devtrackers.gg
7 months
ago -
Ashes of Creation
-
Direct link
Transcript (by Youtube)
| 0s | [Music] |
|---|---|
| 15s | hello everyone and Welcome to our |
| 18s | glorious ashes of creation June |
| 20s | development update we hope that everyone |
| 22s | has been well and safe we know this |
| 24s | one's coming a little hot for you that |
| 26s | just means you got to look on the bright |
| 27s | side you get two in July technically I'm |
| 30s | no this this is like a Fourth of July |
| 32s | stream we should have had uh we |
| 34s | fireworks and cration just explode some |
| 38s | money um obviously you see a different |
| 41s | face uh here as well I'm Margaret CR |
| 44s | your director of communications and with |
| 45s | me as always is Steven Sharie our |
| 47s | creative director but we have uh Brian |
| 49s | lford who you've seen on the stream |
| 51s | before but he's joining us for this |
| 53s | little special one and a while Ash's |
| 55s | merch you're G to make him |
| 58s | jelly yes look good yeah you can see |
| 61s | some behind him as well but uh as a |
| 64s | little sneak peek we are working on |
| 65s | merchandise we know that we've been |
| 66s | getting a lot of people asking us for it |
| 68s | takes time you have to go through |
| 70s | manufacturing and you have to find the |
| 71s | right people that you want you have to |
| 73s | get all your designs ready and then it's |
| 75s | back and forth getting samples and then |
| 77s | you have to have our models inhouse like |
| 79s | Brian that's testing everything yeah you |
| 83s | got for us um but yeah so we'll just get |
| 88s | right into all of the goodness but as |
| 89s | you can see both below us what we're |
| 91s | going to be covering today is reminders |
| 93s | server meshing technology Studio update |
| 96s | art update and we'll have some Q&A from |
| 98s | the forums we'll also be pulling some |
| 100s | live questions as well as we usually do |
| 102s | after the server meshing technology |
| 104s | stuff so stay tuned for that um but |
| 107s | before we begin those things we have a |
| 110s | our quick reminders and the first one |
| 113s | here is of course our YouTube Spotlight |
| 115s | and if you would like to be spotlighted |
| 117s | for our YouTube uh Spotlight all you |
| 119s | have to do is subscribe to over on |
| 120s | YouTube make sure that you are your |
| 122s | subscription shows publicly it's a |
| 125s | little weird thing um and leave a |
| 127s | comment on our development update the |
| 129s | full development update and you could be |
| 131s | chosen for next month so this one is |
| 135s | from Kevin um and they would like to |
| 137s | know love the vid are there any plans in |
| 141s | the future of making skills that require |
| 143s | team effort like 10 Summoner players in |
| 146s | a guild sacrificing 50% of their HP to |
| 148s | summon a giant Golem that can break down |
| 151s | walls and Gates and sieges or act as a |
| 155s | front to block damage from enemy arrows |
| 158s | maybe 10 Mages sacrificing 50% of their |
| 160s | Mana to cast an ice storm in certain |
| 163s | area that significantly reduces movement |
| 165s | speed which can be used on holding down |
| 168s | choke points that was a lot they had |
| 170s | they had some imagination go into that |
| 172s | one I mean regardless regardless of the |
| 175s | specific examples there right um I think |
| 178s | that the Crux of the question is do we |
| 181s | intend for there to be Synergy between |
| 184s | multiple archetypes or different |
| 186s | archetypes in skill activation and I |
| 189s | think we've talked about this in the |
| 191s | past we gave an example actually I think |
| 193s | I heard a Gollum in this example but we |
| 195s | gave an example of Summoners summoning a |
| 197s | you know using combined powers to summon |
| 199s | a larger Golem there is a desire to |
| 203s | explore that path I think our ability |
| 205s | system facilitates that already um one |
| 209s | of the things that we look at like right |
| 210s | now with the current archetypes for A2 |
| 213s | there's a wall ability on the tank it' |
| 215s | be nice that as you're in a party or in |
| 217s | a raid you as a tank have the option to |
| 220s | set on that ability whether or not you |
| 223s | want that wall to connect with others so |
| 226s | that it it it uh it snaps to another |
| 229s | wall potentially creating a stronger |
| 231s | barrier those types of interactions I |
| 233s | think are interesting I think they're |
| 234s | cool I know the combat team thinks |
| 236s | they're interesting and cool um we |
| 238s | definitely want to aim for that being a |
| 239s | potential |
| 241s | um right now that hasn't yet been |
| 242s | implemented but it is something in the |
| 244s | future that I think we're open to and |
| 245s | want to explore really awesome and of |
| 249s | course we have our Dev discussions we |
| 251s | love whenever you all uh give us your |
| 253s | feedback we're wrapping up The Siege in |
| 255s | war on so if you have any feedback on |
| 257s | that one definitely go go over there but |
| 259s | right now we have camera perspectives up |
| 261s | which we want to know about camera |
| 262s | perspectives what camera perspectives do |
| 264s | you prefer in terms of both angle and |
| 267s | distance from your character do you do |
| 269s | different situ ations call for different |
| 271s | perspectives let us know and of course |
| 273s | our next one coming up after that will |
| 274s | be storytelling with archetypes and as |
| 277s | we've alluded in the past already |
| 279s | there's some Bard action coming up in |
| 281s | the future so we know folks are excited |
| 284s | for that coming up you know what's |
| 287s | interesting um s sorry just it reminded |
| 289s | me um that |
| 291s | question a game I played a long time ago |
| 295s | had a um had these like three different |
| 298s | types of wizards one was wind one was |
| 301s | water and the other was fire and they |
| 304s | each had an elemental infusion that you |
| 306s | could cast on another wizard of a |
| 308s | different type oh I know what G you're |
| 309s | talking about yeah I like |
| 312s | that and if you did cast it there was |
| 315s | like a 10 to 20 minute long cooldown on |
| 318s | the infusion but it powered them up to |
| 320s | do a mega like Nuke right on a Target |
| 324s | now I didn't find the efficacy of that |
| 326s | you know to be anything particularly |
| 328s | game-changing it was a kind of nifty |
| 330s | for a single nuke but like what did that |
| 333s | do in the long run of me constantly |
| 335s | casting my normal nuke over and over and |
| 337s | over again right but I thought it was an |
| 339s | interesting attempt at creating some |
| 341s | interclass dependencies or Synergy from |
| 344s | a party perspective with those classes |
| 347s | and if you if you dive a little bit |
| 348s | deeper into that maybe you can come up |
| 350s | with some fun you know regular momentto |
| 352s | moment |
| 353s | gameplay 100% And with that we move |
| 358s | into our server meshing technology this |
| 362s | is going to be a long 1 hour and 18 |
| 365s | minute segment uh so buckle in if I told |
| 369s | everybody in the beginning I was in chat |
| 371s | I was like get your snacks ready because |
| 372s | you're going to maybe take a pee now and |
| 376s | and you know I as as we were talking |
| 379s | through this and you know Brian and I |
| 380s | have talked about this in the past as |
| 382s | well you you really kind of have to |
| 385s | wonder what segment of our community is |
| 388s | interested in this depth of a technical |
| 391s | discuss that are here we told them in |
| 392s | advance coming we're like this is GNA be |
| 395s | very nerdy talk prepare it's a super |
| 397s | nerd that yeah totally yeah it is a |
| 399s | super it is a super nerd thing um and to |
| 402s | be honest with you you know it's it's |
| 404s | not what your general population cares |
| 407s | to see it all right they want to see the |
| 409s | game play they want to feel the game |
| 411s | play they want to talk about that and |
| 413s | that's like if you think about the ratio |
| 416s | of streams that we do what is gameplay |
| 419s | versus technical discussions maybe it's |
| 421s | like a 1 to 10 right it's like yeah it's |
| 423s | like a 1 to 10 or every 10 or so streams |
| 425s | we have a super in-depth you know |
| 427s | technical discussion this one let me |
| 430s | preface is going to be the most |
| 432s | technically oriented discussion we have |
| 435s | ever done as a live stream but we made |
| 438s | our lovely Mike Veo our Graphics |
| 440s | designer made some great IM imagery to |
| 443s | help thought it was |
| 445s | phenomenal so hopefully you're you won't |
| 447s | be too bored visually either I I I I |
| 451s | thought I thought it was phenomenal but |
| 452s | let me just let me just speak to this um |
| 456s | there is a lot of magic that goes on |
| 458s | behind the scenes of what makes these |
| 460s | kinds of games possible yeah and um |
| 463s | there are a lot of individuals who are |
| 465s | not the VFX artists who are not the |
| 467s | combat designers but who are in the |
| 471s | literal weeds of the deepest code |
| 474s | possible that is so nuanced and and |
| 478s | complicated |
| 480s | that it is you know without a without a |
| 483s | degree in this specific field you you |
| 486s | really have a very difficult time even |
| 488s | following along so I would just want to |
| 490s | give kudos to um the guests that we had |
| 494s | on this particular live stream Zach |
| 496s | Hunter andon they did a phenomenal job I |
| 499s | think even with the visual aid that that |
| 501s | Veo did which he did a great job with um |
| 504s | in kind of walking through from a |
| 506s | Layman's |
| 507s | perspective just the level of complex |
| 510s | that's necessary and also the magic that |
| 512s | we're Brewing up here at Intrepid um |
| 515s | because that magic is going to be on |
| 517s | full display for Alpha 2 you guys are |
| 520s | going to be experiencing it and without |
| 522s | an in-depth showcase like the one you're |
| 524s | about to have you would have no idea of |
| 527s | what's going on in the background it |
| 528s | would just feel like seamless gameplay |
| 530s | to you but there is really if the pi |
| 533s | folks cuz the pi folks who uh joined us |
| 536s | for our node Wars testing even realized |
| 538s | what was going on behind the scenes |
| 540s | because the Ser measuring technology |
| 542s | isn't just something we're talking about |
| 544s | today it exists in the game and they |
| 546s | experienced it so and we've been working |
| 548s | on for for years now right so um I think |
| 551s | they do because sometimes as as an |
| 553s | individual server worker might go down |
| 555s | we're like why did I get disconnected |
| 557s | and they get disconnected but I don't |
| 559s | know if they put two and two together |
| 560s | that it's the same technology for sure |
| 563s | yeah right absolutely so these guys were |
| 565s | so excited they were like we could have |
| 567s | talked even longer oh yeah for sure this |
| 570s | is total total nerdgasm area uh but um |
| 574s | let me say this so as a preface before |
| 576s | we get into this this is going to be up |
| 577s | on YouTube afterwards we're going to |
| 579s | post it there it's a it's a standalone |
| 581s | discussion like Margaret said hour 18 |
| 583s | minutes this is very in-depth technical |
| 585s | discussions we're going to answer some |
| 587s | questions afterwards so make sure you |
| 588s | guys are posting Margaret's going to be |
| 589s | watching um feedback Point necessarily |
| 592s | only uh we have our wonderful Community |
| 594s | team and and social team that will be |
| 596s | we'll be grabbing questions from all the |
| 598s | places absolutely so so make sure as you |
| 601s | guys are watching this you know H the |
| 603s | types of questions we would love to hear |
| 605s | from you is how are these types of |
| 607s | technical discussions received by our |
| 609s | audience do you enjoy them do you like |
| 611s | getting a look under the hood um try to |
| 613s | formulate you know the types of |
| 614s | questions you can better understand it |
| 615s | Margaret and the community team will |
| 617s | capture them we'll answer them |
| 618s | afterwards but please understand |
| 619s | everything here continues to be a work |
| 621s | in progress all of the conversations |
| 623s | that we're having about this technology |
| 625s | is what's going to be tested as part of |
| 626s | Alpha 2 um and sit back and enjoy and |
| 629s | insight to what most companies really do |
| 632s | not ever reveal about what's under the |
| 634s | hood all right time to get nerdy see |
| 638s | y'all on the flip side |
| 643s | [Music] |
| 661s | hello everyone Welcome to our glorious |
| 663s | server mesh technology showcase segment |
| 667s | exciting it is I'm Margaret cron your |
| 671s | director of communications and of course |
| 673s | with me as always is Sten Sharie our |
| 676s | wonderful creative director I'm excited |
| 678s | we have a fun showcase today I mean it's |
| 681s | this is something that I don't think a |
| 682s | lot of companies go into a lot of |
| 685s | projects don't really talk about and and |
| 687s | you know it can be a little bit on the |
| 688s | techie side but I think this is stuff |
| 690s | that a lot of players uh like hearing |
| 693s | about and I know we got some we got some |
| 695s | exciting people to chat with today what |
| 697s | do you think Mar and um of course we |
| 699s | tried to make it as visually appealing |
| 701s | as possible shout out to Mike beio as |
| 703s | you you may know him from all of our |
| 706s | other Graphics but he's our Graphics |
| 707s | designer and he did a wonderful job on |
| 709s | these slides so hopefully they help |
| 712s | portray what our wonderful engineering |
| 714s | team is trying to toh showcase so we |
| 718s | will kind of introduce them we have some |
| 720s | new people who you've never met before |
| 723s | we'll kick it off first with Zach if you |
| 725s | want to talk about who you are what you |
| 727s | do at the studio and maybe a little fun |
| 730s | fact hi guys uh I'm Zach I'm the |
| 732s | technical director here uh at Intrepid |
| 735s | Studios I run the networking team and |
| 738s | the engine team |
| 740s | um I'm generally responsible for |
| 743s | performance and making sure that the |
| 744s | game scales to the amount of players |
| 746s | that we wanted to a hefty and lofty goal |
| 750s | many of us players will have thoughts on |
| 754s | that and of course we have Hunter with |
| 756s | us which some of you if you hang out on |
| 758s | our Discord maybe have interacted with |
| 762s | before hi yeah thanks for having me um |
| 765s | my name is Hunter I am a game networking |
| 768s | engineer at inet studios I've been here |
| 770s | for almost about four years now and |
| 773s | first initially started off as a |
| 774s | gameplay engineer and then I slowly |
| 776s | worked my way over to networking because |
| 778s | that's what I've come to find I love and |
| 780s | I love solving the hard problems and |
| 782s | very excited to Showcase to you all what |
| 783s | we've been working on the gameplay |
| 785s | engineering just wasn't torture enough |
| 787s | for Hunter so he wanted to go straight |
| 790s | straight The Impossible |
| 792s | problems oh my God I like I like solving |
| 795s | problems like this hard problems I think |
| 797s | Sol problem F problem to solve yeah um |
| 800s | and then of course last but not least |
| 802s | Anton hello I'm Anton uh I'm a senior |
| 806s | networking engineer over here at |
| 808s | Intrepid I moved straight into this |
| 811s | position Cu uh yeah I wanted to suffer |
| 815s | immediately I didn't want to start with |
| 817s | a game |
| 820s | play all right well let's kick it off |
| 823s | Zach I think you have a little intro |
| 825s | here for us as we move |
| 827s | in yeah um so Zach just I'm sorry just |
| 830s | to start a lot of people you know |
| 833s | sometimes don't know what networking is |
| 836s | yeah talk to talk to us about what is n |
| 839s | networking what is a network layer at at |
| 842s | at its core networking is just the |
| 845s | concept of moving EST state from the |
| 848s | server to the clients that that at its |
| 852s | core that is the problem we are trying |
| 853s | to solve we are trying to do that as |
| 855s | fast as possible with as many players as |
| 858s | possible um since we are making a very |
