Guild Wars 2

Vision? Discouraging zergs?

It seems like all of the decisions made by Arenanet are only encouraging the safety in numbers mentality.

I don’t understand some of the things they are doing. A PTS would be nice… As it stands it feels like the players don’t have much of a voice.

There’s nothing you can really do to prevent zerging because the entire aspect of WvW is designed around it.

sPvP has the same problem with “mini-zerging” as well.

Hope this fixes culling. The complaining on the thief boards is getting grating.

ErraticFaith.9142:

The majority run in zerg format and the changes address that much. Culling wise, for the short term its a step forward. WvW isnt ment to be small groups vs the zerg, mostly they are just zerg fodder. Its supposed to be huge armies. The people roamers like to prey on are weak 1v1 often because they arent optimised for it.

Structured is where anyone with small group pvp and ‘actual’ skill should be, not complaining about culling changes for a small demographic most dont care about.

Have you played structured? It’s one mode on like 4 maps… No real Mmr or ranks, one can only do that for so long.

WvW is much more dynamic and fun right now. (IMO)

Habib Loew.6239:

Hi all, I think a little more explanation is in order.

First, let’s call the culling system that we’re switching to “affinity culling” just to make it easier to talk about. Under affinity culling the system handles enemies and allies independently. This means that the maximum number of allies that you can see under affinity culling is 1/2 the maximum number of characters (combined enemy & ally) that you could see under the original culling. Ditto for enemies. In exchange for that cost he benefit that we get is that running with (or through, or past) a large group of allies (e.g. a guild) won’t prevent you from seeing the enemies who are closest to you.

…

Thanks for the details.

How are the lists organized? My impression back in december was that both the enemy and the friendlies list are FIFOs. I was part of a zerg that did a golem rush on Overlook and while we were running from SM to Overlook the golems would repeatedly appear and disappear for me despite the fact that I was only maybe 5m behind them. The same thing happened when we reached the gate and it was very irritating that they appeared sometimes for me in front of the gate and a couple seconds later they were gone from my view. This is the behavior that I would expect if the friendlies list is a FIFO because then a new ally that enters my visibility area would kick off the front most entry from the queue which in my case was the golem(s).

In any case, I think that the friendlies list should be a priority queue where the priority is based on the something like the following criterias:

a) commanders.

b) mobile siege equipment aka golem since they can have a big impact on the outcome of a battle.

c) siege equipment operators. It is important to be able to see them in order to provide heal and protection.

d) people that are in the same group as I am.

e) people that are in the same guild as I am.

f) people that are in the same squad as I am.

g) anyone else.

Now the question is whether the enemy list should use a similar priority system. I would say no because this would mean that enemy commanders, golems, siege operators, etc would de facto always be visible and consequently focused fired down in no time. Instead the enemy list should be a simple FIFO to ensure more fairness.

Great news! Culling needs to be fixed one step at a time.

Sunreva.8714:

For me and my friends, ‘affinity’ culling resulted in the following:

3 of us in the middle of no where, away from any action. We see a 3 shadows on the ground approaching us (no character models have loaded yet, no nametags). As the shadows get to about 1200 distance away, the character models finally load in, still no nametags. They proceed to run by us, not realizing that we are enemies (they must of been having the same culling issues). As they run by us, the nametag loads in, and low and behold, they’re enemies.

It was this bad. I dread the thought of this returning.

This. was. asset. load. delay. That’s what’s being fixed with this update.

And when can we see a fix for this culling abuse and obvious skill abuse causing Thiefs to become the only class you need in WvW.

(Skip to 4:30)

This type of tower, keep and even Stonemist capping is really breaking WvW and the feel of fair sportsmanship as a whole. Noone should have to spend 20 mins AoEing every wall, gate and siege equipment looking for Thiefs perma-stealthing.

Hope this fixes the issues!

maybe an easy fix for that and maybe it already does this is that everyone in your group never culls.
or place a minimum on friendly culling. say 10.

Great job Anet! how about you add an option that let us “keep enemies as fallback models”, This should increase the client performance in WvW, the last thing that many of us care about in WvW fights is the skins of the enemy models.

Kerithlan.1659:

Is it possible to temporarily increase server caps until the 28th? There are a lot of people moving around right now and it’s getting very difficult to get everyone where they want to be before they’re no longer able to move.

I understand that you guys want populations to disperse more, but the reality is that we (and other servers) have several guilds and groups of friends split between different servers right now, and it’s nearly impossible for them to join their friends.

