57 replies to this topic

[Ncyphe]

Bottoms_Up, on 24 January 2013 - 01:02 AM, said:

As far as I understand, culling as Anet uses it is a method by which they ensure that the flow of data between their servers and players' game clients is smooth. Culling in this sense does not mean that they destroy data, but rather distribute it over time so that it doesn't all flow at once and cause a bottleneck.

Consider the simplified process:

1. Player A presses the move forward button and their client sends this message to the server
2. The server receives the message, processes it and sends it to Player B
3. Player B's client receives the message and renders Player A's movement on their screen

Culling affects step 2. Imagine that there are 50 players who all move into Player B's view at once; to avoid a bottleneck, culling could mean that the messages that those players are there are sent to Player B's client in clumps rather than all at the same time. The net effect is that Player B will see only a portion of those players at first and then more later.

This is vast a simplification because other factors come into play (how many players are there, how many are already on Player B's screen, are they friendly or enemy, do any of them go invisible, etc.).

Fallback models on the other hand I think affect only step 3. Although culling and fallback models are connected in the same process and together affect the final result, they are different problems.

But I'm not an expert and am happy for corrections if I got something horribly wrong.

Culling is actually a 3D-graphics definition.  It's the process of analyzing a scene and deciding what to render and what not to render.  What's happening, is your client learns of a visible player, but has no information on hoe that character looks, so ANet originally designed the client to just ignore the player until visual data is received (not supposed to sound bad).  It also has a lot to do with what a graphics card can handle.

So, in essence, ANet is now redesigning culling to place low def. place holders so that players can see all the players in the area.

[Bottoms_Up]

Ncyphe, on 24 January 2013 - 01:08 AM, said:

Culling is actually a 3D-graphics definition.  It's the process of analyzing a scene and deciding what to render and what not to render.  What's happening, is your client learns of a visible player, but has no information on hoe that character looks, so ANet originally designed the client to just ignore the player until visual data is received (not supposed to sound bad).  It also has a lot to do with what a graphics card can handle.

So, in essence, ANet is now redesigning culling to place low def. place holders so that players can see all the players in the area.

I understand what you are saying and think that logically this is what one would think when talking about culling. I have got the impression, however, that Anet use the word culling in a different sense (this is from comments made by Anet employees in various places).

Take, for instance, the comment in the 'announcement' at the beginning of this thread referring to fallback models:

"Please note that this change does not eliminate delays due to culling, it only addresses delays due to asset load times."

This indicates I think that asset load times (including rendering) and culling delays are separate issues. I am pretty sure from past comments that the culling problem is very much a server-side problem but I don't have links to those and am in no way qualified to go deeper into it so will bow out now. Thanks for your input.

[Impmon]

Anet to acknowledge my idea at the very least.  They never mentioned using the generic models previously.  All they talked about was server lag etc.  They didn't have a clue until they probably heard my idea to use the same models for everyone.

[kidawk]

typographie, on 24 January 2013 - 12:33 AM, said:

I'm not. I can understand that something like this that cuts to the core of their game engine requires some time to get done. I was disappoint a month or so ago when I wasn't sure they were serious about addressing it. But now I'm glad to see they're really on it.

You know, I ain't a programmer or what not but this type of issue pertaining to culling reminds me of the stuck bugs that riddled WAR. Look what happened to that game ~_~...

Quite a bit of flip flopping from ANet lately. First they say no ascended item additions in 2013. Then it is revealed that ascended amulets will be attainable in the coming Jan. patch.

First they claim that culling will be eliminated in the same patch. Now, it is just a work around.

I like the passion of some of these devs but they really need to L2PR without utilizing soup nazi moderators that still populate their official boards.

It is only lately that I feel this but TESO looks very slightly appealing only because of the disappointment that has sprung up. Generally, not really impressed with TESO neither..

[dannywolt]

kidawk, on 24 January 2013 - 02:00 AM, said:

Quite a bit of flip flopping from ANet lately. First they say no ascended item additions in 2013. Then it is revealed that ascended amulets will be attainable in the coming Jan. patch.