| 861s | very large game uh we have a lot of |
| 864s | problems that most games don't usually |
| 868s | have yeah and and you know with with |
| 872s | regards to like an mmpg versus like a |
| 874s | peer-to-peer versus like a small |
| 875s | dedicated server like our networking |
| 878s | needs are intense because the types of |
| 881s | numbers we're looking at are extreme |
| 885s | yeah we we we every everyone on the |
| 887s | networking team here has worked on other |
| 888s | multiplayer games before and this is |
| 891s | definitely the project where we have to |
| 893s | get the most creative with how we are |
| 895s | going to solve uh our scaling problems |
| 898s | you know like we |
| 900s | we we we are working on an MMO most MMOs |
| 902s | don't don't do what we're about to do |
| 904s | here they don't talk about how they you |
| 906s | know made it so they can support |
| 907s | thousands of players per server um so |
| 911s | it's very much we have to you know kind |
| 912s | of like figure out what what has been |
| 914s | done and and do a lot of um R&D to |
| 917s | figure out what what we need to do for |
| 918s | our game right and kind of answering |
| 921s | that question of what we need to do for |
| 923s | our game starts out with well what is |
| 925s | our game trying to achieve right yes |
| 927s | yeah there are a lot of different types |
| 929s | MMOs some that heavily you know utilize |
| 932s | instancing um some that you know maybe |
| 935s | don't care about replicating to hundreds |
| 937s | of players at a given time um so first |
| 940s | it starts with a vision what is ashes of |
| 943s | creation and then that kind of leads you |
| 946s | down the architectural path of okay what |
| 948s | does that mean we have to set up because |
| 951s | this isn't out of the box Unreal Engine |
| 953s | five this is a custom Network layer yeah |
| 956s | yeah this is this is this is fully |
| 957s | custom Taylor Made for the needs that |
| 960s | ashes has like this is not none of this |
| 963s | is is is off the shelf none of this is |
| 965s | standard unreal this is all built for |
| 967s | the kind of game that we are |
| 969s | making right and so how do you go about |
| 972s | kind of you know determining that that |
| 973s | architecture what does that look like |
| 975s | for a game like ashes and and you know |
| 977s | we've been working on this for a number |
| 978s | of years yeah for years um so I gu I |
| 981s | guess the first the first step is we |
| 983s | need to figure out the requirements like |
| 986s | what do we actually need to like what |
| 989s | what what are Ash's requirements um you |
| 992s | know the the first place we start is you |
| 995s | know we're an open world PVP game so we |
| 997s | can't really control where players go we |
| 999s | can't do match uh typical matchmaking |
| 1002s | where we just you know spin up more |
| 1003s | servers SP put them on the new |
| 1006s | matchmaking servers and you know not |
| 1008s | have these issues we're an open world |
| 1009s | game anyone can go anywhere um you know |
| 1013s | we're a highly PVP game so we need to |
| 1015s | have low latency people can't just uh we |
| 1018s | we can't do what other games do where |
| 1020s | you just like lower your tick rates |
| 1022s | really low and |
| 1024s | then um just simulate at a really really |
| 1027s | low rate since we're you know action PVP |
| 1030s | we can't do that um one of our I guess I |
| 1034s | I I would do you call this a game design |
| 1036s | pillar Stephen like this is we that we |
| 1038s | don't want to do layering like that's a |
| 1040s | that's that's a very like |
| 1042s | absolutely game thing we wanted we |
| 1044s | wanted so like you can go anywhere in |
| 1047s | the world stand there and and the people |
| 1050s | who are there are there like |
| 1053s | yeah and I was just going to say like |
| 1055s | game laring you know obviously everyone |
| 1057s | on this call is a gamer who's played |
| 1059s | MMOs and we played games that do that do |
| 1061s | layering yep you know it's just a very |
| 1063s | different experience when you can switch |
| 1065s | channels you know yeah yeah yeah it's |
| 1069s | like a different world a core like MMO |
| 1073s | memory for me is like you know old wow |
| 1076s | you would go to the same place like |
| 1079s | uh and and there would be the same |
| 1081s | person there you know dancing on top of |
| 1082s | a mailbox every single time or dancing |
| 1085s | on top of a corpse and looking at you |
| 1087s | with Hungry Eyes they were they were |
| 1089s | always they were always there and they |
| 1091s | were always and they were always there |
| 1093s | and with layering you don't get that uh |
| 1095s | you don't get that that same level of |
| 1096s | experience that we that we that we want |
| 1098s | to give uh to our |
| 1100s | players um and then I guess our last |
| 1102s | requirement which is like you know just |
| 1103s | a typical MMO requirement is that we |
| 1105s | have you know very very high player |
| 1107s | count goals um |
| 1109s | and you know typically uh High player |
| 1112s | count goals with no layering don't |
| 1114s | really go together so right that's where |
| 1116s | it comes into like you know we need to |
| 1118s | get really creative with how we're going |
| 1119s | to solve these problems and we and some |
| 1120s | of us you know Hunter Anton we've played |
| 1124s | these games that have high player |
| 1125s | accounts but a lot of ways that they get |
| 1128s | around having higher player accounts is |
| 1131s | by being less server adjudicated less |
| 1134s | server authoritative and more player |
| 1136s | which has its own risks and isn't really |
| 1139s | preserving the Integrity of that player |
| 1142s | experience indeed like some some games |
| 1145s | do not um do proper physics simulations |
| 1148s | on the server side to validate player |
| 1150s | movements in order to achieve that kind |
| 1152s | of thing which means like cheaters uh |
| 1155s | can have an easier time right going |
| 1157s | through walls and stuff like that or |
| 1160s | they well I guess they do layering or |
| 1163s | Shing where uh they yes they have real |
| 1167s | high player counts but technically not |
| 1170s | really cuz they're kind of broken up I |
| 1173s | can't tell you how many times I've I've |
| 1174s | jumped into either um you know some |
| 1177s | session based game or I don't want to |
| 1179s | call any any of them out but where |
| 1181s | there's just one single player that's |
| 1183s | either hacking or or or doing something |
| 1185s | and they're just running around and |
| 1187s | ruining the experience for everyone and |
| 1188s | it's because the way they've architect |
| 1191s | it in such a way it's more it's easier |
| 1192s | for those types of experiences to occur |
| 1195s | always trust the client right yeah |
| 1198s | exactly |
| 1201s | you just made everyone sweat Hunter do |
| 1203s | we want to kind of transition that into |
| 1205s | talking about the issues that we are |
| 1208s | trying to address yeah of |
| 1212s | course yeah so um I guess the issue |
| 1215s | start with it's like you know we're |
| 1217s | making MMO and unreal um the amount of |
| 1221s | entities we want on the server um it's |
| 1224s | just not really going to work with |
| 1225s | standard out ofth box unreal replication |
| 1228s | unre replication and unreal simulation |
| 1231s | so there there are two problems here |
| 1233s | it's like okay even if we could fit all |
| 1235s | the players we want on a single server |
| 1237s | um you |
| 1239s | know we we still can't replicate them |
| 1242s | because those are two separate problems |
| 1243s | we have the simulation and then the |
| 1245s | replication problem um so since this is |
| 1249s | is a problem and you know there there is |
| 1252s | no standard unreal solution for this we |
| 1255s | we we had to go about you know making |
| 1258s | our own customs solution to solve these |
| 1261s | problems um and this is where Intrepid |
| 1263s | net um I guess was born um entrep net is |
| 1267s | a fully distributed networking stack |
| 1269s | it's tailor made for as creation fully |
| 1271s | built in house um none of this is offthe |
| 1274s | shelf technology it's all custom built |
| 1276s | for our game |
| 1278s | um we you know we'll we'll get into |
| 1282s | these more in a bit but Intrepid net is |
| 1284s | made up of quite a few different |
| 1286s | Technologies um the the one that you |
| 1290s | know people are most interested in |
| 1291s | people are always talking about is |
| 1292s | server meshing we'll talk a lot about |
| 1294s | server meshing and how we how we |
| 1296s | implemented it for our game um we have |
| 1299s | inter server replication uh which is you |
| 1302s | know we have Ser we have multiple |
| 1303s | servers and they're replicating to each |
| 1305s | other how do we deal with that problem |
| 1307s | we have our own microservice layer for |
| 1311s | servicing our |
| 1313s | um uh uh communication I guess between |
| 1316s | our servers um we have uh Dynamic |
| 1321s | griding which is very exciting and you |
| 1323s | guys and you guys will like a lot and |
| 1325s | then we have the replication graph um |
| 1327s | which is just more um custom replication |
| 1332s | uh I would almost call them formulas um |
| 1335s | for our game so we'll be going into all |
| 1337s | this it's very this is very detailed |
| 1339s | very Tech heavy |
| 1341s | so buckle yeah we need some like theme |
| 1344s | music for the Intrepid net logo there |
| 1346s | too I know they were so excited when mik |
| 1349s | made a logo they're |
| 1350s | like nice of a logo I'm just thinking of |
| 1353s | like the Indiana Jones theme song right |
| 1356s | now right right |
| 1358s | totally so where do we start off with |
| 1361s | this yeah so I'm gonna kick it over to |
| 1363s | Hunter and he'll kind of like start with |
| 1366s | the like giving some of the base |
| 1367s | knowledge for like how unreal works and |
| 1370s | then he'll start with you know talking |
| 1372s | about how some of our server technology |
| 1373s | our server meshing technology |
| 1376s | works yeah hello um so before before we |
| 1379s | start diving too deep into Intrepid net |
| 1381s | um there's probably a few terminologies |
| 1383s | that we wanted to preface with just so |
| 1385s | that it helps the community and everyone |
| 1386s | else here follow along better with like |
| 1388s | the lingo and the terminology that we |
| 1389s | will be |
| 1391s | using um the first concept that we |
| 1393s | wanted to touch on is an actor um in |
| 1396s | Unreal Engine you can think of an actor |
| 1398s | as like almost any physical |
| 1399s | representation of an object or an entity |
| 1402s | that's inside the |
| 1404s | world as you can see on the photo that |
| 1407s | an actor can be anything such as as an |
| 1409s | NPC a mount a tint a piece of card cargo |
| 1414s | um pretty much anything that you can |
| 1416s | visually see and represent is what we |
| 1418s | call an actor and these actors they can |
| 1420s | be replicated which we'll get into on |
| 1422s | the next slide and PC would be player |
| 1425s | character for those who don't |
| 1429s | know personal |
| 1433s | Caravan so touching on replication here |
| 1436s | um replication is a term you might often |
| 1438s | hear in network working I believe Zach |
| 1440s | touched upon that a little bit briefly |
| 1441s | in his |
| 1442s | introduction um replication is the |
| 1445s | process of synchronizing some sort of |
| 1446s | State you can think of it like a value |
| 1448s | or a property that's on an actor on the |
| 1450s | server and then it's synchronizing that |
| 1453s | state from the server to all the clients |
| 1455s | that are relevant to the said actor um |
| 1458s | this is an important because of a |
| 1460s | concept called server authoritative |
| 1463s | networking um the server when you think |
| 1466s | of like networking in video games the |
| 1467s | server normally has a authority over |
| 1469s | your game world and the properties |
| 1471s | they're normally updated and checked |
| 1473s | against the server which can then be |
| 1474s | shared and synchronized between your |
| 1476s | clients which is through the form of |
| 1479s | replication um this is also part of why |
| 1483s | you never want to trust the client as we |
| 1485s | said earlier um because when networking |
| 1487s | you want when the client does some |
| 1489s | action to the server for instance you |
| 1491s | always want to verify that action on the |
| 1493s | server in some way or shape or that's |
| 1495s | how you can get hackers or cheaters or |
| 1497s | something disrupting your game because |
| 1499s | they're able to just send anything they |
| 1501s | want to to the server and your server |
| 1502s | will just trust it without verifying and |
| 1504s | then that's where you can get the |
| 1505s | concept of like I said hacking Andor |
| 1508s | cheating um so as long as the server |
| 1510s | kind of holds your source of Truth the |
| 1512s | client sends the data that your server |
| 1513s | doesn't agree with you can decline the |
| 1515s | actions and or kick or ban the user |
| 1517s | depending on how severe the action is |
| 1519s | they're taking that you're not trying to |
| 1524s | allow and obviously we've seen you know |
| 1526s | a lot of games that we've played in |
| 1528s | where |
| 1529s | as um as as the latency becomes uh worse |
| 1534s | for certain players or as the load on |
| 1536s | the server becomes higher um to process |
| 1540s | that replication can can show some weird |
| 1543s | effects uh in games right wait a minute |
| 1547s | what's that person doing that doesn't |
| 1548s | look right yep that's when you would |
| 1550s | that's typically when you would get like |
| 1551s | rubber banding like if your servers |
| 1553s | really slow down because you would only |
| 1554s | get an update every you know X period of |
| 1557s | time |
| 1559s | yeah so as you can kind of see in the |
| 1561s | the video we showed you can kind of see |
| 1562s | like player one is going into the water |
| 1565s | this is from the perspective of the |
| 1566s | server so as player one enters the |
| 1568s | server it'll be like okay player one is |
| 1569s | swimming so now player two and three |
| 1572s | since they're visually seeing player one |
| 1574s | swimming they're going to get a state |
| 1576s | for player one saying they are swimming |
| 1578s | and then that can kind of perform like |
| 1580s | the necessary actions of like seeing |
| 1581s | them play the swim animation or whatnot |
| 1584s | and and and and like I guess to add on |
| 1586s | to Hunter is saying this is the this is |
| 1588s | pretty much the entire problem like |
| 1590s | you're doing this all the time for |
| 1592s | everything in the game that's replicated |
| 1594s | all the time every movement every |
| 1596s | ability everything you click on interact |
| 1599s | with touch yeah all the Goblins that are |
| 1602s | running around they're playing |
| 1604s | animations uh the players moving stuff |
| 1606s | in their inventories the the trees |
| 1610s | leaves moving subtly yeah the the trees |
| 1613s | is falling being cut down |
| 1615s | yeah so not everything needs to be |
| 1617s | replicated across cross the server right |
| 1620s | some things can be client only like the |
| 1622s | ways the leaves might move specifically |
| 1625s | but when the tree gets cut down that |
| 1627s | needs to get replicated across yeah well |
| 1629s | and and and you you can do interesting |
| 1632s | things to try and like reduce this |
| 1634s | problem like like if we're talking about |
| 1636s | the wind for example you can just |
| 1637s | replicate you know the wind direction in |
| 1640s | speed yes and then the client can just |
| 1642s | apply that to all the trees instead of |
| 1644s | you having you know a specific |
| 1646s | replicated you know wind and speed |
| 1649s | for example right and that's part of the |
| 1651s | Finesse of replication trying to |
| 1653s | identify the problem exactly that's |
| 1655s | right yeah a big part of our job |
| 1657s | especially with working with gameplay is |
| 1659s | trying to figure out how to make that as |
| 1660s | efficient and optimized as possible so |
| 1662s | that our game and servers can perform to |
| 1664s | the expectations we need speaking of |
| 1666s | servers the server meshing what |
| 1668s | everyone's been excited to talk about |
| 1671s | right so now we do replication so now |
| 1674s | originally we just talked about how |
