you and hit you on the run unless you actually evade or run out of range. ALL ranged NPC attacks with any kind of cast timer should do this by default game wide, not just our pets. But ESPECIALLY our pets that are expected to face other players should AT LEAST be given this advantage. Melee range attacks should happen without stopping the pets movement, unless they literally have zero activation time.

Read my post above this: cats and birds will never hit moving targets because they’re intentionally designed not to. I go into detail as to “why”.

If I remember correctly… Anet never said it would be too difficult to give us better pet AI, but that it wouldn’t be “balanced” or “fair” if they gave us better pet AI.

Really, from what I find while playing GW2 it isn’t that we need some insanely better AI or anything, but that we need more precise control over our pets actions to keep them alive better. Every other class has a ton more control over their mechanic; why are the rangers left out?

When I call my pet back to me why does it run away, then off to the side, then around in circles; all the while running through every AOE possible before it just can’t take it anymore and dies, before actually making it back to me?

Why do I often have to spam my attack my target and return to me keys just to get the stupid thing to run back to me or attack the kitten target?

Why does it seemingly purposely stand in that one spot, in that huge radius around you that it could be anywhere within while on standby, that it gets hit by the AOE just perfectly and dies? I mean really my only other option is to have it “attack my target” in which there is a very high likelihood that there are AOEs in front of me in which it would run through anyways and still die…

There is a serious problem with the ranger mechanic when compared to the other classes mechanics… some people are just in denial. Sure you can call your pet back to you…

• IF there aren’t any AOE feilds in the way
• IF your pet decides to run straight to you instead of going on tour visiting as many AOE fields as it can before dying
• IF it actually comes close enough in that rather large radius around you that is considered “returning” to you…

• IF you don’t have it traited as a consumable resource to dump your conditions onto (because there really isn’t any other way for rangers to really effectively remove enough conditions to keep up with the condition spam that goes on)

… which is another thing that just irritates me in general about pets: it seems the Devs can’t make up their minds about what the pet “mechanic” actually is… They seem to want us to keep it alive as a sustained resource in order to keep up in the damage department, but then they go essentially force us to use them as a consumable resource to use up rather than sustain.

Try this:

Go to a higher level than where your pet actually stands. Jump on a lift where you can’t get up without jumping on. If it doesn’t follow you the way up, use F2 – works “best” with the dogs or the wolfs howl, and with “best” I mean it won’t…

solrik.6028:

Yet my PC is able to simulate AI for 500 soldiers in Warband, or thousands of units in the Total War games. Even half life 2 can have better AI with barely any cost.

What you are saying is almost fully irrelevant because the AI our pets need is not as advanced as (say) Arma 3. Arma 3’s AI has all sort of things like reaction time, accuracy, ‘logic’ for when to change stances.

GW2’s pet AI is only an algorithm for finding the path and then a bunch of if statements that trigger animations and effects.

Basically,

findPath(target);
if(range(target) == 400) skills();
————————
skills…(..)
{
//determines which skills is used based on if statements
//e.g. use the skill that immobilizes if the target is moving around or
//use the skill that heals if “this” is on low health
}

findPath…(..)
{
//this one also has a bunch of if statements. This can be made in hundreds of different ways but if each object has it’s own pathfinder assigned to it then the pathfinder can keep track of changes.
}

I could write all kind of examples but that would take all day.
MY POINT IS THIS:
Gw2 already has AI implemented. The problem with it is that it’s not used frequently enough. There is at least 500ms delay before the pet realises that something is now blocking it’s would-be path. The solution to this can be and is not limited to doing parts of the calculations on the client while finishing them on the server.

Your comparing 3 different types of AIs which none of them are similar in the least. Such as I can if you really wanted give you full detail how an RTS AI works and its not how you think it works. GW2 has multiple AIs in it already not just pet AI which currently is the most simplistic out of all of them to lower bandwidth usage. If you want to know something the pet AI is on the client side its not a server side function. There isn’t a 500ms delay do to path finding not being done enough, but there is a delay for how long it takes to do the calculation.

I can tell you this right now that you fail to realize. Path Finding is an NP-Complete algorithm. Look it up since you have proven you do not know what it means. Finding a path from point A to B that is not a straight line is not easy to do in programming. Since it is an NP-complete problem, Computer scientists have come up with approximation algorithms that run faster, but are by far less accurate.