Most players would rather play with their friends than not, and for a lot of them transferring isn’t an option because free transfers are going to close before they can choose a smaller server.

I don’t know if it was me that prompted the server cap increase, but I cannot thank you guys enough either way. You’ve just made many of your players incredibly happy.

Jalmood.3705:

Great job Anet! how about you add an option that let us “keep enemies as fallback models”, This should increase the client performance in WvW, the last thing that many of us care about in WvW fights is the skins of the enemy models.

How about reading the big red posts in this thread, it only has three pages:
https://forum-en.gw2archive.eu/forum/pvp/wuvwuv/January-WvW-culling-loading-changes/page/2#post1275716

Habib,

What’s the number of allies that get reported to clients before culling kicks in? I’m really not interested in playing the “find the commander” game again when running with guild people.

Kuldred.2436:

The December culling ‘fix’ didn’t actually fix anything, you just made the culling different. It is/was still rediculously rampant.

So no actual fix or even progress on culling for the January update, got it.

You got that wrong then or didn’t fully read.
No it isn’t permanently fixed / removed yet but there is progress:
New culling method that was triald has been improved by combining it with fallback models to improve asset loading times, because the issues some players reported with the culling method were actually asset load related and not culling.
It’s already an improvement to the current version, and they did some groundwork for future updates to it and keep working towards the goal of completely removing it.
It is progress, maybe not as big as you like due to your impatient nature, but progress nonetheless.

Please for the good of your game and the future of your development can you not consider a public test server as enquired about many times? Do your metrics provide you with actual large scale data that you use in your internal tests? Do you have alpha testers numbering in the hundreds?

Players will willing take time out from their regular play schedule for mass testing before deploying to live in the interest of your game. Is it so bad to learn from other developers that employ a public test server like CCP (Eve Online)?

Tiglie.5834:

So this confirms it. You change is not addressing culling just loading.

<nods>

That is what I thought too: in this patch, nothing gets fixed for people that have fast PCs (with SSD drives, for example).

We can only hope that things gets fixed later for everyone, but this month won’t help me at all.

Tallis.5607:

Tiglie.5834:

So this confirms it. You change is not addressing culling just loading.

<nods>

That is what I thought too: in this patch, nothing gets fixed for people that have fast PCs (with SSD drives, for example).

We can only hope that things gets fixed later for everyone, but this month won’t help me at all.

He says confidently without actually having tested or seen the system in action.

Intigo.1653:

Tallis.5607:

Tiglie.5834:

So this confirms it. You change is not addressing culling just loading.

<nods>

That is what I thought too: in this patch, nothing gets fixed for people that have fast PCs (with SSD drives, for example).

We can only hope that things gets fixed later for everyone, but this month won’t help me at all.

He says confidently without actually having tested or seen the system in action.

And he’s normally wrong, considering cached models/textures will load faster then models/textures loaded from disk (even if it’s an ssd).

Thanks for your effort against culling Habib:)

Keep it up.

Habib Loew.6239:

The engine change we’re making uses fallback models to represent characters until their detailed models are fully loaded. The fallback models are cached so that they can display without any asset load delay and there is a distinct fallback model for each race/gender/armor-class combination.

It would be even better if there was an option for something like “always use fallback models” available in the client for those with low specs machines.

This would help reduce the load on machines with limited graphics capabilities.

there is a distinct fallback model for each race/gender/armor-class combination

Why make it nso complicated? That vstill makes 30 different models to take into account, which is still a lot. Scrap the gender and race visuals. I have no need to know that a female charr or a male asura is in front of me. That would reduce the amount of generic skins needed to only 3. That would also help a lot in world events like the Karkas.

I don’t think it would be visually satisfying to have nothing but fallback models on the screen but we’re definitely exploring the right balance.

But some (many?) of us appears to think it would be satisfying. How hard would it be to make such an option available if it would satisfy some (a lot?) of your player base?

I care more about playability than visual satisfaction.

Always better to satisfy your clients’ wishes than your own wishes….

Quick note:
One thing they haven’t mentioned is if these “Fallback models” can be attacked or Tagged.
My money is on that no they can’t

EnoLive.2367:

Quick note:
One thing they haven’t mentioned is if these “Fallback models” can be attacked or Tagged.
My money is on that no they can’t

Of course they can, I don’t see what on earth would ever make you think otherwise. It is nothing but a visual change to speed loading of the models/textures, it would be ridiculous for it to prohibit attacking.