| 1675s | replicating a one server works so now |
| 1678s | first part of intrepid net is what we |
| 1680s | would like to call what we basically |
| 1682s | normally know as as server |
| 1685s | meshing |
| 1686s | um in Unreal Engine um your game |
| 1689s | normally works under a single process |
| 1690s | server scenario that means all your |
| 1693s | clients are connected to a singular |
| 1695s | server that runs the entire game now for |
| 1697s | an MMO RPG like as of creation um this |
| 1701s | can't work for us like for a few reasons |
| 1702s | Zach talked about in the previous slide |
| 1704s | there's just with the mass amount of |
| 1706s | entities and actors we have simulating |
| 1708s | and processing the replication of all |
| 1710s | that for everyone every actor in one |
| 1713s | server that's pretty |
| 1715s | insane um our solution first and |
| 1718s | foremost here is basically we had to |
| 1721s | modify unreal engin so that it is |
| 1723s | capable of having Min server processes |
| 1726s | that are kind of meshed together or |
| 1728s | quilted |
| 1729s | together that allows us to use all these |
| 1732s | servers together to make up your whole |
| 1734s | entire game world or your game |
| 1737s | room um this all us to split up that |
| 1739s | simulation work between many of the |
| 1741s | server processes and each server process |
| 1744s | has its own authoritative state of the |
| 1746s | world that it kind of owns and |
| 1749s | controls um this task it was not a |
| 1752s | trivial task and it required us ripping |
| 1754s | apart deep layers of unal engine and |
| 1756s | their networking stack in order to |
| 1758s | really modify it to fit the architecture |
| 1760s | that we have really |
| 1763s | envisioned um outside of this too |
| 1765s | there's like a lot of proprietary |
| 1767s | programs as well that we made that kind |
| 1769s | of run in the background to help it |
| 1770s | orchestrate all this functionality as |
| 1773s | well and by using underall engines API |
| 1777s | um and doing all this we were also able |
| 1779s | to add a lot of minimal overhead for our |
| 1781s | Engineers When developing for this |
| 1783s | solution and it allows us General |
| 1785s | control over like what how things |
| 1788s | replicate and where they replicate that |
| 1790s | we'll get into in the next few |
| 1793s | slides |
| 1795s | um so right now you see we have like the |
| 1798s | static Dynamic world grid okay that's |
| 1799s | great but now you're probably thinking |
| 1802s | okay now that we stitch many servers |
| 1804s | together what is my experience like that |
| 1806s | in regards to traversing the world will |
| 1809s | I be able to see into the other servers |
| 1810s | from mine or will I randomly just start |
| 1812s | seeing actors pop in and out of |
| 1813s | existence right in front of |
| 1815s | me that kind of takes us to this |
| 1818s | slide um so our answer to that of course |
| 1821s | is we want to make this experience as |
| 1822s | seamless as possible for our players if |
| 1824s | you cannot tell what's going on behind |
| 1826s | the scenes and everything that we've we |
| 1829s | are making then that means we did our |
| 1831s | job right if you can tell then it |
| 1833s | probably means something's wrong and we |
| 1835s | need to look at what we're doing here |
| 1836s | and how we can improve and optimize it |
| 1839s | further |
| 1841s | um we've developed what we call here |
| 1844s | inter server replication for this case |
| 1846s | which means that instead of traditional |
| 1849s | replication as you might have as we |
| 1851s | explained was like server to clients |
| 1854s | this is basically representing between |
| 1855s | servers now or replication between |
| 1857s | servers |
| 1859s | and we do this replication to what we |
| 1862s | call a proxy |
| 1864s | actor um as you can see in the video |
| 1866s | here we can see there was like a |
| 1868s | transparent white uh wall going across |
| 1870s | the screen that represents our in-game |
| 1872s | debug tools with our border regions |
| 1875s | being |
| 1875s | drone you can see that vexent is on one |
| 1878s | server and then Kevin is on another one |
| 1881s | as he's on the other side of that border |
| 1883s | so what vexin is seeing right now is |
| 1885s | actually not Kevin but a proxy of Kevin |
| 1888s | is what we have deemed it is it really |
| 1890s | Kevin yeah who knows if only could |
| 1893s | actually duplicate Kevin after expand a |
| 1896s | little bit on you know a lot of a lot of |
| 1898s | people who obviously aren't familiar |
| 1900s | with game development they might not |
| 1902s | understand the the terminology of |
| 1904s | debugging tools right so talk a little |
| 1906s | bit about what a debug tool is right so |
| 1909s | when making a networking solution as |
| 1911s | complicated as this we need a lot of |
| 1914s | information about the state of the world |
| 1916s | what's happening where things are going |
| 1918s | on |
| 1919s | so we've developed a lot of like |
| 1920s | debugging tools that can be used inside |
| 1923s | the game and outside of the game which |
| 1925s | these tools kind of help us collect that |
| 1927s | information see where all the important |
| 1929s | things are happening see where a failure |
| 1931s | might be going on so in this particular |
| 1934s | instance we have a debug tool that's in |
| 1936s | game that allows our developers to kind |
| 1938s | of draw these server borders so that you |
| 1941s | can kind of see while you're playing |
| 1943s | like okay there's a border here um and |
| 1946s | then if QA is testing something they can |
| 1947s | be like okay we transferred across the |
| 1949s | server and then this broke so that means |
| 1951s | something's probably wrong with the |
| 1952s | server transfer so that can kind of just |
| 1954s | help you like narrow down more of the |
| 1955s | problem space and give you that |
| 1957s | necessary information of like what was |
| 1959s | going on at the |
| 1960s | time um because without that if if we |
| 1963s | have it turned off you pretty much |
| 1966s | hopefully should not be able to tell the |
| 1968s | difference between the proxy we talked |
| 1969s | about or the actual |
| 1971s | actor um you can kind of think of the |
| 1974s | proxies here as a puppet any |
| 1979s | actual um action that was done by the |
| 1982s | actual actor on its own like Authority |
| 1985s | server which is the server that owns it |
| 1987s | it's then replicated to this proxy on |
| 1990s | your server which then gets replicated |
| 1992s | to yourself so that you can kind of see |
| 1995s | the representation of Kevin as if he was |
| 1998s | on the same server as |
| 1999s | you um we do it in a form of like a we |
| 2003s | call it lightweight because it's an |
| 2005s | up-to-date copy of the original actor |
| 2007s | but we can specifically our developers |
| 2008s | can specifically control like what does |
| 2011s | and doesn't get sent to this actor so |
| 2013s | that we're more optimal in like what |
| 2016s | we're sending how we're sending it and |
| 2017s | when we're sending |
| 2020s | it that was a |
| 2025s | mouthful I I think I think the other |
| 2027s | thing you know to think about Hunter |
| 2029s | Zach Anton is that um not a lot of |
| 2032s | players obviously follow development of |
| 2035s | the games that they're excited about |
| 2036s | right they they wait until |
| 2038s | you know the game is ready to play and |
| 2040s | and marketing begins and you know they |
| 2042s | can join in and and purchase or whatever |
| 2044s | obviously ashes has followed a little |
| 2046s | bit of a different path where we're |
| 2047s | giving people a bit of insight into you |
| 2049s | know what it actually takes to make |
| 2051s | these types of games and we generally |
| 2055s | tend not to delve too deeply into the |
| 2058s | highly technical components of our |
| 2060s | development process because that's where |
| 2063s | a little bit of a cliff starts to form |
| 2065s | of player interest even you know viewers |
| 2068s | ship interest um because uh it's nice to |
| 2072s | have the magic work in the background |
| 2074s | and you don't necessarily care or want |
| 2076s | to know how or why it's working just |
| 2078s | that it feels good that the you know |
| 2080s | it's responsive and that you know it |
| 2082s | accomplishes from a gameplay Loop |
| 2084s | perspective what you're expecting it to |
| 2087s | but you know because of the level of |
| 2089s | complexity because of the amount of |
| 2091s | effort because of what we're trying to |
| 2094s | create is a is a Next Generation level |
| 2098s | of complexity of gaming um we feel it's |
| 2102s | important to provide you guys an |
| 2105s | opportunity to get some insight into |
| 2108s | what we're doing that's unique into what |
| 2110s | we're doing that is going to service a |
| 2114s | long-term |
| 2116s | MMORPG at the scale that we need Ashes |
| 2119s | to be and that requires some really |
| 2122s | smart people in the room who are |
| 2124s | creating proprietary technology that |
| 2127s | allows us to achieve that in a way that |
| 2129s | presents players with a game at the |
| 2131s | Quality they expect and so I you know I |
| 2134s | think this is a lot of a lot of fun to |
| 2136s | kind of you know have you guys come and |
| 2138s | and explain this front facing it can be |
| 2140s | a mouthful it can be a headful to kind |
| 2143s | of digest but at the same time yeah I |
| 2146s | know we're nowhere near done but at the |
| 2148s | same time it gives you now an |
| 2149s | opportunity to to really see how the |
| 2152s | sausage is made um and I think that's |
| 2155s | cool right I can talk about this all day |
| 2159s | people don't talk about this there are |
| 2161s | so many games that are that are doing |
| 2163s | things like this not exactly but like |
| 2166s | you know there are other games that have |
| 2167s | to solve these problems and you know |
| 2169s | other other games don't talk about this |
| 2170s | you know this is this is a first |
| 2173s | absolutely |
| 2176s | right yeah so as you see in the video um |
| 2180s | you can see into the other server but |
| 2182s | now you're probably thinking okay what |
| 2184s | if I want to interact with this proxy |
| 2187s | how how do I do that across the server |
| 2189s | boundary well um because the proxy is |
| 2193s | like a puppet controlled by the higher |
| 2195s | authority we basically just now need |
| 2198s | events that we can send from the proxy |
| 2200s | up to its Authority and then we tell the |
| 2203s | authority that we want to modify data or |
| 2205s | execute some action on the authority |
| 2207s | server so that it can then replicate |
| 2209s | down to the proxy and then back to you |
| 2212s | um we have what we call a cross server |
| 2214s | event or an inter server event that's |
| 2216s | done through what it's normally known as |
| 2219s | an RPC or remote procedure call which |
| 2221s | allows pretty much just for |
| 2224s | this um our goal with entrep net is to |
| 2227s | make all the routing and the actions |
| 2229s | that we do as fast as possible so that |
| 2231s | we can keep the latency and interactions |
| 2233s | between servers a pretty seamless |
| 2235s | experience and you hopefully shouldn't |
| 2238s | even be able to tell the difference |
| 2239s | between like I said a proxy or an |
| 2241s | authoritative actor yeah and I guess to |
| 2244s | add here this is so important because |
| 2247s | you know these these server borders that |
| 2248s | we're talking about they're invisible |
| 2250s | you you you won't know where they are um |
| 2254s | so the the I think the big important |
| 2257s | thing here is like we need to have we |
| 2259s | need to make it so PVP can happen on |
| 2260s | these borders without people noticing um |
| 2264s | and that's that's that's that's really |
| 2266s | that's the real Crux of it here that's |
| 2267s | why it has to be so |
| 2268s | fast right and as you can see in like |
| 2272s | the video here we have some fighting |
| 2273s | across the border um you can see while |
| 2275s | we're fighting kind of over the boundary |
| 2278s | combat and everything seems pretty |
| 2279s | seamless and smooth as you would expect |
| 2281s | it without any noticeable issues um this |
| 2284s | is kind of showcasing like what we can |
| 2286s | do with just those events and that |
| 2288s | replication we talked about we can have |
| 2290s | these like in-depth battles or |
| 2293s | interactions between the servers with |
| 2295s | hardly any downside yep |
| 2298s | um so what what we're trying to achieve |
| 2302s | here is the seamless experience that you |
| 2304s | hopefully can't tell where that border |
| 2306s | crossing is and and the other important |
| 2308s | thing about this particular you know |
| 2311s | this cross server uh information and and |
| 2314s | the Border itself which I think you guys |
| 2316s | are getting into a little bit later is |
| 2317s | that these borders are not static they |
| 2319s | can change they are Dynamic you're |
| 2321s | you're spoiling it I'm I'm sorry I'm |
| 2324s | sorry I'm so sorry this is my first time |
| 2326s | seeing the |
| 2328s | slide okay go no no go I'm sorry I think |
| 2331s | you were talking about seamless |
| 2332s | transition so we |
| 2334s | can yeah so if you look in or go back |
| 2337s | one more Maggie I'm |
| 2339s | sorry um so if you kind of see in the |
| 2341s | video there as well um now that we can |
| 2344s | kind of do interactions to the proxies |
| 2346s | across the border you may have noticed |
| 2347s | in the video I did charge myself across |
| 2350s | the border and then I basically attacked |
| 2352s | what we call the proxy of |
| 2354s | Kevin however during that charge we did |
| 2358s | basically a server promotion that took |
| 2360s | place behind the scenes that you |
| 2361s | probably didn't |
| 2363s | notice um so if we go to the next slide |
| 2368s | we can kind of see now we're introducing |
| 2370s | the concept of server transfers this is |
| 2372s | kind of where Intrepid net gets very |
| 2376s | complex um this allows us to not only |
| 2379s | share and split the statically defined |
| 2381s | server areas but we can also transfer |
| 2383s | actors to new server areas that are |
| 2386s | moving across the world or happen to |
| 2388s | have a new server come up on top of |
| 2391s | it into a different server and then we |
| 2394s | do and then this allows us to do |
| 2396s | basically that Global load balancing |
| 2397s | across the |
| 2399s | room and then so basically when a player |
| 2402s | or an actor is within range of this |
| 2404s | border our system on the back end kind |
| 2406s | of begins a negotiation and the creation |
| 2410s | of a proxy so that it can replicate it |
| 2412s | to the nearby |
| 2413s | server um this doesn't only happen on |
| 2417s | like a one by one scenario that you |
| 2419s | might see here even if it's a 2 by two |
| 2421s | when you're in like the corner of four |
| 2422s | different |
| 2423s | servers your proxy will then get |
| 2426s | replicated be replicated to on three |
| 2428s | different servers from the one Authority |
| 2429s | so that way no matter where you are |
| 2431s | whichever server you're in around that |
| 2433s | area you'll still be able to see this |
| 2435s | actor on a different |
| 2440s | server |
| 2442s | um if that actor then fully crosses the |
| 2446s | Border going outside of the authority |
| 2448s | region of the server it's currently in |
| 2450s | um this the the two servers then kind of |
| 2452s | start negotiating what we call a |
| 2454s | promotion or a transfer this is when now |
| 2457s | server a is passing over that Authority |
| 2460s | we previously defined to server |
| 2462s | B and then promotions they're like a |
| 2465s | whole different Beast um in order to do |
| 2467s | a lot of like the promotions there's a |
| 2468s | couple things we have to do such as like |
| 2470s | ensuring the first you have to ensure |
| 2472s | the actor is like fully made with all of |
| 2474s | its components and |
| 2475s | children um we have to ensure they're |
| 2477s | all up to date with the entire state |
| 2479s | that was deemed necessary for the new |
| 2481s | copy to have because we don't want to |
| 2483s | lose any data during this entire |
| 2486s | transfer um |
| 2488s | and Hunter you people hear the term like |
| 2490s | you know children and the state and |
| 2492s | whatnot elaborate a little bit about you |
| 2494s | know what what that means at a high |
