Jump to content


Welcome to Ehmry Bay Server Forum for Guild Wars 2

The Ehmry Bay server forum is a community for Guild Wars 2 players to discuss the game, plan events, coordinate WvW, and get to know each other better. The interaction on the forum will build social capital which will make for a more entertaining gaming experience.

Please be sure to not use your Guild Wars 2 account password when registering on this forum.

how ArenaNet plans to deal with display culling


  • Please log in to reply
8 replies to this topic

#1 News

News

    Internet Information Extractor

  • Community Support
  • 98 posts

Posted 22 November 2012 - 02:36 PM

This is from the German Guild Wars 2 forum posted by German Community Manager Ramon Domke.  You will find the original post quoted here and then there will be a translation in the second post.  It is disappointing that there is no English equivalent posted by the US ArenaNet team seeing that this is an issue that's on the minds of many players.

RamonDomke said:

Derzeit arbeiten wir an einer Lösung (oder vielmehr an einer Sammlung von Lösungen) für die Anzeigeverzögerung (Culling) und einige Probleme beim Laden von Charakteren. Dazu müssen an etlichen Spielsystemen Änderungen vorgenommen werden, deren Implementierung Zeit in Anspruch nimmt. Wir sind derzeit noch nicht soweit, ein Veröffentlichungsdatum für die Lösung der aktuellen Probleme preiszugeben, möchten euch aber verraten, dass wir intensiv an einer Verbesserung eurer Spielerfahrung arbeiten.

Und jetzt zur langen Version:
Wir setzen derzeit auf der Serverseite Anzeigeverzögerungen bei der Grafik ein, um die Anzahl der Charaktere zu limitieren, die dem jeweiligen Client gemeldet werden. Indem wir die Anzahl der Charaktere eingrenzen, die wir dem jeweiligen Client melden, grenzen wir auch die (von Server und Client) benutzte Bandbreite ein und verhindern Situationen, in denen der Client von der Anzahl der Charaktere, die bearbeitet und gerendert werden müssen, überfordert ist. Dieses System hat zwar ganz offensichtliche Vorteile und funktioniert im PvE ausgesprochen gut, doch die groß angelegten Schlachten, die für das WvW charakteristisch sind, unterstreichen gelegentlich die Schwächen dieses Ansatzes.

Außerdem gibt es einige clientseitige Probleme, die sich darauf auswirken, wie unser System der Anzeigeverzögerung funktioniert. Nachdem ein Charakter an den jeweiligen Client gemeldet wurde, wird zum Laden und Anzeigen der Ressourcen für diesen Charakter eine gewisse Zeitspanne benötigt. Zusätzliche Ladezeit hängt davon ab, wie stark der Client-PC ist (PCs mit mehr Speicher, schnelleren Hauptprozessoren, mehr Prozessorkernen und schnelleren Laufwerken laden schneller). Einer unserer Engine-Programmierer hat vor Kurzem einen Optimierungsdurchlauf für den Charakterladevorgang abgeschlossen, wir sollten also bald Verbesserungen an diesem Aspekt des Problems sehen. Trotzdem liegt der überwiegende Teil des Problems bei der Anzeigeverzögerung auf der Serverseite; wie schnell euer Client einen Charakter laden und anzeigen kann, tut nämlich nichts zur Sache, wenn er nicht weiß, dass dieser Charakter überhaupt vorhanden ist.

Wie ihr vielleicht schon wisst, haben wir im sPvP bereits eine Lösung für die Anzeigeverzögerung auf der Serverseite implementiert. Da es im sPvP dramatisch weniger Spieler gibt, konnten wir die Problemlösung dort sofort implementieren, ohne uns um eventuell dadurch entstehende Nebenwirkungen sorgen zu müssen. WvW funktioniert aber auf einem viel größeren Maßstab als sPvP. Deshalb müssen einige zusätzliche Hürden bewältigt werden, bevor wir das Problem auf der Serverseite angehen können. Wir müssen dazu sicherstellen, dass Clients, darunter auch Clients mit den Mindestsystemanforderungen, viel mehr Charaktere rendern und verarbeiten können. Außerdem müssen wir sicherstellen, dass die vom jeweiligen Client benötigte Bandbreite in einem vernünftigen Rahmen bleibt und weiterhin unseren Mindestanforderungen für Verbindungen entspricht. Das WvW-Team arbeitet derzeit an beiden Problemen – der Bandbreite und der Clientleistung. Die Änderungen, die wir vornehmen, sind komplex und wirken sich massiv darauf aus, wie die Spiel-Engine funktioniert. Da die erforderlichen Änderungen so komplex sind und die zentralen Systeme betreffen, wird die korrekte Implementierung dieser Lösungen länger dauern. Daher kann ich euch auch noch kein fixes Datum dafür geben.