Your solution your wanting is impossible by now days unless you can prove that NP = P. Since your wanting an NP-Complete algorithm to run constantly just so you can have flawless pathfinding in a dynamic situation. A dynamic NP-Complete algorithm would consume all your processor. I have stated this more then once and told you guys to look it up if you do not know what NP-complete means.

edit: Your line of Findpath() is what takes the most cause that is calling a method which could be 1000 lines of code for all we know about it.

I’m enjoying people defend the kittenty pet AI while I’m destroying people now on my mesmer, thief and guardian class compared to my ranger which I no longer play because of it’s lack of effectiveness in wvw compared to other classes. Oh the irony.

Sorry, but delusional people play ranger, then they play another class, and realize the world of difference between the two. It’s not just pet AI that’s the issue with rangers, it’s utility , weapon synergy and coefficients.

Things would probably be much better if they used an AI similar to what UT2004 had with path nodes baked into the map. That game really had good AI in my opinion. It seemed to hold up well with a large number of bots in the map too. It only bogged down when you had 500+ enemies and a few thousand rockets with limited tracking capability.

However, what we have now is probably what we’re stuck with. To make the best of it, I think they should let the pets cheat a little. Some of this has already been mentioned, but basically they should be allowed to teleport to their destinations, and have their F2 behavior looked at.

If a pet is sent to attack something, the game should figure out how long it should take to reach the target and teleport the pet to the target if it hasn’t reached it in 125% of the optimal time. F3 is easier. Just teleport the pet back.

For the F2 abilities they should consider classing the abilities as offensive or support and use that to decide whether the ability should be used at the current position or if the pet needs to move into range of the selected target. It would be even better if the pet would move to a targeted ally to use support abilities.

I usually put on some Benny Hill when this happens

Chopps.5047:

Read my post above this: cats and birds will never hit moving targets because they’re intentionally designed not to. I go into detail as to “why”.

Can you give me a reference to ANet stating this in some form or fashion? Or is this opinion based on observation of numbers and effectiveness?

That’s pretty funny. Actually that’s how I stay away from Hillbillies in SAB. I run at them when they’re running at me. They get confused and keep chasing me as I keep running past them.

Khayoss.2019:

Chopps.5047:

Read my post above this: cats and birds will never hit moving targets because they’re intentionally designed not to. I go into detail as to “why”.

Can you give me a reference to ANet stating this in some form or fashion? Or is this opinion based on observation of numbers and effectiveness?

Its based on observation based on the fact all animals in the game use stationary attacks. Wargs, birds, wolves, boars, ectcan’t attack without stopping first. The player when transformed into animals as well can’t move and attack at the same time. It is a design choice. Since pets are animals the follow the same rules.

Suddenflame.2601:

It is a design choice.

And a really, really crappy design choice it is, for a pet based class.

Anet could quite easily fix many of the issues with pets hitting moving targets. I’d say most of the issues are due to the eye-candy – having animations that looks cool, but are hilariously inefficient because of the fact that they are so incredibly slow to execute – just look at birds, the no#1 contenders for first place when it comes to “the pet that is the worst of all to hit moving targets” – just look at their animations for crying out loud!

For ranged pets, the only issue really is the slow projectile speeds they suffer from. Increase their projectile speed and they will probably be just fine/better.

As for survivability issues like aoe fields that pets have to run through when recalled from danger, that could easily be resolved as well by allowing pets to shadowstep back. Of course that would bring other [balancing]issues, but I’m sure there would be ways to balance that as well.

OGDeadHead.8326:

Suddenflame.2601:

It is a design choice.

And a really, really crappy design choice it is, for a pet based class.

Anet could quite easily fix many of the issues with pets hitting moving targets. I’d say most of the issues are due to the eye-candy – having animations that looks cool, but are hilariously inefficient because of the fact that they are so incredibly slow to execute – just look at birds, the no#1 contenders for first place when it comes to “the pet that is the worst of all to hit moving targets” – just look at their animations for crying out loud!

For ranged pets, the only issue really is the slow projectile speeds they suffer from. Increase their projectile speed and they will probably be just fine/better.