It would be nice if instead of just splitting allies and enemies, which made very little actual improvement last time they tried it, they would implement something similar to the suggestions I made here which would at least improve things a lot more until they can eliminate culling completely. The placeholder models are certainly going to help a lot though.

Chaba.5410:

Scarn.6810:

Mif.3471:

/hugs my SSD

:P

(Seriously, do yourself a favor and buy one)

I have one, but I installed on my large capacity disk drive. I guess I should stop being lazy and just reinstall on the SSD already.

Me too! LOL! I’m kind of afraid to uninstall and reinstall even. Don’t want to miss out on game time.

You don’t have to. All you have to do is copy all the game files over to the new drive. It works, helped a friend set up his new account I bought for him by copying all the files off my system through my LAN. It’s faster than installing by disk and then having to let it update online.

RyuDragnier.9476:

Personally I feel they should make the classes that can stealth the priorities for loading enemies. For example, make thieves the first thing the game loads, that way when they come out of stealth, the culling doesn’t give them an advantage.

^Agree 100%.

Tiglie.5834:

The current culling situation (just picking arbitrary numbers here), allows say 50 models to load, be it either enemy or ally. The December system splits the cap to say 25 each side. For your standard zerg this is an improvement. They were previously using their 50 cap on themselves, so by cutting down the cap to only 25 of their allies, it allows at least 25 of the enemy to load to get a rough idea of position.

I wonder what happens at the 50 cap? Do we get with placeholders for every extra person over 50? That would make sense to me and would allay your concerns.

Svarty.8019:

Tiglie.5834:

The current culling situation (just picking arbitrary numbers here), allows say 50 models to load, be it either enemy or ally. The December system splits the cap to say 25 each side. For your standard zerg this is an improvement. They were previously using their 50 cap on themselves, so by cutting down the cap to only 25 of their allies, it allows at least 25 of the enemy to load to get a rough idea of position.

I wonder what happens at the 50 cap? Do we get with placeholders for every extra person over 50? That would make sense to me and would allay your concerns.

How is your client going to use a placeholder for something it doesn’t have data about? If you only get info of 50 players from the server only those exists according to your client.

Intigo.1653:

Tallis.5607:

Tiglie.5834:

So this confirms it. You change is not addressing culling just loading.

<nods>

That is what I thought too: in this patch, nothing gets fixed for people that have fast PCs (with SSD drives, for example).

We can only hope that things gets fixed later for everyone, but this month won’t help me at all.

He says confidently without actually having tested or seen the system in action.

It was confirmed on how the system operates in the third red post in the thread. Go back and read it. This does absolutely nothing to address the server side reporting to the machine. The affinity system “helps” with this. The problem being, the priority system absolutely renders the game unplayable for people not running in zergs. Anyone who runs in small groups before, during, and after the trial can attest to this.

If you think cpu processing of the data was the issue (which this change IS addressing), there would not be 99% of the population crying. The server cap they have on data transfer is the issue, affinity is just a reallocation of bandwitch, a kittenty one at that.

Why not leave the current system as it is and roll out the placeholders for the system lag………you are going to have a ton of people quit over this affinity system.

There ya go, good job Habib, and everyone who’s been working on this solution. Those Fallback models is something which I had, in a way, tried to suggest maybe about a month ago (not sure just how many other people may have made the suggestion, I’m not trying to take credit). I really think this is an enormous step in the right direction!

I absolutely agree with the sentiment to have the ability to render more players (cull less) by choosing to permanently retain the fallback models and simplified spell effects. Perhaps just make that behavior turn on whenever you’re in range of the orange crossed swords?

I have to stop reading this thread. It’s giving me a frustration headache.

Please, people, be civil. I know we’re already three pages into the topic, but at least read the big red posts before you reply. I haven’t seen anything posted yet (and I’ve been reading them all) that I would consider a valid concern.
No, this is not the end of all culling, but neither is it the end of their solutions. They’re still working on it, and this is just a stopgap measure.

If you experienced poor performance with the December update, consider waiting to see what effect the other changes (particularly “fallback models”) have on your experience. They heard your vocal (sometimes vicious) feedback in December, and they investigated the problems before implementing this system. Personally, I had no problems whatsoever in December, but I know my experience isn’t the only one; I’m only mentioning it to prove that not everyone hated it.

While I think it would be neat if they could adjust the culling threshold dynamically (something like the new system, but if there are fewer than 10 or so allies, loading all of them first and then spending the remainder of the resource quota on enemies), I expect the January patch to bring improvements overall.