Worauf es uns schlussendlich ankommt, ist eine dramatische Verbesserung der Spielerfahrung in groß angelegten Schlachten im WvW und eine beträchtliche Erhöhung der Anzahl der Spieler, die von jedem Client angezeigt werden.

source: https://forum-de.gui...gehen-gedenken/


This was posted in the same thread in response to just turn player models into stick figures to make the display perform better:

RamonDomke said:

In dieser Richtung schauen wir uns derzeit ebenfalls an was machbar ist. Natürlich soll es am Ende besser aussehen als “Strichmännchen” und dennoch einen ordentlichen Leistungs-Sprung ermöglichen – z.B. ein standardisiertes Modell für männliche Menschen oder etwas vergleichbares – sodass dann nur 10 unterschiedliche Modelle vorgeladen werden müssen – und dennoch die immersion nicht vollständig flöten geht.

and one more:

RamonDomke said:

Richtig, daher wäre es das Ziel eine berechenbare feste Größe zu haben (wie eben z.B. die 10 Platzhalter-Modelle – 5 Völker, 2 Geschlechter – Eventuell noch 8 Farb-Optionen für die Klassen-Identifizierung (oder nur 3 Farben für die Zugehörigkeit der Welt)) die permanent vorausgeladen ist und im zweifelsfall sofort zur Verfügung steht. Was in der Theorie schnell ausgearbeitet ist, kann in der Praxis allerdings einiges an Programmierkünsten Erfordern.


#2 News

News

    Internet Information Extractor

  • Community Support
  • 98 posts

Posted 22 November 2012 - 03:02 PM

We are currently working on a solution (or rather a collection of solutions) for the culling and other problems associated with loading characters [into the players view].  In order to improve this issue we have to make numerous changes to the game engine which will take time.  We are currently unable to publish any details on the solutions we are working on, but we do want to let you know that we are intensively working on improving your game experience.

Long version:
We are currently employing a server-side delay in displaying characters to limit the number of characters displayed to each client.  By limiting the number of characters displayed to each client we are able to limit the bandwidth used by the server and the client to avoid situations in which the client is overwhelmed with the number of characters that need to be rendered.  This system has obvious advantages and works really well in PvE, but in large battles characteristic for WvW the weakness of this system occasionally surfaces.

Furthermore there are some client-side problems which impact how our culling system works.  After a client is notified that a character is present some time is needed to actually load the resources required to display this character.  The time it takes to load those resources depends on the power of the client computer (RAM, CPU, CPU cores. disk i/o).  One of our game engine programmers recently completed an optimizing routine which will improve character load times and we will soon™ see those improvements.  Still, the main issue with culling is server side an how fast your computer can load character resources is this irrelevant [if your client doesn't even know that the character exists in your vicinity].

As you may know, we implemented a solution to the current culling problems in the sPvP part of the game.  Since in sPvP significantly fewer players present at any one time we were able to implement this solution immediately without to worry too much about unintended consequences.

WvW on the other hand functions on a much larger scale than sPvP and thus we have to deal with additional issues before we can implement a server-side fix.  We need to ensure that clients, especially clients which only meet the minimum system requirements, can render and process more characters.  We also need to ensure that there the bandwidth required [to play the game] remains within a reasonable range and the published minimum requirements.

The WvW team is currently working an both of those issues.  Any changes we will be implementing will be complex and will have massive implications on how the game engine works.  Because those changes are so complex and affect core systems of the game engine it will take some time to properly implement those changes.  Thus I am unable to announce any kind of ETA.

Our goal is to drastically improve the game experience during large WvW battles and to significantly increase the number of characters that can be rendered by the client.

---- this is the end of the first post ---

In response to a player question whether it wouldn't be easier to just display stick figures in WvW Ramon replied the following:

We are looking into it.  Of course it's supposed to look better than just stick figures yet offer a significant improvement in performance.  For example, a standardized character model for male human characters or some such that only 10 different models have to be preloaded yet the immersion isn't completely hosed.

---- this is the end of the reply to that question ---

In reply to the suggestion to color code players:

Right, thus it's our goal to have a well defined finite solution to how characters are displayed, for example those 10 placeholder models, 5 races, 2 genders, potentially 8 color options for classes, or perhaps only 3 colors to indicate which world players are from.  Those models would be permanently preloaded and would thus be instantly available when needed.  The theorycrafting is quickly and easily done though implementing such a solution will take quite a bit of programming skill.

#3 News

News

    Internet Information Extractor

  • Community Support
  • 98 posts

Posted 22 November 2012 - 04:08 PM

So as it turns out there actually is a post on the US ArenaNet forums about this, it's just that @guildwars2 didn't tweet about it and/or it was otherwise missed somehow:

HabibLoew said:

tl;dr: We are working on a fix (a collection of fixes, really) for culling and character loading issues. The fixes require significant changes to a number of game systems and thus require time to implement. We’re not yet ready to discuss a release date for the current set of fixes, but we’re working hard to improve the experience as quickly as we can.


Now for the long version:
We currently use server-side report culling to limit the number of characters that any given game client is aware of. By limiting the number of characters that we report to any given client we also limit the bandwidth used (by the server and the client) and avoid situations where the client is overwhelmed by the number of characters that need to be processed and rendered. While this system has some obvious advantages, and it works well in PvE, the large battles that are the signature of WvW tend to highlight the deficiencies of this approach.