As for survivability issues like aoe fields that pets have to run through when recalled from danger, that could easily be resolved as well by allowing pets to shadowstep back. Of course that would bring other [balancing]issues, but I’m sure there would be ways to balance that as well.

If they allow pets to attack on the move they would also have to allow all other animals attack on the move just to be consistent. Pets are not meant to be the main source of damage. They allow your enemy to escape from the damage. It would be unbalanced to have damage appear from 2 directions and 1 being unavoidable. Live with the design choice especially since 40% of our damage does not come from our pets. From what i seen its more of an added bonus. Also to escape an AOE field you have to walk through it as well (unless you use a dodge).

Suddenflame.2601:

If they allow pets to attack on the move they would also have to allow all other animals attack on the move just to be consistent.

Uhm, no?

Suddenflame.2601:

It would be unbalanced to have damage appear from 2 directions and 1 being unavoidable.

Yes, and not really what I’m asking for either. But instead of missing, what, maybe 80% of their attacks vs a moving target, it would be enough if they missed half of that, at most?

Suddenflame.2601:

Also to escape an AOE field you have to walk through it as well (unless you use a dodge).

Exactly – unless you dodge.

Suddenflame.2601:

OGDeadHead.8326:

Suddenflame.2601:

It is a design choice.

And a really, really crappy design choice it is, for a pet based class.

Anet could quite easily fix many of the issues with pets hitting moving targets. I’d say most of the issues are due to the eye-candy – having animations that looks cool, but are hilariously inefficient because of the fact that they are so incredibly slow to execute – just look at birds, the no#1 contenders for first place when it comes to “the pet that is the worst of all to hit moving targets” – just look at their animations for crying out loud!

For ranged pets, the only issue really is the slow projectile speeds they suffer from. Increase their projectile speed and they will probably be just fine/better.

As for survivability issues like aoe fields that pets have to run through when recalled from danger, that could easily be resolved as well by allowing pets to shadowstep back. Of course that would bring other [balancing]issues, but I’m sure there would be ways to balance that as well.

If they allow pets to attack on the move they would also have to allow all other animals attack on the move just to be consistent. Pets are not meant to be the main source of damage. They allow your enemy to escape from the damage. It would be unbalanced to have damage appear from 2 directions and 1 being unavoidable. Live with the design choice especially since 40% of our damage does not come from our pets. From what i seen its more of an added bonus. Also to escape an AOE field you have to walk through it as well (unless you use a dodge).

“They allow your enemy to escape from the damage. It would be unbalanced to have damage appear from 2 directions and 1 being unavoidable. "

You’re kittening stupid.

There’s a lot of declarative statements in this thread.

Personally, I feel that the AI is lacking in some cases such as pet turning, and if that is a design choice it has very little logic behind it.

Regardless I would love to see some more design iteration on pets!

Ryan.8367:

Suddenflame.2601:

OGDeadHead.8326:

Suddenflame.2601:

It is a design choice.

And a really, really crappy design choice it is, for a pet based class.

Anet could quite easily fix many of the issues with pets hitting moving targets. I’d say most of the issues are due to the eye-candy – having animations that looks cool, but are hilariously inefficient because of the fact that they are so incredibly slow to execute – just look at birds, the no#1 contenders for first place when it comes to “the pet that is the worst of all to hit moving targets” – just look at their animations for crying out loud!

For ranged pets, the only issue really is the slow projectile speeds they suffer from. Increase their projectile speed and they will probably be just fine/better.

As for survivability issues like aoe fields that pets have to run through when recalled from danger, that could easily be resolved as well by allowing pets to shadowstep back. Of course that would bring other [balancing]issues, but I’m sure there would be ways to balance that as well.

If they allow pets to attack on the move they would also have to allow all other animals attack on the move just to be consistent. Pets are not meant to be the main source of damage. They allow your enemy to escape from the damage. It would be unbalanced to have damage appear from 2 directions and 1 being unavoidable. Live with the design choice especially since 40% of our damage does not come from our pets. From what i seen its more of an added bonus. Also to escape an AOE field you have to walk through it as well (unless you use a dodge).

“They allow your enemy to escape from the damage. It would be unbalanced to have damage appear from 2 directions and 1 being unavoidable. "

You’re kittening stupid.