| 2496s | level right |
| 2498s | so when we have all these entities such |
| 2501s | as like an actor actors can then have on |
| 2504s | them what we call components which is |
| 2507s | basically building blocks that you can |
| 2509s | then add to your actor that kind of make |
| 2511s | up the actor what it does how it's |
| 2513s | supposed to act things like that um so |
| 2516s | for example you could have like an actor |
| 2518s | and then that actor has an interactable |
| 2520s | component so that means you can interact |
| 2522s | with this actor or it could have an |
| 2524s | inventory component so that means okay |
| 2527s | this actor has something of like a bank |
| 2528s | or a it it stores goods and to |
| 2533s | contextualize that let's say you know |
| 2534s | you have your Mount out you're not on |
| 2536s | your Mount right another player could |
| 2538s | walk up to your Mount and they would be |
| 2540s | prompted with an interact that says to |
| 2541s | Mount Your Mount as you're moving across |
| 2544s | the boundary that actor has that |
| 2547s | component that is the interactable |
| 2548s | component that needs to go with it right |
| 2551s | because if you transfer this actor or |
| 2553s | you're proxying it it still you still |
| 2556s | need to be able to do that interaction |
| 2557s | across the boundary so we proxy that |
| 2559s | interaction component and just like we |
| 2562s | could do cross server events on the |
| 2563s | actor itself we can also do them at the |
| 2565s | component level so that way it can |
| 2567s | communicate with its authoritative |
| 2569s | one um because we kind of treat the |
| 2571s | proxies as like a readon concept so the |
| 2575s | proxies are just visually kind of |
| 2576s | representating what Authority |
| 2578s | represents so any data modifications or |
| 2581s | anything that has to happen has to |
| 2583s | happen at the authoritative level so |
| 2584s | that way then the proxy can get updated |
| 2587s | appropriately across any server um |
| 2589s | that's kind of how we have like that |
| 2590s | Central truth of the actor is whatever |
| 2592s | server it's Authority on and there's |
| 2594s | only ever one copy of that I just want |
| 2597s | you to know this conversation so far has |
| 2600s | probably induced about 30 to 50 people |
| 2604s | now wanting to take computer science |
| 2606s | degrees that would awesome yeah no it's |
| 2609s | it's pretty cool stuff |
| 2612s | um it definitely took a lot of time and |
| 2614s | a lot of effort to try to figure out how |
| 2616s | we make these things possible and after |
| 2618s | all this work this is we're excited to |
| 2620s | be able to show it um although watching |
| 2622s | Kevin run back and forth is making me a |
| 2624s | little |
| 2626s | naous yeah so in the scenarios 2 this |
| 2629s | also works where you may not be proxying |
| 2631s | to that nearby server because say you |
| 2633s | teleport across the map um this kind of |
| 2637s | means that when promoting there it kind |
| 2639s | of does the same exact route except it |
| 2641s | may just need to make your actor instead |
| 2643s | of making a proxy of it so it all falls |
| 2645s | kind of under the same generalized um |
| 2648s | system y um and then when we normally do |
| 2653s | have that proxy actor on the other |
| 2655s | server that means when you're promoting |
| 2657s | we can reuse your existing two actors |
| 2659s | that are from server a and server B your |
| 2661s | Authority and the proxy and we can |
| 2663s | convert that proxy to the authority and |
| 2666s | then demote the authority one to a proxy |
| 2669s | so this allows us more to have that more |
| 2671s | seamless experience when you're swapping |
| 2673s | servers so that way it's a lot faster |
| 2676s | you don't need to send a lot more data |
| 2678s | to copy over to be created and then |
| 2680s | there's not really any visual hitching |
| 2681s | at all when this happens because we're |
| 2684s | not spawning or destroying a new one on |
| 2686s | the fly during that |
| 2688s | promotion and this helps also with |
| 2690s | client |
| 2691s | performance it it can um not going too |
| 2696s | too into detail but |
| 2698s | yeah actors aren't actors really aren't |
| 2701s | cheap so replicating an actor can be |
| 2704s | expensive and every new actor you kind |
| 2706s | of open and close to the client which is |
| 2709s | basically like a create or a destroy |
| 2711s | that just adds this additional overhead |
| 2713s | that you we didn't want so we can kind |
| 2716s | of negate it by just reusing the |
| 2718s | existing ones we already |
| 2719s | have and then doing that swap so that |
| 2722s | way you don't even notice anything |
| 2724s | happened cuz something Stephen no no he |
| 2728s | got it on the spawn and Destroy nice |
| 2730s | okay yeah yeah like like I said our goal |
| 2732s | is to make it as seamless as possible |
| 2734s | and one of the main pillars when we were |
| 2736s | making this uh entire system |
| 2740s | was we try not doing anything on the |
| 2743s | client if all of our actions are kind of |
| 2746s | done on the server in a fast and |
| 2748s | efficient way the client shouldn't |
| 2750s | notice anything happening at all and |
| 2752s | that's kind of where this comes in |
| 2754s | because we don't want the client to |
| 2756s | notice so every time we're doing |
| 2758s | something we're like okay how can we do |
| 2759s | this without affecting the client's |
| 2761s | experience because we don't want them to |
| 2763s | have a bad time |
| 2765s | yeah okay so um let's talk a little bit |
| 2771s | about how all this stuff is |
| 2773s | working uh because yeah we are |
| 2776s | communicating between servers we're |
| 2778s | sending events between servers we are |
| 2780s | transferring between servers but how is |
| 2783s | that information flowing how do the |
| 2785s | clients end up seeing that stuff |
| 2788s | and typically uh the way this is done is |
| 2793s | uh you create a centralized rep |
| 2796s | replicator and all the different servers |
| 2798s | from all across the realm send all the |
| 2802s | relevant information about all the |
| 2804s | different entities to that replicator so |
| 2807s | let's say one server has um a bunny and |
| 2810s | that bunny moved around it sends that |
| 2813s | information to the replicator and now |
| 2815s | the server has a goblin that took damage |
| 2817s | it will send that information to the |
| 2819s | replicator and then all the players will |
| 2822s | connect to that replicator and the |
| 2824s | replicator will pass pass through all |
| 2827s | those different |
| 2828s | entities decide which player should see |
| 2832s | which entities and send that information |
| 2834s | to the specific |
| 2837s | player however this creates a few issues |
| 2840s | and you might actually be able to see it |
| 2842s | from that graph there that we've just |
| 2844s | created a |
| 2845s | bottleneck uh as this singular process |
| 2848s | now may need to keep track of hundreds |
| 2850s | of thousands of |
| 2852s | entities and |
| 2854s | send potentially hundreds of millions of |
| 2858s | updates to all those clients because all |
| 2861s | those bunnies moving and the goblins and |
| 2863s | the players and the trees falling all |
| 2865s | that information has to be passed |
| 2868s | through the replicator and then sent to |
| 2870s | each individual client and I guess |
| 2872s | something to add here too Antoine is |
| 2874s | like this is the that that would just be |
| 2875s | moving the problem we're trying to |
| 2878s | like like if we had a single server this |
| 2880s | this would this would be the same |
| 2882s | problem where all all the replication is |
| 2884s | happening on one place and we're trying |
| 2886s | to we're trying to not do that Yes |
| 2889s | actually we tested a solution like this |
| 2891s | uh during alha one and this |
| 2895s | bottleneck it kind of affects how many |
| 2898s | players you can have in their realm as a |
| 2900s | total like you can spawn all the servers |
| 2902s | you want but one at some point you hit a |
| 2905s | limit on the replicator itself and it |
| 2906s | doesn't matter how many more servers you |
| 2909s | spawn you are limited by that cap on the |
| 2912s | replicator uh during alpha 1 the number |
| 2915s | we got was like I think about a thousand |
| 2918s | players where after hitting that point |
| 2921s | you can no longer really scale anymore |
| 2924s | and it's important to note that be |
| 2926s | because of this |
| 2928s | more I guess for lack of a better term |
| 2931s | archaic and monolithic approach which |
| 2933s | which many of us have likely experienced |
| 2935s | in older games |
| 2938s | there are certain scenarios in those |
| 2940s | games that are multiplayer where you |
| 2942s | start to see that server is having |
| 2944s | trouble and it's because it's hitting |
| 2946s | these types of Choke |
| 2948s | points yes and in fact when this |
| 2950s | replicator is overwhelmed the entire |
| 2953s | realm will lag it doesn't matter if |
| 2955s | you're alone in one server in a remote |
| 2957s | island you will lag because there is a |
| 2959s | big battle on the other side of the |
| 2961s | realm that's I remember in so many so |
| 2963s | many games where I would be out hunting |
| 2965s | and peaceful nobody's near nearby and |
| 2968s | I'm noticing everything is starting to |
| 2970s | churn and I'm like what is going on and |
| 2972s | then I'll look into Global chat and I'll |
| 2974s | see oh [ __ ] they're having a huge fight |
| 2977s | somewhere across the realm in a totally |
| 2979s | different part of the world and I am |
| 2981s | suffering as a result of it indeed and |
| 2985s | guess I guess it's a good bridge to our |
| 2988s | to what we're doing yeah it is a pretty |
| 2990s | good Bridge because basically we yeah we |
| 2994s | ases of creation is trying to aim for |
| 2997s | player counts that we found out that |
| 2999s | centralized replicator does not support |
| 3002s | so we had to come up with our own |
| 3004s | solution and uh this is from what we |
| 3007s | know not a very common solution possibly |
| 3011s | because it's more difficult to pull off |
| 3015s | uh what we basically did is we |
| 3016s | completely got rid of that centralized |
| 3018s | replicator and instead we had the |
| 3020s | servers themselves collaborate with the |
| 3023s | other servers around them and exchange |
| 3025s | information that way that and that makes |
| 3029s | each individual server have all the data |
| 3032s | they need to serve the clients that are |
| 3035s | connected to them uh in our scenario |
| 3038s | over here you can see we have four |
| 3040s | servers and two clients are connected to |
| 3043s | it's one of those servers if we had a |
| 3045s | centralized replicator all those four |
| 3048s | servers would be connected to that one |
| 3050s | replicator and then all eight clients |
| 3052s | would be connected to that one |
| 3054s | replicator too so we've actually |
| 3057s | distributed the problem we've eliminated |
| 3059s | the bottleneck and we've also eliminated |
| 3063s | this one point of failure that existed |
| 3067s | beforehand where if that one replicator |
| 3069s | fails the entire realm goes |
| 3072s | down and we can actually yeah have a |
| 3076s | look at the uh next slide where we see |
| 3080s | uh how we are more fault |
| 3083s | tolerant uh than if we have a |
| 3085s | centralized replicator |
| 3088s | uh and we can actually yeah you can swap |
| 3091s | to the next slide there you go we have |
| 3093s | the uh the entire server grid over there |
| 3096s | and if one server goes down everyone |
| 3099s | else can uh keep playing it like normal |
| 3103s | we patch that server up we fix the issue |
| 3106s | we bring it back up and everyone else |
| 3109s | never experienced anything the player is |
| 3111s | that we're connected to that one server |
| 3112s | got booted and then can rejoin just fine |
| 3115s | uh if we had a central leisure |
| 3117s | replicator if that replicator went down |
| 3119s | everyone gets booted |
| 3122s | so yeah so that's another uh another |
| 3126s | good uh benefit from our approach |
| 3128s | although the the really big gain we |
| 3130s | achieved with this was the elimination |
| 3133s | of the bottleneck uh and it actually |
| 3136s | lets us scale we are truly distributed |
| 3141s | we can just spawn new servers to a |
| 3143s | certain point of course and this is this |
| 3146s | is not and by the way like this is not |
| 3148s | just a strength obviously of the of the |
| 3150s | replication model of of the of the |
| 3152s | server grid itself you know being |
| 3154s | distributed in such a good way but when |
| 3156s | you have to bring such a monolithic |
| 3158s | server down because of that because of |
| 3160s | let's say whatever era you can't |
| 3162s | determine that has a Cascade effect |
| 3164s | across you know everything you're having |
| 3167s | to to affect the database to affect the |
| 3170s | login servers to have the player cues |
| 3172s | going off again the ability for this to |
| 3175s | truly be distributed and for us to be |
| 3178s | fault tolerant is an entirely different |
| 3181s | player experience when it comes to these |
| 3183s | large massive MMO RPG uh type settings |
| 3187s | yeah I mean it's it's very typical you |
| 3189s | know like with you know an MMO like that |
| 3192s | something else is happening in the world |
| 3194s | somewhere someone's hacking someone's |
| 3196s | you know there's a really large maybe |
| 3197s | there's a really large raid or or Siege |
| 3200s | happening and the the server Falls over |
| 3203s | and then everyone gets kicked with our |
| 3206s | model that doesn't happen the the rest |
| 3208s | of the server is fine performance is |
| 3209s | still fine in the rest of the server |
| 3210s | people can keep playing if that server |
| 3213s | goes down we you know we'll deal with it |
| 3215s | we'll spin it back up and it we'll just |
| 3217s | seamlessly rejoin the grid once it's |
| 3219s | once we're back in a good |
| 3222s | State and this like eliminating the |
| 3226s | replicator was um definitely not easy uh |
| 3231s | and it actually it caus us quite a few |
| 3234s | problems but let's investigate one of |
| 3236s | those problems |
| 3237s | and the solutions we came up to uh solve |
| 3240s | it |
| 3243s | so the because we did not want to create |
| 3249s | an really large connection network |
| 3252s | between all the different servers we |
| 3254s | essentially connect the servers to the |
| 3258s | nearby uh servers the neighboring or |
| 3260s | like some that are close |
| 3263s | by uh and that is fine in most |
| 3265s | situations over here here we can see |
| 3267s | there's two players in um two |
| 3269s | neighboring servers those those servers |
| 3272s | communicate so if those two players are |
| 3274s | in a party even though they might be a |
| 3276s | little bit too far away to see each |
| 3278s | other they can still see each other on |
| 3280s | the |
| 3281s | map however in a more complex situation |
| 3284s | where that other player is maybe on the |
| 3288s | other side of their realm now those |
| 3291s | servers do no longer have a connection a |
| 3293s | direct connection to each other so we |
| 3296s | have created a problem how can player |
| 3300s | one see player two when player two is on |
| 3303s | the other side of the world when if we |
| 3305s | had a centralized replicator sure we |
| 3307s | have the data just there we can figure |
| 3309s | it out but we do not uh what do we do |
| 3312s | about |
| 3313s | that and uh to solve uh that issue we |
| 3318s | created a concept called |
| 3321s | microservices microservices |
| 3325s | are Imagine them kind of |
| 3328s | like servers on top of the servers like |
| 3332s | um like some database for example but |
| 3334s | it's active it's not just storage |
| 3337s | they're doing |
| 3338s | work uh we have a node service that is |
| 3342s | responsible for um leveling up and |
| 3345s | keeping track of all the nodes and the |
| 3347s | um upgrades that have to be done to to |
| 3350s | nodes we have a population service that |
| 3353s | is doing the spawning across all the |
| 3355s | realms |
| 3357s | and we have a Social Service which would |
| 3359s | be for example what |
| 3362s | um what would affect the party example |
| 3365s | that we just uh talked about this means |
| 3369s | that server one and server 8 on the |
| 3372s | diagram can connect to that social |
| 3374s | service and exchange the locations of |