Source? I distinctly recall A-Net saying no new gear tier in 2013 and that Ascended gear would be rolled out throughout 2013.

[dawdler]

Ncyphe, on 24 January 2013 - 01:08 AM, said:

So, in essence, ANet is now redesigning culling to place low def. place holders so that players can see all the players in the area.

Which you might as well also add this is also a defined behaviour in 3D known as LoD (Level of Detail). Most modern games use LoD rendering. Some with good design where you can barely tell its being used, some with ugly LoD popping all over the place (Total War shifting from 3D characters to 2D sprites comes to mind).

Edited by dawdler, 24 January 2013 - 05:29 AM.

[kidawk]

dannywolt, on 24 January 2013 - 02:14 AM, said:

Source? I distinctly recall A-Net saying no new gear tier in 2013 and that Ascended gear would be rolled out throughout 2013.

Twas mentioned in an itnerview with Colin Johanson. You can find it

[Gilles VI]

kidawk, on 24 January 2013 - 07:21 AM, said:

Twas mentioned in an itnerview with Colin Johanson. You can find it

It was mentioned they wouldn't introduce a new gear tier.
Ascended has been said from the start it would be rolled out in 2013.

[BrownPatrick]

Impmon, on 24 January 2013 - 01:25 AM, said:

Anet to acknowledge my idea at the very least.  They never mentioned using the generic models previously.  All they talked about was server lag etc.  They didn't have a clue until they probably heard my idea to use the same models for everyone.

Oh, Oh, Oh, Oh,
Thank god you were there

On topic:
Sounds good, but really depends on the implementation.
Hope this and february WvW improvements will be good enough to start playing WvW again.

[Vihar]

Lordkrall, on 23 January 2013 - 10:44 PM, said:

Did you even read the whole first post?
Where do they state that they simply take the exact same thing that they used in December and just put it back in?

It is quite clear that they are using that trial and the results as a basic for the new system and they also stated that they are adding fallback models until the real models are loaded.

"In December we ran a one matchup trial of an updated culling system. Based on all the feedback we received, both during and after the trial, we will be transitioning to the updated culling system that we used in the trial. This update allows the culling system to handle allies and enemies separately so that being surrounded by a group of allies will not impact the culling of enemies (and vice-versa)."


  Yes, I did.

  Did you?

  The fallback models are for those with lower end video cards. Nothing to do with culling.

[The Comfy Chair]

Vihar, on 24 January 2013 - 11:40 AM, said:

"In December we ran a one matchup trial of an updated culling system. Based on all the feedback we received, both during and after the trial, we will be transitioning to the updated culling system that we used in the trial. This update allows the culling system to handle allies and enemies separately so that being surrounded by a group of allies will not impact the culling of enemies (and vice-versa)."


  Yes, I did.

  Did you?

  The fallback models are for those with lower end video cards. Nothing to do with culling.

I found December to be a much better solution personally. If you're in a small group and you encounter an enemy zerg, you run anyway. So only seeing 50 of them instead of 100 wont be a big deal considering you were going to get steamrolled by attacking them head on anyway. Culling is mainly an issue in zerg versus zerg, and i don't care so much about seeing my allies in that situation.

Since culling renders the closest enemies to you, you'll see the front of the incoming enemy zerg as a small group (that's all you need to see to know it's time to run), and you'll be able to see far more of a zerg when with allies. You can't ague 'it'll hide the size of the enemy zerg in zerg v. zerg' because the previously culling did that anyway!

Edited by The Comfy Chair, 24 January 2013 - 11:57 AM.

[Vihar]

The Comfy Chair, on 24 January 2013 - 11:56 AM, said:

I found December to be a much better solution personally. If you're in a small group and you encounter an enemy zerg, you run anyway. So only seeing 50 of them instead of 100 wont be a big deal considering you were going to get steamrolled by attacking them head on anyway. Culling is mainly an issue in zerg versus zerg, and i don't care so much about seeing my allies in that situation.