Personal insults don’t help your position rather they weaken them. Explain why it is stupid in a rational manner before your view is valid. Your wanting a pet attacks to be unavoidable when they should be able to be avoided. Why should other classes not be allowed to avoid the damage?

40% of the ranger damage does not come from the Pet as some people have brought the math into it. Pets have a set damage based on the level. So that means the pet is a form of bonus support such as other classes mechanics. Such as mesmers clones normally don’t do damage but are a strong support skill. Thieves have the ability to steal and initiative which in both cases don’t deal damage but support the class.

Our pet is not there for pure damage but rather as a form of support. Such as our pet has a F2 skill that has a unique ability with a skill set and can be casted while the enemy is not aware of the pet. The pet also by extension is our ability to control. Such as the spider snaring our enemy or the bear removing conditions and being a damage soaker. Our pet is also able to save us from life threatening situations such as being downed underwater. We are the only class that while downed in water is still a threat. Hell I took on 4 people in underwater combat and still killed 1 even thou I was downed the entire time just because my pet soaked the damage while rezing me. No other class has that function.

Ryan.8367:

I’m enjoying people defend the kittenty pet AI while I’m destroying people now on my mesmer, thief and guardian class compared to my ranger which I no longer play because of it’s lack of effectiveness in wvw compared to other classes. Oh the irony.

Sorry, but delusional people play ranger, then they play another class, and realize the world of difference between the two. It’s not just pet AI that’s the issue with rangers, it’s utility , weapon synergy and coefficients.

I have been looking for someone to spar with. Maybe you can help me test my ranger builds in the field?

Chopps.5047:

Ryan.8367:

I’m enjoying people defend the kittenty pet AI while I’m destroying people now on my mesmer, thief and guardian class compared to my ranger which I no longer play because of it’s lack of effectiveness in wvw compared to other classes. Oh the irony.

Sorry, but delusional people play ranger, then they play another class, and realize the world of difference between the two. It’s not just pet AI that’s the issue with rangers, it’s utility , weapon synergy and coefficients.

I have been looking for someone to spar with. Maybe you can help me test my ranger builds in the field?

Sure, ill duel you on my ranger

Whisper Tanbin

Suddenflame.2601:

Ryan.8367:

Suddenflame.2601:

OGDeadHead.8326:

Suddenflame.2601:

It is a design choice.

And a really, really crappy design choice it is, for a pet based class.

Anet could quite easily fix many of the issues with pets hitting moving targets. I’d say most of the issues are due to the eye-candy – having animations that looks cool, but are hilariously inefficient because of the fact that they are so incredibly slow to execute – just look at birds, the no#1 contenders for first place when it comes to “the pet that is the worst of all to hit moving targets” – just look at their animations for crying out loud!

For ranged pets, the only issue really is the slow projectile speeds they suffer from. Increase their projectile speed and they will probably be just fine/better.

As for survivability issues like aoe fields that pets have to run through when recalled from danger, that could easily be resolved as well by allowing pets to shadowstep back. Of course that would bring other [balancing]issues, but I’m sure there would be ways to balance that as well.

If they allow pets to attack on the move they would also have to allow all other animals attack on the move just to be consistent. Pets are not meant to be the main source of damage. They allow your enemy to escape from the damage. It would be unbalanced to have damage appear from 2 directions and 1 being unavoidable. Live with the design choice especially since 40% of our damage does not come from our pets. From what i seen its more of an added bonus. Also to escape an AOE field you have to walk through it as well (unless you use a dodge).

“They allow your enemy to escape from the damage. It would be unbalanced to have damage appear from 2 directions and 1 being unavoidable. "

You’re kittening stupid.

Personal insults don’t help your position rather they weaken them. Explain why it is stupid in a rational manner before your view is valid. Your wanting a pet attacks to be unavoidable when they should be able to be avoided. Why should other classes not be allowed to avoid the damage?

40% of the ranger damage does not come from the Pet as some people have brought the math into it. Pets have a set damage based on the level. So that means the pet is a form of bonus support such as other classes mechanics. Such as mesmers clones normally don’t do damage but are a strong support skill. Thieves have the ability to steal and initiative which in both cases don’t deal damage but support the class.