And yes, we know it would be awesome if the fallback/placeholder models could be extended to other areas, but:

This change also lays the groundwork for more extensive uses of fallback models in future updates.
…
While this update only contains a couple of visible changes to WvW it lays a lot of important groundwork for future updates. We have some exciting changes coming and this update is just the beginning!

They’re working on it.

Habib Loew.6239:

…reported to the client. Once the specific, detailed model for a given character is completely loaded from disk the fallback model will be replaced with the detailed model. …

While basic models is a nice optimization, one could off course start sending the detailed static info – including armor pieces – to the existing clients when someone joins a Wvw server so that the clients have this information ready when this player comes within viewing range. Similarly the new client could be given (all) this info when he has just joined. That way the server(s) will have less work when two zergs collide.

I look forward to seeing it in action, the december trial was an improvement for me, glad to see it coming back

frans.8092:

Habib Loew.6239:

…reported to the client. Once the specific, detailed model for a given character is completely loaded from disk the fallback model will be replaced with the detailed model. …

While basic models is a nice optimization, one could off course start sending the detailed static info – including armor pieces – to the existing clients when someone joins a Wvw server so that the clients have this information ready when this player comes within viewing range. Similarly the new client could be given (all) this info when he has just joined. That way the server(s) will have less work when two zergs collide.

Sending all that information to everbody would make things worse. Especially in such a dynamic game mode where many people are constantly joining and leaving.

The December trial was an improvement for zerglings. I’ll reserve judgement for Saturday, but if December was any indication, you can probably say goodbye to every single DAoC player in WvW.

If you don’t want people zerging, then stop making changes that cater towards zerging.

I was hoping for the best, but I am having a bad feeling about this. For the most part, the trial system was a total bust to most players. Sure a few thought it was better, but the vast majority didn’t.

I am not sure why the ‘bubble of influence’ can’t be expanded to well past normal render distance to give you more time to relay the information to the client until you can optimize your code to perform better. I assume your network and server config is expandable and not in the stone age? You should be able to just throw more hardware/bandwidth at it.

The client will be aware of the models before they are in render distance and can load up the default models until they get the rest of the information from the server. This way everything that we are currently experiencing in rendering range now occurs off screen.

This is one of the things DAOC did a decade ago (and was responsible for the ‘lagdar’ effect once in a while). Why can’t you do this now?

Edit: Not only that, you need to make better use of memory on GPUs. For the most part, this game hardly touches mine.

covenn.7165:

I was hoping for the best, but I am having a bad feeling about this. For the most part, the trial system was a total bust to most players. Sure a few thought it was better, but the vast majority didn’t.

I am not sure why the ‘bubble of influence’ can’t be expanded to well past normal render distance to give you more time to relay the information to the client until you can optimize your code to perform better. I assume your network and server config is expandable and not in the stone age? You should be able to just throw more hardware/bandwidth at it.

The client will be aware of the models before they are in render distance and can load up the default models until they get the rest of the information from the server. This way everything that we are currently experiencing in rendering range now occurs off screen.

This is one of the things DAOC did a decade ago (and was responsible for the ‘lagdar’ effect once in a while). Why can’t you do this now?

Edit: Not only that, you need to make better use of memory on GPUs. For the most part, this game hardly touches mine.

https://forum-en.gw2archive.eu/forum/pvp/wuvwuv/Update-on-Culling/first#post1137307

PS: using more GPU memory is not a sign of quality. GW2 is well optimized in this regard.

Iruwen.3164:

https://forum-en.gw2archive.eu/forum/pvp/wuvwuv/Update-on-Culling/first#post1137307

PS: using more GPU memory is not a sign of quality. GW2 is well optimized in this regard.

Link to old information thus a useless post. ANyone that follows or cares about this issue has read that already.

I also never said that using more GPU memory is a sign of high quality. You assumed that, but it was not what I said. (One issue is how CPU bound this game is, to see where I was going with my comment)

covenn.7165:

Iruwen.3164:

https://forum-en.gw2archive.eu/forum/pvp/wuvwuv/Update-on-Culling/first#post1137307

PS: using more GPU memory is not a sign of quality. GW2 is well optimized in this regard.

Link to old information thus a useless post. ANyone that follows or cares about this issue has read that already.

I also never said that using more GPU memory is a sign of high quality. You assumed that, but it was not what I said. (One issue is how CPU bound this game is, to see where I was going with my comment)

You obviously didn’t read it, or you wouldn’t have made that statement. Apart from that, your last comment about memory makes it even more useless. If you want to say something: say it. Don’t make shady assumptions to be interpreted by others and complain afterwards.