  Except that it's not 50, it's more like 10 enemies you see.

  I generally know how many allies are around me, anyway.

  When you run into a supply camp and see 10, you keep going...and therein lies the problem.

[typographie]

kidawk, on 24 January 2013 - 07:21 AM, said:

Twas mentioned in an itnerview with Colin Johanson. You can find it

Quite easily in fact, it wasn't long ago:

Colin Johanson said:

Our reward systems need to be exciting, and include things you want to earn over time, but we don’t want to force our players on endless gear treadmills for new tiers of gear we add every 6 months. You won’t see another tier between Ascended and Legendary in 2013 for example. Our goal will be to use our existing reward systems and build new ones that are fun and exciting that step away from the stale gear grind reward systems you see elsewhere.

Source

Edited by typographie, 24 January 2013 - 02:20 PM.

[Norn Osprey]

Bottoms_Up, on 24 January 2013 - 01:19 AM, said:

This indicates I think that asset load times (including rendering) and culling delays are separate issues

They are separate issues. Rendering delay only happens player-side after the players computer gets the information and creates an image. The delay depends on how powerful the players computer is. Culling is done server side and it is a willful omission of data sent to the player about nearby targets that should not be able to harm the player. Culling saves them bandwidth (primarily) and they claim, benefits players on low end machines. However, I have 3 guildies that play off mid level 5 yr old laptops that did not have a single problem with zergs during beta.

It is my belief, that after the first Beta weekend, ANet saw how big their ISP charges were going to be, and started a group of people working on Culling to save them money. Culling wasn't in any of the beta weekends or stress test events. It wasn't added to the game until a week or so after launch. Even then, it wasn't until players started complaining about invisible characters in WvW and massive events in Orr, Hello Temple of Balthzar, that ANet finally admitted they added Culling.

EDIT: Clarification

Edited by Norn Osprey, 24 January 2013 - 02:56 PM.

[Doctor Overlord]

Azure Skye, on 23 January 2013 - 10:41 PM, said:

Like these? I thought my gaming files were corrupted when they did this update. Lol

I actually did not know ArenaNet had pulled something like that before!   That's hilarious    And it really makes me want to see what they'll be doing for April Fools in GW2.

[Azure Skye]

Doctor Overlord, on 24 January 2013 - 03:31 PM, said:

I actually did not know ArenaNet had pulled something like that before!   That's hilarious    And it really makes me want to see what they'll be doing for April Fools in GW2.

Yea, I can't wait,if they do it again. Lol  Tho, i love their 2008 to 2012 April's Fool's day prank. Lol

And they did Gwen-chan and when they shrank everyone to mini size. It was funny to be playing GW1 then. Lol

Spoiler

Though i can't wait for this culling to be fixed, i need to stop reminiscing about GW1. =P

[Var]

kidawk, on 24 January 2013 - 02:00 AM, said:

Quite a bit of flip flopping from ANet lately. First they say no ascended item additions in 2013. Then it is revealed that ascended amulets will be attainable in the coming Jan. patch.

First they claim that culling will be eliminated in the same patch. Now, it is just a work around.

They haven't been flip-flopping at all recently (and your citation on ascended is wrong as has been pointed out), in fact Habib has been as candid and up front with his progress as can be, which is saying a lot with how hush hush they were before (though, honestly, I think that was because they were caught overwhelmed by how successful their game was at launch (they're catching up to the seven years worth of sales in GW1...), I honestly don't think they expected it, got flooded, some departements (read: WvW/sPvP) were understaffed in the panic to work out the rest of the game).

kidawk, on 24 January 2013 - 02:00 AM, said:

It is only lately that I feel this but TESO looks very slightly appealing only because of the disappointment that has sprung up. Generally, not really impressed with TESO neither..

Considering that all of the claims from TESO have been nothing but claims, I remain extremely skeptical. Its also P2P...