| 3377s | those two players that way player one |
| 3379s | and player 8 can have exactly the |
| 3382s | information they need to be able to see |
| 3385s | each other on the map |
| 3388s | and we act we yep what was it |
| 3391s | Stephen oh nope that wasn't me oh that |
| 3394s | wasn't you can't sorry my bad um so uh |
| 3398s | yeah and this actually gave us some |
| 3399s | extra benefits we were able to make some |
| 3401s | pretty cool tools with it because we |
| 3405s | wanted to be able to see to have like an |
| 3407s | overview of what happens to the entire |
| 3410s | realm we wanted to be able to keep track |
| 3412s | of all the players all the NPCs see all |
| 3415s | the different gatherable being spawned |
| 3417s | see all the |
| 3419s | trees and that's kind of hard to do if |
| 3422s | we just connect to a single server and |
| 3424s | then see what's on that one server and |
| 3426s | then jump on another server and see |
| 3427s | what's there so we created a service |
| 3430s | that keeps track of all entities in the |
| 3433s | world and a tool that can allow us to |
| 3436s | view those |
| 3438s | entities and this is a pretty cool tool |
| 3441s | because we can see the players moving |
| 3443s | around we can see them uh participa in |
| 3445s | events we can see the PC is dying we can |
| 3447s | see them collect |
| 3449s | Gables so if there's any issue with the |
| 3452s | spawning on GA of Gables we'll detect it |
| 3455s | and hopefully fix it it's I think the |
| 3458s | big thing to say here is like you know |
| 3460s | we can we can see the entire world state |
| 3463s | which we yes generally can't do because |
| 3466s | we're so distributed and everything's so |
| 3468s | server Central but since we have these |
| 3470s | Services we can make a debug service |
| 3472s | like this and we can see the entire |
| 3474s | world yeah watching |
| 3477s | yeah I also wanted to piggyback off of |
| 3478s | that Anton of your services um the |
| 3481s | services as well are also pretty much |
| 3484s | distributed like going for our model so |
| 3486s | that if one of those systems go down |
| 3488s | it's not taking down the entire all |
| 3490s | every game system so if we just if the |
| 3493s | chat service goes or the Social Service |
| 3494s | for example goes down you may just not |
| 3496s | be able to chat or form a party for a |
| 3498s | few minutes until it comes back up but |
| 3500s | you should still be able to interact |
| 3501s | with like your guilds your nod your mail |
| 3504s | Etc and we try making the service is as |
| 3507s | well fault tolerant like we explained |
| 3508s | earlier um so that way they can come |
| 3511s | back up go down come back up on the Fly |
| 3513s | get hot fixed and so that we can still |
| 3516s | provide that seamless experience while |
| 3518s | the feature may not be working |
| 3519s | everything else should still be working |
| 3521s | in the meantime right and because |
| 3522s | they're they're componentized in this |
| 3524s | you know microservices approach right |
| 3527s | there is not an interconnectivity where |
| 3529s | you have one failure in a single service |
| 3531s | causes a cascading effect across all the |
| 3533s | services right exactly they're they're |
| 3535s | independent of one another that allows |
| 3537s | for a recovery time a recovery process |
| 3540s | to stand back up while the other |
| 3541s | services continue their operations |
| 3544s | exactly and then these debugging tools |
| 3545s | allow us to see where those problem |
| 3547s | areas are or what what situation or |
| 3549s | actor might be causing a problem right |
| 3552s | exactly yeah we don't have all the tools |
| 3554s | slide for all the tools but we do have |
| 3556s | monitoring tools for all the different |
| 3558s | Services we know when they're lagging we |
| 3560s | know when they're down we know we can |
| 3561s | monitor pretty much everything it's |
| 3563s | pretty cool yes it's very cool |
| 3566s | cool now let's get let's get into the |
| 3568s | really fun stuff now next one is this |
| 3570s | one is something the community has |
| 3572s | actually been talking about quite a bit |
| 3573s | yeah I saw a bunch I saw a bunch of |
| 3575s | videos people are people are have been |
| 3577s | talking a lot about this yeah so go into |
| 3580s | it Anton because this is this is your |
| 3582s | thing yeah oh God I have been I've been |
| 3585s | looking forward to like since I joined |
| 3588s | the company I've been wanting to |
| 3589s | implement this and uh this is um an |
| 3595s | answer to a problem |
| 3597s | and that problem that is trying to |
| 3599s | answer is sure you have all those |
| 3602s | different servers but usually player |
| 3604s | activity tends to condense to certain |
| 3607s | areas let's say a large PVP battle is |
| 3610s | happening at some place um and when that |
| 3614s | happens that one server can be |
| 3618s | overwhelmed and you usually wouldn't |
| 3620s | want to just create tiny servers for |
| 3624s | everything because that can be |
| 3626s | incredibly expensive and |
| 3628s | wasteful but having those smaller |
| 3632s | servers would be really useful so what |
| 3635s | if you can do it dynamically what if you |
| 3637s | can detect when a server is overwhelmed |
| 3640s | and then split it this is what dynamic |
| 3643s | griding it is and it can be a little bit |
| 3648s | of a little bit tricky to implement uh |
| 3650s | the difficulty of it is that you just |
| 3653s | now split a server and you want the |
| 3656s | players to not realize they were casting |
| 3659s | abilities they were running around and a |
| 3662s | new server just spawned they were moved |
| 3664s | there and no one realized that is a |
| 3666s | difficulty there and something I want to |
| 3668s | add here is like these cases in which |
| 3669s | we're splitting a server they're complex |
| 3672s | because if we're splitting a server |
| 3673s | there's a lot going on so it's not like |
| 3676s | we can like there's just a couple people |
| 3678s | we can split them we can do like a a |
| 3679s | quick fade and it will be fine if we're |
| 3683s | splitting a server it's because it's |
| 3685s | it's underload |
| 3687s | and there's there's a lot of action |
| 3688s | going on so we want to keep that action |
| 3690s | going on and not interrupt it while we |
| 3692s | while we balance the load across |
| 3693s | multiple servers and and it's important |
| 3695s | to you know call back again to the very |
| 3697s | beginning of this conversation right how |
| 3700s | do you determine the types of |
| 3702s | architecture necessary you must first |
| 3705s | understand what is the mission statement |
| 3707s | of the game and Ashes is built with a |
| 3710s | lot of open world Dynamic and emerging |
| 3715s | player behaviors that are attempting to |
| 3718s | encourage High player densities at any |
| 3721s | point within the open World locations |
| 3724s | which makes this particular problem a |
| 3727s | requirement when determining |
| 3729s | architecture of how do we address this |
| 3732s | right and I'm not familiar with and you |
| 3735s | know Zach Anton Hunter I'm not sure if |
| 3736s | you guys are but I'm not familiar with |
| 3739s | any other Tex stack or networking |
| 3741s | approach that has a dynamic griding |
| 3743s | solution in order to better accommodate |
| 3746s | and facilitate that emergent player |
| 3750s | density Behavior yeah I mean we we are |
| 3753s | we are not the first people to try this |
| 3755s | this is this is something that a lot of |
| 3756s | people have tried it's it is you know |
| 3759s | the the answer to to scalability if you |
| 3762s | if you can get it right and that is you |
| 3764s | know that is our our mission here is to |
| 3766s | get this right we want to be the people |
| 3768s | that solve this problem and because if |
| 3771s | we solve this the the the the level of |
| 3775s | battles and uh dynamicism I guess we'll |
| 3778s | be able to have in ashes is going to be |
| 3780s | unparalleled to any other game yeah and |
| 3783s | to make something clear we're not |
| 3785s | talking about the far future 5 years |
| 3787s | from now uh this is current currently in |
| 3791s | development and we are uh hoping at |
| 3795s | least to have it available by Alpha 2 um |
| 3799s | and the server meing part of it is |
| 3800s | already in game yeah the dynamic aspect |
| 3805s | of yeah so the static part yeah is |
| 3806s | already is already going it's been going |
| 3809s | for manys now yeah years years now um |
| 3812s | and you know the the all of our Dev test |
| 3815s | servers people are on them every day |
| 3816s | running it it's fine uh this is this is |
| 3819s | like you know if you where where we're |
| 3821s | investing our Dev time right now is |
| 3823s | getting this |
| 3825s | perfect get getting this ready for the |
| 3827s | entrance of Alpha 2 yes and hopefully |
| 3831s | ases of creation is going to be uh the |
| 3833s | first game that implements this |
| 3835s | technology and actually deploys it yes |
| 3837s | that would be really fun you got to work |
| 3840s | quickly guys if you want that moving |
| 3845s | moving very cool this is |
| 3847s | exciting right do you want talk about |
| 3850s | the it up to uh to Hunter yeah yeah I I |
| 3853s | guess now that we've you know all these |
| 3856s | slides before this we were talking about |
| 3857s | Distributing the problem separating the |
| 3860s | problem um into multiple servers so we |
| 3863s | don't have to deal with like you know |
| 3865s | thousands and thousand thousands of |
| 3866s | players and on on the same server doing |
| 3869s | replication um but you know we still |
| 3872s | have like the core problem here right we |
| 3875s | can only split servers so far we can't |
| 3877s | you know make them like 10 meters by 10 |
| 3879s | meters we can't do that it's going to be |
| 3880s | too small um so we need to we need to |
| 3883s | still support you know a high you get a |
| 3885s | server you get a server exactly exactly |
| 3889s | so we still need to solve the like the |
| 3891s | core problem of how many players can we |
| 3894s | get on a server have them replicate and |
| 3896s | simulate at the same |
| 3898s | time so uh I'll pass it over to 100 |
| 3902s | he'll kind of go into like I guess the |
| 3904s | problems that we have when we're when |
| 3906s | we're with with with unal replication |
| 3908s | and how we've gone about uh speeding |
| 3910s | them up so we can uh support many many |
| 3913s | many uh hundreds of players on a |
| 3916s | server yeah um so kind of piggybacking |
| 3920s | off what Zach said like what do we do |
| 3922s | when a large amount of players are say |
| 3924s | following your favorite streamer into a |
| 3926s | a small area and then like we can't just |
| 3929s | keep dividing our servers into such tiny |
| 3931s | areas and |
| 3934s | that you're needing a single server |
| 3936s | that's still of a reasonable size to |
| 3940s | then be able to process the heavier load |
| 3942s | and replication that's happening in that |
| 3944s | exact area at that exact |
| 3946s | moment um so in order to do that we have |
| 3950s | to then figure out how we can increase |
| 3952s | the number of players and entities that |
| 3954s | a single server can support |
| 3956s | and this doing so will also help us |
| 3959s | increase our overall like concurrent |
| 3960s | user limits per realm um because as a |
| 3963s | rough napin math of course you know you |
| 3965s | take the players per server multiply |
| 3967s | that by the number of servers and |
| 3968s | there's your CCU right it's that easy |
| 3970s | yeah that's how it |
| 3973s | obviously um so I'm G to kind of go into |
| 3976s | now uh we're going to touch back up on |
| 3978s | kind of what we explained earlier like |
| 3980s | why is replication |
| 3982s | expensive like what's going on um |
| 3987s | so we kind of wanted to explain that |
| 3990s | like on this slide here for example you |
| 3992s | can see an example of a CPU performance |
| 3996s | profile that we took during a large |
| 3998s | scale bot |
| 3999s | test um you can see here that like the |
| 4002s | single frame we're staring at took 150 |
| 4004s | milliseconds for example now 150 |
| 4007s | milliseconds may not seem like much but |
| 4009s | when you want to for example you're |
| 4011s | talking about servers and tick |
| 4013s | rates um and if you're trying to Target |
| 4015s | say 30 or 60 frames per second or ticks |
| 4018s | per second server you only get a budget |
| 4021s | of about like 33 milliseconds or 16 Mill |
| 4024s | seconds per frame in order to keep that |
| 4026s | stable High tick time can you talk a |
| 4030s | little bit about what you're referring |
| 4031s | to when you're when you're talking about |
| 4033s | tick rates and ticks people |
| 4035s | not |
| 4037s | so as your game server runs |
| 4041s | every it it runs and processes basically |
| 4044s | the entire world at like an interval of |
| 4046s | ticks or frames is what it's called the |
| 4049s | faster you can do all that processing |
| 4051s | the faster your gameplay and experience |
| 4053s | is going to be um if you're targeting |
| 4056s | say roughly 30 to 60 frames per second |
| 4058s | which is a pretty good um targets to |
| 4061s | have for like a game server of like an |
| 4063s | MMO that needs to be very fast-paced and |
| 4065s | have like very good performance is |
| 4069s | different than client FPS yes but but |
| 4072s | they but they they might like 30 is not |
| 4074s | acceptable |
| 4076s | so they're vaguely coupled so you one |
| 4079s | one of the things you would notice like |
| 4080s | if server performance went down is you |
| 4082s | would notice rubber banding like when |
| 4084s | you see like people just jumping around |
| 4085s | the world or you would notice your |
| 4087s | latency goes up and you would notice |
| 4089s | that by like you're you know you're |
| 4090s | trying to you know do PVP and you feel |
| 4094s | you know you you you cast Fireball and |
| 4096s | you feel the latency between the time |
| 4098s | you casted it and when it actually |
| 4100s | executed on the server so the the this |
| 4103s | this this time these like these tick |
| 4105s | times that were talking about that's |
| 4107s | what it correlates to for the player |
| 4109s | right because normally when you do an |
| 4110s | action the client triggers something it |
| 4112s | tells the server hey do this and the |
| 4114s | server then replicates back to your |
| 4115s | client if these frames are taking a long |
| 4118s | time to process in between that means |
| 4120s | your server just has artificial latency |
| 4122s | and it's taking a lot longer to then |
| 4123s | process that incoming message to then |
| 4126s | return back to you um so like in this |
| 4129s | 150 millisecond scenario if our game was |
| 4132s | steadily running every frame at like 150 |
| 4134s | milliseconds of time to process |
| 4136s | that would drop our server down to |
| 4138s | roughly like six ticks or six frames per |
| 4140s | second which then you will really |
| 4142s | heavily around that time start filling |
| 4144s | like latency rubber banding uh dropped |
| 4147s | packets Etc and and then that's not |
| 4150s | acceptable for an action combat PVP MMO |
| 4153s | like it just doesn't work |
| 4156s | right so I played those games it |
| 4160s | definitely doesn't |
| 4161s | work we all have blamed lag before right |
| 4164s | right so so we know that this is |
| 4167s | unacceptable what do we have to do to |
| 4169s | optimize this right so now this kind of |
| 4172s | tells us replication and now it's like |
| 4174s | okay let's dive a bit deeper into why |
| 4176s | this replication is very |
| 4179s | expensive um one of the big excuse |
| 4184s | me one of the big um major factors |
| 4188s | replication can be slow is because first |
| 4191s | when you want to do your replication you |
| 4193s | have to gather and determine basically |
| 4195s | relevancy of every actor in the game to |
| 4199s | not only your connection but every other |
| 4200s | connection that's being processed in |
| 4202s | that |
| 4203s | frame so relevancy is used and needed to |
| 4207s | determine basically how far away an |
| 4208s | actor is from your player and then there |
| 4211s | might be some other like comment some |
| 4213s | other checks that they may need to do |
| 4214s | like is it visible or not but for |
| 4217s | Simplicity sake we're just going to base |
| 4218s | it off a distance for |
| 4219s | now so then it's basically comparing |
| 4223s | that distance to you and saying okay |