Our pet is not there for pure damage but rather as a form of support. Such as our pet has a F2 skill that has a unique ability with a skill set and can be casted while the enemy is not aware of the pet. The pet also by extension is our ability to control. Such as the spider snaring our enemy or the bear removing conditions and being a damage soaker. Our pet is also able to save us from life threatening situations such as being downed underwater. We are the only class that while downed in water is still a threat. Hell I took on 4 people in underwater combat and still killed 1 even thou I was downed the entire time just because my pet soaked the damage while rezing me. No other class has that function.

So your explanation of the shoddy AI is that we’re able to survive underwater and that pets are an extension of us. Of course pets are an extension of us, they’re a very poorly designed extension of us is the issue. There’s a reason why F2 doesn’t work half the time on your pet, it’s poorly written, such as the rest of the pet AI in general. Anybody defending the pet AI on this game needs their head thoroughly examined. Yes, making comments like ‘our pets are poorly written is so that enemy players can avoid damage’ needs to be told their ideas are kittening stupid because that is some dumb kitten.

Ryan.8367:

Suddenflame.2601:

Ryan.8367:

“They allow your enemy to escape from the damage. It would be unbalanced to have damage appear from 2 directions and 1 being unavoidable. "

You’re kittening kittenupid.

Personal insults don’t help your position rather they weaken them. Explain why it is kittenupid in a rational manner before your view is valid. Your wanting a pet attackkitteno be unavoidable when they should be able to be avoided. Why should other classes not be allowed to avoid the damage?

40% of the ranger damage does not come from the Pet as some people have brought the math into it. Pets have a set damage based on the level. So that meankittenhe pet is a form of bonus support such as other classes mechanics. Such as mesmers clones normally don’t do damage but are a kittenrong support skill. Thieves have the ability to kitteneal and initiative which in both cases don’t deal damage but support the class.

Our pet is not there for pure damage but rather as a form of support. Such as our pet has a F2 skill that has a unique ability with a skill set and can be cakittened while the enemy is not aware of the pet. The pet also by extension is our ability to control. Such akittenhe spider snaring our enemy or the bear removing conditions and being a damage soaker. Our pet is also able to save us from life threatening situations such as being downed underwater. We are the only clakittenhat while downed in water is kittenill a threat. Hell I took on 4 people in underwater combat and kittenill killed 1 even thou I was downed the entire time jukitten because my pet soaked the damage while rezing me. No other class hakittenhat function.

So your explanation of the shoddy AI ikittenhat we’re able to survive underwater and that pets are an extension of us. Of course pets are an extension of ukittenhey’re a very poorly designed extension of us ikittenhe issue. There’s a reason why F2 doesn’t work half the time on your pet, it’s poorly written, such akittenhe rekitten of the pet AI in general. Anybody defending the pet AI on this game needkittenheir head thoroughly examined. Yes, making comments like ‘our pets are poorly written is so that enemy players can avoid damage’ needkitteno be told their ideas are kittening kittenupid because that is some dumb kitten.

As Chopp and I pointed out earlier the pet AI is fine. Your failure is to understand what AI stands for. Let me educate you on the issue with pets and what AI is. AI is artificial intelligence which covers combat scenarios and path finding. The only AI issue people have brought up is the Pets inability to dodge AOE which by all accounts is a design choice to make the Ranger actually do something such as tell your own pet to move out of it.

The issue that people are having thou is that the pet stops to attack which is an animation not an AI issue. It is a chosen design choice considering all animals in the game stop to attack even when the player is turned into an animal. This animation is what allows the pet to miss since the pet has to stop moving, has an animation cast time, then damage is applied. The enemy player moves out of the range before the damage is applied. As I stated in one my earlier posts 2 options can be done to fix this, but it must be done to all animals in the game as well.

1.) Lower animation cast time
2.) Allow animation on the move

Which ever is done it must be done across the board which also means all other animals would get the same treatment not just pets. Which means largely more player deaths overall and harder time outrunning some animals in PVE. It is an Anet design choice you may not like it but I have learned to live with it and control my pet.

I’m sorry, but I can’t agree that pet AI is fine. Maybe it is for PvP and roaming WvW, but in zerg-based WvW and PvE you will not have enough control over the pet to override it’s AoE death wish.

The pet AI has serious issues. Especially as more and more PvE content is gaining mechanics that require dodging/evading to survive and pets are incapable of that.