-----------------------------------------------

A quick question for the more savvy:

Upon dying, I will usually see a bunch of models and players load in almost instantly the moment my body becomes a corpse. Were these people culled or were they asset load time issues? If it is the latter, a lot of my problems are about to disappear. I'm under the impression its the latter because I'd imagine that these people weren't all instantly reported to me at the moment I died but had been reported long since any battle had started but my game and computer hadn't rendered them.

If they are killing me, they have been reported, am I right in thinking of it this way? So anything that can kill me will now be loaded by dummy models, so I won't be dying to ghosts anymore. Or can unreported people (data culled) still kill me? Seems like that would be impossible since the server sending me data has no idea these people exist and as such their attacks do not yet exist either. Or do attacks go on a different cull stack?

[Izokka]

i dindt notice any change in my wvw xperience an anyone post a SC of how does fellback players look?

[Var]

Izokka, on 24 January 2013 - 04:49 PM, said:

i dindt notice any change in my wvw xperience an anyone post a SC of how does fellback players look?

Update goes live on January 28th, today is January 24th.

Edited by Var, 24 January 2013 - 04:50 PM.

[Hecksa]

The Comfy Chair, on 24 January 2013 - 11:56 AM, said:

I found December to be a much better solution personally. If you're in a small group and you encounter an enemy zerg, you run anyway. So only seeing 50 of them instead of 100 wont be a big deal considering you were going to get steamrolled by attacking them head on anyway. Culling is mainly an issue in zerg versus zerg, and i don't care so much about seeing my allies in that situation.

No, if you're in a small group against a zerg you don't necessarily run away. If you're in a unorganised group, sure - when you see larger numbers the correct decision is to run, and I'd do exactly the same.

I really enjoy fights in organised groups against larger numbers, though. In order to win those fights, you need to know how many people you're charging into, and where they are. With affinity culling, as it's being called, we can't see that. It makes killing larger numbers very difficult. When you charge into 50 people with 20, you're already at a massive disadvantage. When you can only see half of those 50, and they can see all 20 of you, you're in big trouble. Making it harder for the smaller group to win just turns WvW into a game of "who has more numbers". I thought everyone agreed that zerging was a bad thing?

[The Comfy Chair]

Hecksa, on 24 January 2013 - 05:41 PM, said:

No, if you're in a small group against a zerg you don't necessarily run away. If you're in a unorganised group, sure - when you see larger numbers the correct decision is to run, and I'd do exactly the same.

I really enjoy fights in organised groups against larger numbers, though. In order to win those fights, you need to know how many people you're charging into, and where they are. With affinity culling, as it's being called, we can't see that. It makes killing larger numbers very difficult. When you charge into 50 people with 20, you're already at a massive disadvantage. When you can only see half of those 50, and they can see all 20 of you, you're in big trouble. Making it harder for the smaller group to win just turns WvW into a game of "who has more numbers". I thought everyone agreed that zerging was a bad thing?

I'm talking 5-10 man groups versus 50. 20 versus 50 is easy, first group to have the bravery to charge, wins. The normal zerg will run away in all directions once they see players run at them. Never underestimate the cowardice of other players in WvW.

I count 20 players as zerg myself to be honest, it's overkill for many situations. So it would class as zerg v zerg.

Vihar, on 24 January 2013 - 12:00 PM, said:

  Except that it's not 50, it's more like 10 enemies you see.

  I generally know how many allies are around me, anyway.

  When you run into a supply camp and see 10, you keep going...and therein lies the problem.

A problem that the new system addresses much more than before. You would see a 50 man enemy zerg now, whereas before you wouldn't if you were in a 60-70 man zerg yourself. Here's the breakdown of why the new system is relatively win-win in major fights:

Fixed player counts for allies and enemies means:

If you're ever in a allied zerg or by an allied zerg that breaches the player count limit for allies, you're not going to be making any real intelligent strategical attacks against other equal number zergs. We're talking the 'main zergs' of a server colliding, those battles are a spam fest, with 80-90% of the players (let's base it off steam stats) running at pretty much unplayable framerates, with only those rocking overclocked i5's with any hope of making any real decisions mid fight.