There are also some client-side issues which have contributed to the perception of how our culling system works. Once a character is reported to a given client there’s a non-zero amount of time required to load and initially display the assets associated with that character. Extra load time varies depending on how beefy the client machine is (those with more memory, faster CPUs, more CPU cores, and faster drives experience shorter load times). One of our engine programmers recently completed an optimization pass on the character loading process and so we should be seeing improvements to that part of the issue very soon. Even so, the bulk of the issue remains with the server-side culling as it doesn’t matter how fast your client can load and draw a character if it hasn’t even been told that character should exist yet.

As you may have heard we already have a fix for the server-side culling implemented for sPvP. Because sPvP has dramatically less players we were able deploy our fix immediately without worrying about downstream side effects. WvW, however, operates at a much larger scale than sPvP and so we have a number of additional hurdles to clear before we can turn on the server-side fix. In order to address the culling issue we need to ensure that clients, including min-spec clients, are able to handle rendering and processing many more characters. We also need to ensure that the bandwidth needed by any given client remains reasonable and falls within our min-spec for connectivity. The WvW team is working to address both the bandwidth and the client performance issues even now. The changes that we’re making are complex and have a large impact on the way the game engine works. Because of the level of complexity involved, and the core systems that are impacted, these fixes take time to implement correctly. As such, I can’t give you a date when we’ll be done.

At the end of the day our goal is to dramatically improve the experience of large battles in WvW and provide a substantial increase to the number of players that can be seen by any given client.

So the German Community Manager translated it to German, then I translated it back to English, if nothing else it shows you whether I was close enough with it. ;)

source: https://forum-en.gui...-fix-it-culling


HabibLoew said:

I’d like to correct a common misconception that I see popping up regularly in these threads. The culling that exists in Gw2 today is the same as what we shipped with. It was not introduced after the game launched, but was in fact present on day one. I suspect that people didn’t notice it as much in the beginning due to the game being all shiny and new, but it has been there all along.

I would also like to add a personal observation: I understand that the current state of culling is frustrating. It frustrates us too and we’re working hard to fix it. Often in these threads people propose a variety of options for addressing the issue along with the suggestion that their proposed changes are simple. This is a tempting line of thought but in my experience very little of what goes on in game development is actually simple.

In particular, the changes required to deal with culling involve fundamental changes to core systems. Both the changes and the systems involved are complex, which is why it’s taking us so long to implement them. As I hope I’ve made clear in my posts fixing the culling issues is a major focus of the WvW team right now and I firmly believe that we will be able to achieve significant improvements when our current development efforts are complete.


#4 Snype Cyclos

Snype Cyclos

    Hungry for Pizza

  • Established Members
  • 488 posts
  • LocationSeattle, WA
  • Guild Tag:[HC]

Posted 23 November 2012 - 01:49 AM

The main reason I'm glad about this is to stop the people who have never designed a game in their life and continue to say, "This doesn't seem like that hard of a fix."

Posted Image


#5 Liric

Liric

    Senior Member

  • Established Members
  • 897 posts

Posted 23 November 2012 - 03:20 AM

I'm fine with stick models. Anything, as long as it removes the problems of near perpetual invisibility and not seeing like 30 people until they're all on top of you.

#6 aphero

aphero

    Dedicated Member

  • Established Members
  • 87 posts
  • LocationRight over there
  • Guild Tag:[CoS]

Posted 23 November 2012 - 01:35 PM

Just render them like doom style sprites until the 3d model can be rendered.  So long as there is some form of a tell for their abilities and I can see them I'm fine with whatever I'm looking at.
Posted Image

#7 Fenris

Fenris

    Member

  • Established Members
  • 19 posts
  • LocationRedmond, WA
  • Guild Tag:[CLAW]

Posted 23 November 2012 - 02:24 PM

I would have to agree here.  Rendering something just so we know there are 20 or 30 enemy players charging us would be nice. I can deal with a simple indication as long as I know what I'm facing and how many then wait for the 3D full mode to populate later.  It's also nice to know it's primarilly a server side issue not a client issue so doesn't matter how powerful your PC is it's the server doing the culling.
Officer of Tyrian Claws 500 Charr Strong (mostly) Ehmry Bay Guild
80 Guardian | 100% World Completion | Master Armor/Weapsmith
Current Legendary in Progress : Twilight

Posted Image

#8 Lovisa

Lovisa

    Dedicated Member

  • Established Members
  • 96 posts

Posted 23 November 2012 - 02:39 PM

News, you're hilarious :)

#9 Chappy

Chappy

    Senior Member

  • Established Members
  • 290 posts
  • LocationSydney, Australia
  • Guild Tag:[TDE]

Posted 26 November 2012 - 08:36 PM

Why can't we have player control over the number of other players that can be visible?

Surely somebody running a tank of a PC will have little trouble handling 2 or 3 times the current cap.
Posted Image




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users