Thankyou Arenanet this is the best news i heard as a update for a game.

Been so angry when enemy abuse culling bug with running in extremle big groups so you dont se that they are moving in Golems to gate before they start attacking.

RyuDragnier.9476:

Personally I feel they should make the classes that can stealth the priorities for loading enemies. For example, make thieves the first thing the game loads, that way when they come out of stealth, the culling doesn’t give them an advantage.

That is an idea, additionally you could just keep it simple and prioritize giving a pc information about anyone within view (!) who targets them.

Iruwen.3164:

Sending all that information to everbody would make things worse. Especially in such a dynamic game mode where many people are constantly joining and leaving.

People don’t join and leave at very high rates, it’s probably a handful per minute and everyone gets information constantly anyway. Can’t be a big problem to add a few dozen bytes to describe the character.

What was funny about the December culling trial was that the thread about was full of people who didn’t like it. Then after the original behavior was restored, the thread about that was full of people who liked it and wanted it back. Now that we’re going back to the way it was in the trial the majority of the responses to it are negative. Maybe everyone should actually contribute to the conversation rather than only coming out to post when you don’t like something.

For those of you who don’t like it, you only have to deal with it until culling is removed entirely. As Habib explained a thousand times this is a temporary measure, not a permanent change. As for me, I’m happy that thieves won’t be able to endlessly stealth around in zergs anymore.

lettucemode, culling will never be removed completely so I’m not sure what you’re talking about.

waeren.9743:

lettucemode, culling will never be removed completely so I’m not sure what you’re talking about.

Never is a big word to use especially given the fact that other MMOs can and have suppported hundreds of players on screen together without any culling.

so it’s gonna be months before we see an actual fix. why isnt the thief receiving some sort of temporary nerf until then, in WvW only? im tired of thief wars 2.

I wish this was stated specifically by a dev so people would notice, but can we once and for all eliminate the idea that stealth (in a smaller skirmish) is a culling issue? It is an asset load issue. When you are 1v1 against a thief and he comes out of stealth, your client knows he is there. You can’t see him for 2 more seconds because you haven’t loaded the model. This is what cause the affinity culling to suck for small groups in December.

With the fallback models, when that thief unstealths, his fallback model will load immediately (essentially). No matter how good your computer was in December, it still took time to load the detailed model. Even if it’s top of the line, you still have to load a model.

Don’t be so quick to jump to conclusions, you have not seen the system they are implementing yet. Affinity culling is not the same thing as affinity culling plus fallback models.

@Habib Loew

I got questions.

I was around for the test run of the culling solution ur going to be using. The grouped allies being visable was great and all but the culling distance was still incredible close to the players character still.

Can we expect the same short culling radius as before? becouse to blunt..the radius is waaay to close.
From what iv seen culling can get very close to the character.

When said test run was introduced i went from seeing every1 inside the Stonemist keep main room in eternal battleground., to only seeing people within a small radius around me (same room, same day, same people, just b4 and after the culling was introduced during the WvW reset). Both before and after i had 60 fps, there for culling was not neccesery on my end. There was ‘atleast’ 50 people in the room.

Cant u make culling take into account client side FPS?

So say for example the Server finds that client FPS starts of at 50 when entering WvW, then plumits to 20 in battle .. culling kicks in to keep FPS above 30 (say 35 for leeway). If FPS never reaches 30 then culling stays on. If it goes above 40 the culling radius increases till the FPS is again near 35.
If FPS starts of at 60 and doesnt drop down below 35 ever then no culling is used untill it does.

I would like to have the option of keeping the fallback model as the default model at all times. I realy don’t care what players are wearing and if they’re weilding rainbow weapons.

In the early days of PC gaming, FPS games gave us a lot of customizability on what the engine rendered.

Oh dear. The December trial was simply horrible, we can only HOPE that the additional changes have a significant impact and improve it considerably.

I play on a high end system with SSD, I know I suffer less issues and culling than most guild mates as I’m usually the one calling out on TS a 30 man zerg is about to arrive on our heads because I see them first.

But, if there is no significant improvement from the December trials I won’t be logging in to play until things are changed. That week in December was horrible for everyone in our guild – regardless if they ran in zergs or ran solo/small group.

They should make the placeholder models look like trees or critters. “Oh god, look at that enormous forest heading this way. GTFO!”