Additionally, in that situation, the enemy zerg would be much smaller on screen than it will be after the update, so even if someone could actually make the judgement about the fight AND stop the tunnel vision zerg from charging blindly, they have a better idea of the enemy numbers than before.

If you aren't triggering the ally limit, you're very unlikely to be taking on a zerg that does regardless. Unless you're somewhat close to the limit anyway or a significant force, at which point actual numbers don't matter much as charging in and scaring the players away. This is made easier because the bigger zerg feels smaller now!

Disadvantage: a single player on their own away from allies can no longer 'scout' effectively for larger forces that the enemy player count. But since most scouts at the moment just say 'zerg incoming' and lose the ability to judge well after about 30-40 players in a zerg (especially if you throw pets and minions into the mix), i don't think this is a major thing either.

Edited by The Comfy Chair, 24 January 2013 - 08:26 PM.

[Vihar]

I am skeptical, but we will see.

  I specifically recall the December test being the worst culling effects ever, and was happy when they reverted it back. That opinion was almost universal in my guild.

   I also remember part of the issue would be people moving back and forth just to alter distances, and the culling would kick in and out, and they would thus exploit the terrible rendering issues.

  Hopefully the fallback models will at least remedy that.

[DeagarFA]

Bottoms_Up, on 24 January 2013 - 01:19 AM, said:

I understand what you are saying and think that logically this is what one would think when talking about culling. I have got the impression, however, that Anet use the word culling in a different sense (this is from comments made by Anet employees in various places).

Take, for instance, the comment in the 'announcement' at the beginning of this thread referring to fallback models:

"Please note that this change does not eliminate delays due to culling, it only addresses delays due to asset load times."

This indicates I think that asset load times (including rendering) and culling delays are separate issues. I am pretty sure from past comments that the culling problem is very much a server-side problem but I don't have links to those and am in no way qualified to go deeper into it so will bow out now. Thanks for your input.

I'm really hoping that whomever wrote that is bullshitting because transmitting only the race, the gender and whatever armor class would alleviate the culling problem dramatically since the calculations and transmissions are more simple.

[Bottoms_Up]

DeagarFA, on 24 January 2013 - 10:52 PM, said:

I'm really hoping that whomever wrote that is bullshitting because transmitting only the race, the gender and whatever armor class would alleviate the culling problem dramatically since the calculations and transmissions are more simple.

I'm not sure, because they aren't saying that is the only info being transmitted, rather that that info is being used by your PC to render a more basic model on screen.

I think those three aspects are probably pretty trivial in comparison to other transmitted data, such as movement direction, speed, actions, buffs, etc. If they only sent info for only those three aspects, you might end up being able to see someone is there but have no idea what they are doing or where they are going.

My impression is that culling isn't about filtering the types of data being sent but rather choosing for which of the players in an area data should be sent.

Edited by Bottoms_Up, 25 January 2013 - 02:53 AM.

[dawdler]

Bottoms_Up, on 25 January 2013 - 02:53 AM, said:

I think those three aspects are probably pretty trivial in comparison to other transmitted data, such as movement direction, speed, actions, buffs, etc. If they only sent info for only those three aspects, you might end up being able to see someone is there but have no idea what they are doing or where they are going.

Considering the three are simple integers while movement is complex transformations, yes they are trivial lol.

It really depend on what "issue" we are refering to, yes.

You have two scenarios here - either the client is getting complete player data but the character model (ie armour pieces, colors, etc) hasnt loaded yet. The character is invisible... It isnt technically a subject to culling, its just not loaded yet. Being able to display a simple model (race/height/gender/class armour) until it has loaded the custom model would completely removing the "culling" issue seen there BUT introduce LoD popping if the class armour is very different from custom armour.

The you have the other type of network "culling" where I am guessing that the client is recieving simplified data. The character is invisible because your client only knows there is a player at that location - you have already reached the maximum your client can handle OR there is a delay issue.