| 4226s | are you eligible for me to replicate to |
| 4228s | you because we obviously don't want |
| 4230s | things being considered to replicate to |
| 4232s | you if it's like across the world like |
| 4234s | you don't care what's happening on the |
| 4236s | far Northern end of the continent while |
| 4237s | you're all the way down on the southern |
| 4238s | end like that's just that's just going |
| 4240s | to waste a lot of processing and |
| 4243s | bandwidth um so the relevancy checks |
| 4248s | like this gets way more expensive the |
| 4250s | more players and more actors you have |
| 4251s | that are replicated because then it's |
| 4254s | for every actor for every player |
| 4256s | am I within range of them yeah and when |
| 4259s | we're talking about an MMO where you |
| 4261s | know you're going to have thousands of |
| 4262s | players on the server thousands of NPCs |
| 4264s | uh tens of thousands of replicated |
| 4266s | actors you know you can't you just can't |
| 4268s | check all you can't do a distance check |
| 4270s | on all of those things in a reasonable |
| 4273s | amount of time right and so this is an |
| 4275s | optimization here for networking |
| 4277s | purposes but it also leads to them |
| 4279s | slowing down your CPU resources because |
| 4283s | again you have to do that like every |
| 4284s | tick or at every ever interval you set |
| 4286s | to do these |
| 4287s | checks um as we if we looked at the |
| 4291s | previous example for example I believe |
| 4293s | the I don't think it shows in the |
| 4295s | profile but roughly a little bit over |
| 4297s | onethird of that total time in that |
| 4299s | profile was just doing the relevancy |
| 4303s | checks um a good thing about our model |
| 4306s | though because all of our servers are |
| 4308s | distributed they only have to do those |
| 4310s | relevancy checks for just everything |
| 4312s | within that then distributed Zone as of |
| 4316s | if we had that Central replicator it |
| 4317s | would that replicator would then have to |
| 4319s | still process every single player every |
| 4321s | single actor Etc in the world in some |
| 4323s | way shape or form to then determine if |
| 4325s | it should be relevant to you or not so |
| 4329s | by splitting up and doing the |
| 4330s | distributed server load like we're doing |
| 4333s | this allows us then to kind of reduce |
| 4334s | that problem area but it's still a |
| 4337s | problem yeah it's still really big |
| 4339s | because it's still a really big problem |
| 4340s | now adding more servers to the solution |
| 4342s | that can make it smaller and smaller but |
| 4345s | eventually there's like we want to get |
| 4348s | it as fast as possible so it's like how |
| 4350s | how how do we make this better how do we |
| 4352s | make this faster to be clear that |
| 4354s | profile was using uh inter it had a |
| 4358s | server messing and it still was 40 35% |
| 4362s | uh uh relevancy checks so we need to |
| 4365s | reduce those relevancy checks uh how how |
| 4369s | do we do that how can we tell unreal not |
| 4373s | to do all those unnecessary checks |
| 4377s | well unreal in recent years introduced |
| 4380s | something called the replication graph |
| 4382s | and that is something that gives the |
| 4387s | developers |
| 4389s | the ability to write a custom solution |
| 4392s | into how to handle all those uh relevant |
| 4396s | checks and how to determine which player |
| 4398s | should see which actors when it comes to |
| 4402s | replication uh and this can actually |
| 4405s | actually lead to some really great |
| 4408s | performance games we'll look at it in a |
| 4410s | bit uh but let's start and to and see |
| 4414s | how we use this replication graph to |
| 4417s | solve specifically this relevance |
| 4420s | check on the map over there we see uh we |
| 4425s | you still cannot see too far away but |
| 4427s | there are a couple actors that are |
| 4430s | really close to the player that Goblin |
| 4432s | over there in a loot box and there is |
| 4436s | also um a bear a little bit on a medium |
| 4440s | distance away and a bunny really far |
| 4442s | away now imagine that keep stretching |
| 4445s | for 500 meters more and there's 200 more |
| 4448s | bunnies and without the replication |
| 4450s | graph we would have to check every |
| 4453s | single one of those bunnies and say is |
| 4455s | this within my distance is this within |
| 4457s | my distance and that's that's a problem |
| 4460s | that's expensive to do so what we're |
| 4463s | going to use the replication graph |
| 4466s | um to do is we are going to create this |
| 4469s | grid where we eliminate all the cells |
| 4473s | that are far further away from us and |
| 4475s | only check the nearby |
| 4477s | cells that allows us to only check these |
| 4480s | three actors that are around us skipping |
| 4482s | potentially hundreds of others actors |
| 4484s | that are further |
| 4486s | away and uh this can lead to a pretty |
| 4489s | big performance Improvement we noticed |
| 4492s | and this is like highly varying |
| 4494s | depending of the scenario scario of |
| 4496s | course but on an average scenario we |
| 4499s | noticed about a 1.5 times performance |
| 4502s | Improvement like a 35% |
| 4504s | reduction uh in our replication |
| 4508s | cost this is this is pretty |
| 4511s | good uh but it's still not fast enough |
| 4515s | and if we analyze the problem a little |
| 4517s | bit we'll see |
| 4519s | that we are we we're talking and |
| 4522s | thinking about like the NPCs and the |
| 4524s | bear and the |
| 4526s | goblin but we are also comparing our |
| 4529s | distance with all the trees with all the |
| 4531s | bushes that you can pick up with |
| 4533s | potentially some buildings that are |
| 4537s | destructible why are we doing that can |
| 4540s | we do something to improve this and the |
| 4542s | answer is yes we can uh what we can do |
| 4546s | is we can monitor and see when these um |
| 4552s | ores or these trees are changing |
| 4555s | Ing and when they're not changing we can |
| 4559s | move them out of our grid so we're not |
| 4562s | even considering them at |
| 4564s | all uh we might have 2,000 trees in the |
| 4567s | area well we're not even going to look |
| 4570s | at them because none of those trees just |
| 4571s | got cut down if someone is cutting them |
| 4574s | down then yes we're going to move them |
| 4576s | temporarily into the special grid and |
| 4579s | then once they're done cutting out and |
| 4581s | everyone sees that they were cut down we |
| 4583s | put them back to the dormy bucket |
| 4586s | it's essentially looking imagine like |
| 4588s | the actor is going to slip for it a |
| 4590s | little bit and this this gives us a |
| 4594s | really really large Improvement we |
| 4596s | noticed in our test about five times |
| 4599s | Improvement in replication |
| 4601s | times so so |
| 4604s | far uh if we also include the previous |
| 4606s | optimizations we've improved our |
| 4608s | replication performance by seven |
| 4610s | times and yet that's still not fast |
| 4613s | enough it's still not fast enough one of |
| 4617s | mainly because of our worst case |
| 4620s | scenario and that worst case scenario is |
| 4623s | people keep packing in that one small |
| 4626s | area there's an event happening there's |
| 4629s | some point they need to capture 200 |
| 4631s | people run in the same area yes she can |
| 4633s | ignore the um relevancy to some Far Away |
| 4637s | tree that's great that's a game that's |
| 4640s | uh more than welcome but if 200 players |
| 4644s | can see each other |
| 4646s | that means that 200 time 200 you're |
| 4648s | sending 40,000 updates per tick yeah and |
| 4654s | I guess one thing to add there Anton |
| 4655s | this is typically when a game would |
| 4658s | introduce a new layer or um a new Shard |
| 4661s | this is like typically that case um but |
| 4664s | as we talked about earlier one of our |
| 4666s | goals is we don't want we don't we don't |
| 4668s | want to do laying we don't want to new |
| 4669s | sharting we want everyone who's in the |
| 4671s | same area to be able to see each other |
| 4673s | right and you could you you could then |
| 4675s | use like Dynamic greeting to help in the |
| 4677s | scenario but again we can only make the |
| 4679s | servers so small and dynamic before we |
| 4682s | kind of run out of that room so we still |
| 4684s | have to improve the scenario that's kind |
| 4686s | of happening around you indeed so we |
| 4690s | need to figure out a way to further |
| 4693s | improve this worst case scenario and |
| 4697s | this is where we kind of hit an engine |
| 4701s | limit um and this engine limit is that |
| 4706s | the replication graph of Andre |
| 4709s | is the replication tick in general not |
| 4712s | just the replication graph is single |
| 4714s | threaded and that's because the engine |
| 4718s | in general is single threaded now what |
| 4720s | does that mean what does single threaded |
| 4722s | mean it means that all the processing |
| 4726s | happens in one core of the |
| 4729s | CPU uh and uh we can move on to the next |
| 4733s | Slide by the way but |
| 4735s | but |
| 4739s | the that means that even though the |
| 4741s | server might have four cores um all the |
| 4746s | other three cores are going to stand |
| 4748s | there and wait for the first thread to |
| 4751s | do all the |
| 4753s | processing uh this is done normally in |
| 4755s | unreal because it's it was not designed |
| 4759s | to be |
| 4760s | multi-threaded uh it was it is a very |
| 4762s | single threaded engine and and |
| 4765s | replication is probably the most |
| 4768s | single-threaded system out of all the |
| 4770s | systems in |
| 4771s | unreal uh so making it multi-threaded is |
| 4777s | incredibly difficult and uh at least I |
| 4781s | am not aware of it being done before I |
| 4784s | tried to do some research around it at |
| 4786s | the time and uh everyone was saying |
| 4789s | basically that it's |
| 4792s | impossible but uh we did it anyway uh we |
| 4795s | persevered through and uh we |
| 4798s | multi-threaded the unreal replication |
| 4801s | Tech and uh well looking at the |
| 4805s | results they did not disappoint uh we |
| 4809s | actually gained and there's a graph |
| 4811s | showing it on the next slide but we |
| 4815s | gained a |
| 4817s | 2.5 times Improvement 2.4 times to be |
| 4821s | more exact in a replication Tex huge |
| 4825s | yeah it was a really really big |
| 4827s | Improvement in uh a replication and it's |
| 4831s | probably necessary to achieve the player |
| 4834s | counts that we're targeting for I don't |
| 4836s | I don't think it would be possible |
| 4837s | without doing |
| 4838s | this uh and the the really important |
| 4842s | thing to note about this optimization |
| 4845s | and why it's like such a big deal |
| 4847s | compared to like the five times |
| 4849s | Improvement we saw before from dorcy is |
| 4853s | that this Improvement |
| 4857s | helps equally even on the worst case |
| 4860s | scenario the other ones in the worst |
| 4862s | case scenario they're helpful but |
| 4864s | they're not that helpful |
| 4866s | this is incredibly important even on the |
| 4869s | worst case |
| 4871s | scenario and if we add it on uh all the |
| 4874s | other improvements we've done so far |
| 4876s | we've actually achieved a total |
| 4879s | Improvement of total reduction or |
| 4882s | replication take by 94% so are only at |
| 4885s | 6% of the original cost for uh |
| 4890s | replication uh this was actually a huge |
| 4892s | win for uh our team um but um |
| 4898s | it there are still there's still many |
| 4900s | more things that need to be |
| 4902s | done uh ases of creation is |
| 4906s | um an incredibly ambitious MMO in terms |
| 4909s | of scale uh and uh we need to do |
| 4915s | a lot of work in order to heat our goals |
| 4918s | and you know this is this is obviously |
| 4921s | again going back to the start of this |
| 4922s | whole conversation what is ashes of |
| 4924s | creation trying to be and what does it |
| 4927s | then require of the Tex act we are using |
| 4931s | in order to become that and when we said |
| 4934s | in 2017 that we are looking to bring |
| 4936s | massive back to massively |
| 4939s | multiplayer having an engine with the |
| 4941s | graphical Fidelity the front end that |
| 4943s | unreal provides |
| 4945s | is |
| 4947s | so is so important to the immersive |
| 4951s | nature to the beauty of the world to you |
| 4954s | know the quality of the setting but then |
| 4957s | coupled with a proprietary backend that |
| 4962s | can facilitate the same player |
| 4964s | experience that a massively multiplayer |
| 4967s | World wants to |
| 4969s | provide that's the dream that is what |
| 4972s | ashes is trying to be exactly |
| 4977s | phenomenal work here guys this is this |
| 4979s | is huge yeah I I I'd like to like take a |
| 4983s | moment I guess to shout out the team the |
| 4985s | team has done an incredible job you know |
| 4987s | we |
| 4989s | are we we we are we are pushing MMO Tech |
| 4991s | to where it's never gone uh and we |
| 4993s | continue to we we intend to keep doing |
| 4996s | this moving forward Hunter Alex Kevin |
| 4999s | Andrew Anton Adam ish Doug you know they |
| 5002s | they've all done amazing work uh you |
| 5005s | know to get us to this point where we're |
| 5006s | at today and I'm sure they'll keep doing |
| 5008s | it moving forward um if you're |
| 5011s | interested in working on problems like |
| 5013s | this you know yes send us an application |
| 5016s | you know we're always looking for people |
| 5017s | who want to work on this who have |
| 5020s | experience working on te Tech like this |
| 5022s | please send this an application we we're |
| 5024s | we're always interested and we know that |
| 5026s | like this this this showcase this this |
| 5028s | Tech talk this Ted Talk this this GDC |
| 5031s | talk you know what whatever whatever we |
| 5033s | want to apply to this this uh to this |
| 5036s | indepth technical discussion is going to |
| 5039s | appeal not just to to people who are |
| 5042s | aspiring obviously uh within this field |
| 5044s | to learn it to do it to you know this is |
| 5047s | their career going forward but also to |
| 5049s | those of you who have been within this |
| 5052s | field work and who yearn to work on the |
| 5055s | types of problems that we're discussing |
| 5057s | here in this showcase um we have a |
| 5060s | phenomenal culture at Intrepid we have a |
| 5062s | phenomenal project in our hands that is |
| 5065s | you know no shortage of problems that |
| 5068s | need to be solved um and one thing that |
| 5071s | our community has done a phenomenal job |
| 5073s | with is is helping to evangelize the |
| 5077s | project and the company and we're here |
| 5079s | to say that we have apps open we have |
| 5082s | positions that are available and if you |
| 5084s | want to be a part of this team um or if |
| 5087s | you want to help others find this team |
| 5089s | so that they can apply please share this |
| 5092s | link share this uh video with s go on |
| 5096s | you know social media and let people |
| 5097s | know we're working on some pretty Next |
| 5099s | Generation Tech that is going to change |
| 5102s | the face of massively multiplayer when |
| 5105s | it comes to a genre so you know very |
| 5107s | excited and proud of what your team has |
| 5109s | accomplished you guys I know are working |
| 5111s | every single day and doing a phenomenal |
| 5113s | job um congratulations just very well |
| 5117s | done thank you thank you and it's great |
| 5120s | to experience it in game two and I know |
| 5122s | that you know our PI folks have already |
| 5124s | experienced that in our node Wars test |
| 5126s | as well so maybe they didn't even |
| 5128s | realize it um yeah they might not have |
| 5131s | noticed um but there were definitely |
| 5133s | server workers working hard in that um |
| 5136s | and we were capturing data so we super |
| 5138s | duper appreciate it and as we head into |
| 5140s | Alpha 2 we're only going to get more |
| 5142s | data and be able to continue to push on |
| 5145s | and as uh the folks here have stated um |
| 5148s | moving towards more of that Dynamic |
| 5150s | State yeah um it's been in progress It's |
| 5153s | going to be in progress still for many |
| 5155s | years as we continually find new ways to |
| 5157s | advance it and make it better um we're |
| 5160s | committed to providing kind of like that |
| 5162s | quality we want for our players and as a |
| 5164s | lot of us have played MMOs in the past |
| 5166s | we just like networking was the big |
| 5170s | stopper for a lot of things and we're |
| 5172s | trying to do something different and |
| 5174s | trying to make that a better experience |
| 5177s | for not only our game but many more |
| 5179s | games to come if other companies were |
| 5181s | able to like wanting to adapt to a new |
| 5183s | tech style or try to figure out their |
| 5185s | their networking |
| 5187s | issues yeah and of course all of us here |
| 5189s | are huge MMO fans and I know those of |
| 5192s | you who are watching are as well |
| 5193s | obviously if you're sitting through this |
| 5195s | um but of course the the plan for us is |
| 5198s | to continue to create an amazing game |
| 5201s | and push forward and continue to change |
| 5203s | things even after post launch we'll |
| 5204s | continue to try to drive things forward |
| 5207s | so thank you for joining us and thank |
| 5208s | you to Hunter and Zach and Anton for |
| 5211s | your time and putting together an |
| 5212s | amazing deck a little extra shout out to |
| 5214s | Mike for putting together all the graphs |
| 5216s | gra Mike did such a good |
| 5219s | job Kyle for helping capture some of the |
| 5222s | little footage pieces to make sure that |
| 5223s | they looked great we super appreciate it |
| 5225s | any final words Stephen to WP yeah |
| 5227s | absolutely just you know out in the |
| 5229s | community let us know you know these are |
| 5231s | always the more risky takes that we |
| 5233s | provide insights to in our monthly |
| 5235s | updates because they are so Tech heavy |
| 5237s | and you know we know that this has an |
| 5240s | appeal to perhaps a smaller audience but |
| 5243s | give us your feedback give us your |
| 5245s | thoughts what did you think of this |
| 5246s | presentation how do you enjoy you know |
| 5248s | learning a little bit more indepth under |
| 5251s | the hood conversation about what it |
| 5253s | takes to make these types of games to |
| 5256s | make you know ashes of creation that |
| 5259s | that dream of all of ours um how does it |
| 5262s | appeal to you what are your questions |
| 5263s | what are your thoughts um leave us |
| 5265s | comments in the YouTube channel in the |
| 5268s | uh in on Twitter and different social |
| 5269s | medias and Discord and on our forums |
| 5272s | obviously the community team continues |
| 5273s | to do a bang up job of collecting your |
| 5275s | insights and bringing it back to the |
| 5276s | team um but we enjoy you know sharing |
| 5279s | this and and giving giving our perhaps |
| 5283s | um perhaps more sheltered uh |
| 5286s | developers who are who are in the behind |
| 5289s | the scenes making all the magic and |
| 5291s | everything work an opportunity to come |
| 5293s | out and talk about how they are |
| 5295s | literally the magic that makes the |
| 5297s | wheels turn for this bus um and it it's |
| 5300s | it's a great opportunity I think for |
| 5301s | them to come out see a little daylight |
| 5303s | and let and bask the of what is the |
| 5305s | glory that our community looks to them |
| 5307s | and says thank you for providing an |
| 5309s | opportunity to enjoy a game at this |
| 5311s | scale because it would not be possible |
| 5313s | obviously without your guys's hard work |
| 5316s | um and so with that I think we'll see |
| 5318s | you guys back on stream we'll see you on |
| 5321s | the flip |
| 5329s | [Music] |
| 5332s | side welcome back everyone we'll play |
| 5334s | the video in the background so you have |
| 5336s | something visually to watch while we |
| 5339s | discuss some things but of course uh we |
| 5341s | do we do grab a few questions from the |
| 5343s | chat uh most of them I feel like a lot |
| 5345s | of your questions we answered throughout |
| 5347s | later in the deck um but it seems like |
| 5350s | you guys like the presentation so again |
| 5352s | Mike thank you for putting that together |
| 5353s | and for Hunter Anton and Zach for |
| 5356s | putting together such an amazing very |
| 5358s | thorough uh explanation of the things |
| 5360s | that we're working on technology- wise |
| 5362s | but with that comes some information |
| 5365s | that we want to share with you guys of |
| 5367s | course we always like to be transparent |
| 5368s | we like to give you guys a heads up on |
| 5370s | what's coming and what we're doing and |
| 5372s | one of the things is that with this |
| 5373s | technology you know we had stated |
| 5375s | previously that we weren't going to |
| 5378s | expand we closed down our Alpha 2 packs |
| 5380s | for a reason and that reason was we |
| 5382s | wanted to make sure that we could |
| 5383s | service all the users that we had um one |
| 5385s | of the things that we really want to |
| 5386s | make sure of is that no matter who we |
| 5388s | have joining us that they're going to |
| 5390s | have a good experience obviously Alphas |
| 5394s | an alpha things are going to be buggy |
| 5396s | this is not a game we've always |
| 5397s | continually said that it's going to be a |
| 5399s | testing phase but with that we still |
| 5401s | want it to be a decent experience for |
| 5403s | those of you who join us and because of |
| 5406s | this amazing technology that these guys |
| 5408s | have been working on we can now service |
| 5410s | more users which means that we will be |
| 5413s | opening up Alpha 2 key access sales in |
| 5416s | the near future as we get closer to |
| 5418s | alpat 2 or with along with Alpha 2 so |
| 5421s | kind of stay tuned for that we do want |
| 5422s | to make sure that folks are aware of |
| 5424s | that but some caveats um these will not |
| 5427s | be pre-order packs these will be key |
| 5429s | access so these will be you know to |
| 5431s | access the product once it's available |
| 5433s | and then of course the pricing is still |
| 5434s | TBD but we will we're going to try to |
| 5437s | make sure it's fair as possible |
| 5439s | especially for those of you who have |
| 5440s | already purchase packages we understand |
| 5442s | that you're like hey I purchased a pack |
| 5444s | package we want to make sure that that |
| 5445s | price is comparable so we are Stephen |
| 5448s | and I have gone back and forth and we |
| 5450s | we'll make sure that it is fair for you |
| 5452s | guys and they will not be um when we do |
| 5455s | release the access to |
| 5457s | to they will not be um additional |
| 5460s | rotating Cosmetics either um so these |
| 5463s | will be static key access that players |
| 5465s | will have an opportunity to purchase in |
| 5467s | and access Alpha 2 and participate |
| 5469s | because ultimately Alpha 2's purpose is |
| 5471s | to test the game and more players |
| 5474s | testing the game is better um we wanted |
| 5476s | as Margaret said be sure that the tech |
| 5478s | that we are utilizing for Alpha 2 uh is |
| 5480s | capable of facilitating that um that |
| 5483s | Dynamic server griding and by the way |
| 5484s | watching the watching the commentary |
| 5486s | watching the feedback during the uh |
| 5488s | during the discussion was interesting |
| 5491s | there's a lot of people who I I didn't |
| 5493s | expect we're going to be really |
| 5494s | interested in the |
| 5496s | conversation um and it seemed like you |
| 5498s | know we I was watching the viewership |
| 5500s | was staying still around you know what |
| 5502s | it was I was like okay this is better |
| 5505s | than I expected we knew not everybody's |
| 5507s | G to be interested and we I we try to be |
| 5509s | uh set expectations like we've made it |
| 5512s | very clear that this is what we're going |
| 5513s | to be talking about for like a month now |
| 5516s | so hopefully everyone was prepared we |
| 5518s | tried to like make sure our social copy |
| 5519s | was like hey it's going to get nerdy |
| 5521s | we're going to talk a lot of tech so you |
| 5523s | know that's not your thing that's okay |
| 5525s | it definitely did we could have got more |
| 5526s | nerdy I feel like um but also I wanted |
| 5529s | to note that the upgrade packs uh for |
| 5532s | the previously purchased packs are still |
| 5534s | going to be available so if you already |
| 5535s | purchased a pre-order pack those |
| 5537s | upgrades are still available to you as |
| 5539s | well so those aren't going away um at |
| 5542s | some point obviously Alpha 2 keys will |
| 5544s | go away and of course those upgrade |
| 5546s | packs will go away um but those will be |
| 5549s | closer to the end of Alpha 2 and Alpha 2 |
| 5551s | is meant for to be a much longer um |
| 5554s | running um testing phase for us and |
| 5557s | we'll be we'll be having updates that |
| 5559s | will be coming frequently for those bug |
| 5561s | fixes that will be coming frequently for |
| 5563s | those so um in the persistence wise of |
| 5566s | aspect of it um because I think that |
| 5568s | that is something that needs to be set |
| 5570s | because people have different |
| 5571s | understanding of what persistance means |
| 5572s | we will have times where this servers do |
| 5574s | go down so that we can do updates so |
| 5576s | that we can check things and make sure |
| 5578s | things are good our goal is to not have |
| 5580s | to wipe but obviously if there are |
| 5582s | situations where we have to we will do |
| 5583s | that um but persistency is the main goal |
| 5586s | for us we want to get all that data that |
| 5587s | we possibly can when it comes to |
| 5590s | persistence absolutely that's not it we |
| 5593s | also have another announcement of an |
| 5595s | announcement another |
| 5597s | announc we I think people will like this |
| 5600s | one though that's yeah you think I think |
| 5602s | some people will like and some people |
| 5603s | won't uh so we are also going to start |
| 5607s | expanding our NDA testing um obviously |
| 5610s | we've been working with pi you guys have |
| 5611s | known that we've kept them pretty NDA we |
| 5614s | did uncover the vocal NDA in regards to |
| 5617s | the node War spot testing um but |
| 5619s | obviously we're doing other testing and |
| 5620s | by the way that's that's been a lot of |
| 5622s | fun to watch and I'm very encouraged to |
| 5624s | continue doing that as part of |
| 5626s | additional testing going forward because |
| 5628s | it's been great to see our test groups |
| 5631s | go out and sit on different content |
| 5633s | creators platforms and talk about what |
| 5634s | their experiences were and allow you |
| 5637s | know more than just us as an interface |
| 5639s | to you know what development is looking |
| 5642s | like right now what the game state is |
| 5643s | looking like right now so I'm very |
| 5645s | encouraged by those interactions I think |
| 5647s | we have a phenomenal content creator |
| 5649s | team group out there that is following |
| 5652s | the game who are you know servicing a |
| 5654s | lot of different viewers and communities |
| 5656s | who want to learn about what's happening |
| 5658s | they have great channels to go to and do |
| 5660s | that and you guys are choosing great |
| 5661s | guests and having really good |
| 5662s | conversations so kudos to you guys as a |
| 5665s | tangent because you brought that up um |
| 5668s | we are going to be putting up a role for |
| 5669s | a content creator relations manager and |
| 5672s | as soon as we fill that role you'll be |
| 5673s | starting to see some of those uh that |
| 5676s | program starting to come online so just |
| 5678s | you know that's very exciting for us we |
| 5680s | have so many applicants who have come in |
| 5682s | and one of the things that Stephen and I |
| 5684s | had discussed is that we really want our |
| 5686s | content creator program to be a very |
| 5688s | interactive thing you are part of like |
| 5689s | our family um so we want to help you you |
| 5692s | help us you know it's going to a mutual |
| 5694s | thing it's not going to be like a |
| 5695s | program that you've seen before and we |
| 5697s | did you know provide a little bit of an |
| 5699s | outline of what that program looks like |
| 5701s | and a lot of people were pretty excited |
| 5702s | about it but you know the bigger you |
| 5705s | grow the bigger we grow so we want to |
| 5707s | help you just as much as you help us and |
| 5709s | we aren't looking at just big content |
| 5711s | creators we're looking for really just |
| 5713s | quality content uh creators who are |
| 5715s | consistent so you could have 200 |
| 5717s | followers you could have 200,000 so it |
| 5720s | just to us that's not what we're looking |
| 5722s | for what we're looking at is we're |
| 5723s | really watching all your content we're |
| 5725s | looking at all your social yes exactly |
| 5728s | which is very timec consuming and we've |
| 5729s | already started funneling through some |
| 5731s | of that stuff and looking through it but |
| 5733s | obviously once we fill that role that's |
| 5734s | going to be the person who's going to |
| 5735s | help us manage it cuz it's going to be a |
| 5738s | full-time job um it might we might need |
| 5740s | more people than just one um but to go |
| 5742s | back to the announcement of the |
| 5743s | announcement um we obviously have a pi |
| 5747s | joining us and testing but we need more |
| 5750s | testers so we are going to be starting |
| 5751s | to expand that we're going to do a slow |
| 5753s | roll of what that expansion looks like |
| 5755s | and so for these initial testing we are |
| 5757s | going to be starting to add some of the |
| 5759s | alpha 1 users now how we're going to |
| 5761s | pick that I know that's going to be a |
| 5762s | big contentious thing but we uh are |
| 5765s | definitely going to do our best and so |
| 5767s | the first thing that we want to do is |
| 5768s | for this first round is we're going to |
| 5770s | add the top 500 from the alpha 1 |
| 5773s | contributors so that is looking at |
| 5775s | people who actually participated in game |
| 5777s | in alpha 1 we are going to be grabbing |
| 5779s | those users and then we're going to have |
| 5781s | another 500 that will be randomly chosen |
| 5784s | you guys will all be emailed the goal is |
| 5786s | to email you within this July time frame |
| 5788s | and uh you'll be joining us in testing |
| 5791s | like soon and and this is really this is |
| 5793s | you know this approach of invitation to |
| 5797s | um randomize but also activity like we |
| 5800s | know all the people who participated in |
| 5802s | alpha 1 what they did the statistics |
| 5804s | around how often they played what level |
| 5806s | they got to the you know amount of time |
| 5808s | that they played and the detail they |
| 5810s | spent in their Forum accounts and their |
| 5813s | notes and response like those are our |
| 5815s | dedicated testers from the alpha 1 group |
| 5818s | so we want to give them a first stab |
| 5821s | exactly we want them we want them back |
| 5822s | first stab as we begin to expand outside |
| 5826s | of the P1 Group excuse me I'm sorry |
| 5828s | outside of the uh um Pi group uh into |
| 5832s | alpha 1 testers over the month of July |
| 5834s | and then into August as we expand that |
| 5836s | to include many more Alpha One testers |
| 5840s | all all of the Alpha One testers |
| 5842s | eventually right uh before A2 so this is |
| 5845s | just how we start to ramp that up in in |
| 5847s | the first stages and that's just |
| 5849s | ensuring that you know we give you guys |
| 5850s | a decent experience as we move forward |
| 5852s | and that's not to say that uh you won't |
| 5854s | run into bugs not to say that we won't |
| 5856s | have some client performance issues or |
| 5858s | whatnot but like we are going to be |
| 5860s | testing a lot of bigger pieces and we're |
| 5862s | for a little longer test runs um |
| 5864s | obviously these will be all under NDA |
| 5866s | there may be cases where we unlock some |
| 5868s | of that vocal NDA like we have stated |
| 5870s | before we might unlock some images that |
| 5872s | people take um like screenshots um we |
| 5875s | let them screenshot but only within our |
| 5877s | like private uh channels um obviously |
| 5880s | not externally um but yeah so we are |
| 5883s | we're we're excited we're excited to |
| 5886s | exciting news very exciting and you know |
| 5888s | obviously I know that runs into folks |
| 5891s | asking other things about Alpha 2 but uh |
| 5894s | I will let uh Brian talk a little bit |
| 5896s | about that when he goes into Studio |
| 5897s | update I do have some questions that we |
| 5899s | do want to answer in regards to server |
| 5901s | meshing technology so I'm going to reel |
| 5903s | us back in but that was relevant to it |
| 5905s | I'll try my best I'm I am cracking my uh |
| 5909s | technical brain skills let's go I think |