Its really too technical for me, but it still doesnt change one rather important thing - the december test didnt go that well. I would be more interested in seeing ONLY the LoD rendering introduced than a return to that code...

Edited by dawdler, 25 January 2013 - 10:39 AM.

[baels]

Wondering how useful this fix will be. When I command it's more or less "Push in and AoE the crap out of everything, don't stop, just keep AoEing. You can't see them, but they're here."

[asmodess]

{ Quote  }  

Considering that all of the claims from TESO have been nothing but claims, I remain extremely skeptical. Its also P2P..


Never understand the issues people have with p2p games , i remember when wow was releasing pre launch the boards were full of people whining about how 15 dollars a day was to much money to pay for a game .

I live in texas which has a low cost of living  ie pay is less then somewhere else but since we have no personal income tax other then it`s not bad . my 25 dollars a hour here would be 35 ish in say cali.

just amazes me i guess that so many have a issue paying 50 cents a day for unlimited access to a game  not counting scheduled maint etc .

I mean 1 day of overtime just 1 day pays for a 2 year sub  ohwell .


  Yep I`m skeptical how TESO will turn out  also  supposedly the lead desgin dev was a lead dev in the org  DAOC  so that`s a plus  as  DAOC  still has the best rvr    GW2  while being a good copy of  DAOC   isn`t as good .

  While DAOC and warhammer were both made by the same company , they didn`t have the same people working on them and it showed  how warhammer RvR was
.
  All they needed to do to make Warhammer a good game was use a better game engine , and move DAOC entire RvR  system, complete with crafting, guild and player housing.

Then slap Warhammer artwork, graphics, classes , lore on top , adjust OP CC aspects that were in DAOC and that game would have been  10 times better then what they gave us .

While id like TESO to be good , I suspect it`ll turn out like every other MMO in the last 8 years , dumbed down , easy mode grind treadmill to pull in the most people and make investors happy

Edited by asmodess, 28 January 2013 - 03:54 PM.

[notrub]

asmodess, on 28 January 2013 - 03:34 PM, said:

{ Quote  }  

Considering that all of the claims from TESO have been nothing but claims, I remain extremely skeptical. Its also P2P..


Never understand the issues people have with p2p games , i remember when wow was releasing pre launch the boards were full of people whining about how 15 dollars a day was to much money to pay for a game .

I live in texas which has a low cost of living  ie pay is less then somewhere else but since we have no personal income tax other then it`s not bad . my 25 dollars a hour here would be 35 ish in say cali.

just amazes me i guess that so many have a issue paying 50 cents a day for unlimited access to a game  not counting scheduled maint etc .

I mean 1 day of overtime just 1 day pays for a 2 year sub  ohwell .


  Yep I`m skeptical how TESO will turn out  also  supposedly the lead desgin dev was a lead dev in the org  DAOC  so that`s a plus  as  DAOC  still has the best rvr GW2  while being a good copy of  DAOC   isn`t as good .

  While DAOC and warhammer were both made by the same company , they didn`t have the same people working on them and it showed  how warhammer RvR was
.
  All they needed to do to make Warhammer a good game was use a better game engine , and move DAOC entire RvR  system, complete with crafting, guild and player housing.

Then slap Warhammer artwork, graphics, classes , lore on top , adjust OP CC aspects that were in DAOC and that game would have been  10 times better then what they gave us .

While id like TESO to be good , I suspect it`ll turn out like every other MMO in the last 8 years , dumbed down , easy mode grind treadmill to pull in the most people and make investors happy

The problem with the monthly subscription model isn't the cost. It is that you HAVE to pay every month even if the developer is giving you a crappy game. In a payment model like GW2 I decide when to reward Anet by a purchase for what they have given me in game. The whole reason P2P is dying out is because players as a whole are starting to recognize the difference. If/when TESO decides to go with another method than monthly subs I'll check it out. But, after all the recent disasters in MMOs I won't even be tempted till then.