| 5912s | Brian said earlier he was like I think I |
| 5914s | got smarter after watching that yeah I |
| 5918s | did I did all right so our first |
| 5920s | question here is what is your plan and |
| 5922s | what systems are in place to avoid the |
| 5924s | first few months of server cues |
| 5927s | congestion and when players drop off how |
| 5929s | do you plan on handling that if such |
| 5931s | when the time comes |
| 5934s | sure um you know that I I think that |
| 5937s | obviously is a problem that every MMO |
| 5939s | launch suffers right I mean that is you |
| 5941s | know the balance between your active |
| 5944s | users your expected Peak concurrencies |
| 5947s | right um where that fall-off begins and |
| 5950s | how you consolidate servers you don't |
| 5952s | want to expand too much at the beginning |
| 5954s | to alleviate server cues and then end up |
| 5956s | with like 30% POP servers that you have |
| 5959s | to consolidate later the same is going |
| 5961s | to be true for Alpha 2 right and so |
| 5964s | we're going to be leveraging obviously |
| 5965s | we know how many users we have available |
| 5969s | uh when it comes to Alpha 2 entrance and |
| 5972s | then as we uh as we allow additional |
| 5975s | people to purchase in with these new key |
| 5976s | sales that we'll be doing later um we're |
| 5979s | going to get an idea of what that is |
| 5980s | generally I would expect you can you |
| 5982s | know we're going to see 30 to 40% Peak |
| 5984s | concurrencies when it comes to like |
| 5985s | opening the servers when it comes to |
| 5987s | Prime Time when it comes to events that |
| 5989s | happen right and |
| 5991s | so when we talk about um server queuing |
| 5995s | and and and the and the server |
| 5997s | congestion we want to make sure that we |
| 5999s | have healthy Alpha 2 servers so we're |
| 6001s | going to leverage server queuing as much |
| 6003s | as possible that means that sometime |
| 6006s | those server cues might get into the |
| 6007s | thousands um during those peak times |
| 6010s | during the servers spinning up or during |
| 6012s | those Prime Times or events um and |
| 6015s | that's okay it's not going to be the |
| 6016s | most fun for a lot of players to have to |
| 6018s | sit in a queue obviously we've all done |
| 6020s | that before in MMOs that we played but |
| 6022s | the alternative is that we have too many |
| 6025s | servers available we don't have those |
| 6027s | cues during peak times and then when |
| 6029s | it's not peak times those server |
| 6031s | populations dilapidate and that's not |
| 6033s | what we want to have obviously um but |
| 6036s | we'll have live operations watching |
| 6037s | those making sure we're looking at the |
| 6040s | health of our our realms for this case |
| 6042s | we'll say Realms because servers mean a |
| 6044s | different thing and it can get confusing |
| 6046s | um so we'll make sure that we have |
| 6047s | enough Realms to to accommodate for the |
| 6050s | users that we expect corre and and we |
| 6052s | are able to spin up new Realms at a |
| 6054s | moment's notice right so so as we |
| 6059s | watch at a devops no |
| 6063s | moment yeah I just want I I want to say |
| 6066s | that because like when we said it before |
| 6068s | the guys were like uh they're like we |
| 6070s | can do it quickly but not like well and |
| 6073s | I mean generally you have a new new |
| 6076s | realm on standby right that's spooled up |
| 6078s | and ready to go and you need to pull the |
| 6079s | trigger on it or not we just don't |
| 6081s | unlock the tags for you guys to be able |
| 6083s | to get in so we have those those ready |
| 6085s | we just don't unlock them for you but |
| 6087s | it's easy for us to do all right then |
| 6090s | the next one here is is there a higher |
| 6093s | likelihood for bugs or server hiccups |
| 6095s | near the edges of servers and during |
| 6097s | testing of this meshing have there been |
| 6100s | issues surrounding item duplication |
| 6102s | server you know hiccups back and forth |
| 6104s | between travel Ser on these server |
| 6106s | borders so like a lot of people had |
| 6108s | questions regarding this because sure |
| 6110s | people are worried about speed hackers |
| 6111s | and stuff like that yeah absolutely I |
| 6114s | mean around these borders and especially |
| 6116s | early on as we as we um rolled out this |
| 6120s | multi-grid formation we're looking for |
| 6122s | bugs when it comes to server Authority |
| 6125s | transfers when it comes to these proxies |
| 6127s | across boundaries you know all those |
| 6128s | types of things however um these days |
| 6131s | not so much right because we've been |
| 6133s | spending the last nine 10 months or so |
| 6136s | the networking team around identifying |
| 6138s | those bugs across border issues we |
| 6140s | started with a one by one and then we |
| 6142s | did a 2 by two 3 so' we've been scaling |
| 6146s | it up and testing and yes we we found |
| 6149s | bugs before that's right but um what you |
| 6152s | say about you know as as we kind of |
| 6154s | enter into Alpha 2 and as we continue to |
| 6157s | expand our testing to include now |
| 6159s | additional real clients um that is where |
| 6163s | the networking team obviously focuses on |
| 6165s | they watch those things they get reports |
| 6167s | and profiles around where those errors |
| 6170s | can occur and they address them as |
| 6172s | necessary |
| 6175s | all right uh the next one is does in |
| 6177s | trepid currently have a plan on how to |
| 6180s | accomplish dynamic griding or is it just |
| 6183s | we want it but don't know how um and oh |
| 6186s | no we talked about it so so right right |
| 6188s | right now that right now Dynamic griding |
| 6190s | is in place right now what we are doing |
| 6192s | is we are optimizing fixing performance |
| 6196s | around that and the the current goal |
| 6199s | which is on target is that Dynamic |
| 6201s | griding will be online by the start of |
| 6203s | Alpha 2 |
| 6204s | um so you know all of our Tech was built |
| 6208s | to get to the point where we can do |
| 6211s | Dynamic griding um this has always been |
| 6214s | the end goal overall so Dynamic griding |
| 6216s | is still in active development right now |
| 6218s | but um the plan is that by Alpha 2 |
| 6221s | entrance criteria we will have dynamic |
| 6223s | griding in a bug sturdy place that is |
| 6226s | optimized and and uh working as we need |
| 6229s | it right and how does the longevity of |
| 6233s | servers speak to the relevancy problem |
| 6236s | um it is eventually going to be |
| 6239s | non-financially viable in this person's |
| 6241s | opinion to maintain that amount of |
| 6242s | server Farm space to keep the |
| 6245s | performance optimal I had to put that in |
| 6246s | there because I just yeah yeah I mean |
| 6249s | you know when you talk about um when you |
| 6251s | talk about infrastructure right there |
| 6254s | there's a duality in modern day times |
| 6256s | between what's available in the cloud |
| 6258s | and there's you know business |
| 6260s | relationships that get formed between |
| 6262s | Cloud providers like an AWS or gcp uh |
| 6266s | where you can have you know reserved |
| 6268s | instances available you want to optimize |
| 6271s | you know your server types and machine |
| 6273s | configurations that are necessary so |
| 6275s | that you know costs are not prohibitive |
| 6277s | against the number of expected users per |
| 6279s | instance that it can facilitate um and |
| 6283s | we run all of those projections we have |
| 6285s | that regularly that we are optimizing |
| 6287s | also the types of machine uses that we |
| 6289s | require and that's from a cloud |
| 6291s | perspective now when it comes to testing |
| 6294s | especially for Alpha 2 will be heavily |
| 6296s | leveraging Cloud instances in order to |
| 6298s | provide the types of of of of |
| 6300s | infrastructure that we require because |
| 6303s | it's easily scalable and we can Flex |
| 6304s | into that relatively um uh in an on |
| 6307s | demand fashion however as we move |
| 6310s | towards a proper launch scenario we will |
| 6314s | have bare metal that we service |
| 6316s | ourselves from a service center |
| 6318s | perspective uh that we U utilize and |
| 6321s | still have the ability to flex into the |
| 6323s | cloud when we need to flex uh for |
| 6326s | numbers that we didn't account for for |
| 6328s | popularity we didn't account for that's |
| 6330s | the that is the finess of these types of |
| 6332s | systems and how we are agnostic of |
| 6334s | whether or not we deploy our clusters in |
| 6337s | a bare metal service or in the cloud |
| 6340s | where we can do either or um so so for |
| 6343s | us you know everything we do to |
| 6345s | calculate cost per user is associated |
| 6348s | with how we're optimizing for the types |
| 6350s | of instances that we can use and for |
| 6352s | Alpha 2 that's looking pretty good yeah |
| 6355s | so we're not concerned on the financial |
| 6357s | side of things I think everyone always |
| 6358s | asks those those type of questions but I |
| 6360s | figure Mak they're great questions L |
| 6363s | that's a good question about that so I'm |
| 6365s | like I'll add it to there we talk about |
| 6367s | this stuff regularly all the time I know |
| 6370s | Brian and I Brian and I are reviewing |
| 6372s | projections and spreadsheets that he's |
| 6374s | putting together every day almost |
| 6377s | absolutely it's the stuff we have to do |
| 6379s | I mean I think similarly to the tech |
| 6382s | side of things that the financial side |
| 6384s | of stuff people you know it's not the |
| 6386s | stuff that you get to see a lot of right |
| 6388s | so yeah um and we've been pretty |
| 6390s | transparent about where we're at what |
| 6392s | we're doing we're how we're moving |
| 6393s | forward in regards to that now now don't |
| 6395s | get me wrong let me give some context |
| 6397s | here because I think it's interesting |
| 6398s | for those who in the audience who enjoy |
| 6401s | more the business side of things um |
| 6403s | there have been times in the past at our |
| 6406s | company and other companies do this |
| 6408s | where running a test is exorbitantly |
| 6411s | cost too cost right however you need to |
| 6415s | bite you need to take that bite in order |
| 6418s | to collect certain information even if |
| 6419s | you're not in a state where you're fully |
| 6421s | optimized in order for that to be |
| 6423s | long-term sustainable we've done that in |
| 6425s | the past and and we knew the price we |
| 6427s | were paying in order to do that Alpha 2 |
| 6429s | is not one of those scenarios so just to |
| 6432s | be clear all right and of course that |
| 6435s | was our server mhing technology we hope |
| 6437s | that folks enjoyed it the video for that |
| 6439s | will be uploaded after the end of this |
| 6441s | live stream so you can go check it out |
| 6442s | relive it go back to where you want uh |
| 6444s | read read things look at stuff cuz I |
| 6447s | know there were people who were peeking |
| 6448s | at some assets and stuff like that that |
| 6450s | they were excited about and when we were |
| 6452s | showcasing 3D models in there um but of |
| 6454s | course we do have a feedback thread up |
| 6455s | over on our forums um so if you do have |
| 6458s | any thoughts or questions or um you know |
| 6460s | any any answers to some of the questions |
| 6463s | that Stephen had asked both in there and |
| 6466s | then previous before we uh started the |
| 6468s | the segment uh feel free to head on over |
| 6471s | there and give us your thoughts we are |
| 6472s | constantly reading those putting reports |
| 6474s | together for our Dev team um and if we |
| 6476s | see any additional Fe questions that we |
| 6478s | feel need to be answered we'll make sure |
| 6480s | that we get those answered in future |
| 6481s | updates um and with that I'm going to |
| 6484s | hand it actually over to lenford to do |
| 6486s | our studio update well thank you yeah I |
| 6489s | do I did I did feel smarter after |
| 6491s | watching that incredible and Stephen |
| 6492s | you're right I think I might go try to |
| 6493s | get my engineering degree now there you |
| 6495s | go that's what I'm saying I I'm just I'm |
| 6497s | curious how many people you know are |
| 6500s | interested in that don't have the |
| 6502s | opportunity to to associate thing are |
| 6505s | you interested post a one in chat um if |
| 6508s | you're live on Twitch and if you're |
| 6510s | watching this on YouTube you can put a |
| 6511s | little one emoji in your comment but you |
| 6514s | know you know it's it's one thing to to |
| 6516s | listen to a lecture to take a class or |
| 6519s | to you know GDC talk um where the |
| 6522s | substance of the conversation is related |
| 6525s | to you know |
| 6526s | particular knowledge you want to gain |
| 6529s | it's another thing to have the |
| 6531s | substance also the interest of the |
| 6534s | project coupled right like the it's it's |
| 6537s | great to be I think a fan who's |
| 6539s | interested in the tech side of this or |
| 6540s | who's a software developer or who's you |
| 6542s | know working in unreal and trying to |
| 6544s | build their first game or whatever |
| 6546s | following a project they care about from |
| 6548s | a gamer's perspective and now get to |
| 6549s | couple that with their interest uh from |
| 6551s | a a technology perspective you're |
| 6553s | looking for work we're hiring just |
| 6556s | uh you know go |
| 6559s | ahead things you got cover yeah I I do |
| 6562s | have list things I'd like to cover with |
| 6564s | the hirings on that um so uh yeah so |
| 6567s | look so what they said in the in the in |
| 6569s | the video today was that uh the stuff is |
| 6571s | going to and as Stephen recapped the |
| 6572s | stuff's going to be available coming to |
| 6574s | Alpha 2 for the dynamic uh griding stuff |
| 6578s | uh and I think the first thing to note |
| 6579s | is that the studio in general the the |
| 6582s | fires of production are hot uh the |
| 6585s | there's the yeah there's H the plumes of |
| 6587s | smoke arising from the forges uh and we |
| 6589s | are definitely speeding towards Alpha 2 |
| 6593s | uh so just to reiterate I think off what |
| 6596s | we said last uh live stream we are still |
| 6598s | entirely confident in our plan for a Q3 |
| 6602s | Alpha 2 that's right um so and we're |
| 6605s | planning to announce a date yeah that's |
| 6608s | what I was gonna say we're not quite |
| 6610s | ready to put the date out there yet but |
| 6612s | I think by next live stream I think by |
| 6615s | next live stream I'll have enough Road |
| 6617s | mapping done with the team and we'll |
| 6618s | know for sure yeah the end of this month |
| 6621s | expect to expect to get a date for A2 uh |
| 6624s | start and and actually you know from a |
| 6626s | production standpoint uh many people you |
| 6629s | know obviously don't know we've talked |
| 6630s | in the past about how production runs |
| 6632s | but we just finished a major Milestone |
| 6634s | yes we did we talked about that a little |
| 6636s | last time too Milestone |
| 6638s | 9ine yep yeah give some some stats on it |
| 6643s | yeah yeah great okay cool um yeah so uh |
| 6646s | we did we just finished Milestone 9 uh |
| 6648s | so as a company I don't know uh how |
| 6650s | familiar the audience is with our scrum |
| 6652s | uh stuff and what story points really |
| 6654s | means but we completed 8,900 story |
| 6656s | points that's like a measure of time and |
| 6658s | risk um and that's quite a vast number |
| 6662s | to complete in a milestone uh we were |
| 6664s | about 89% successful to our own |
| 6667s | estimations which is pretty darn good um |
| 6669s | but of course anything that can you know |
| 6671s | can get in the way does and we have |
| 6672s | meetings every day to look at disruptors |
| 6675s | and top five criticals and blockers and |
| 6677s | stuff like that so and just to give |
| 6679s | context you know 89% delivery on the |
| 6682s | product goals for a particular Milestone |
| 6684s | given that a milestone is looking out |
| 6686s | roughly 3 months ahead is pretty darn |
| 6689s | good right and and people might wonder |
| 6691s | well why wasn't it 100% right the things |
| 6694s | that can happen during a milestone |
| 6696s | whether it be you know emergent work |
| 6698s | that was unexpected that might be SE |
| 6701s | zero work for you know build Breakers or |
| 6704s | test blockers memory crash or something |
| 6706s | memory Stomps or crashes right discovery |
| 6709s | of new work or estimations didn't take |
| 6711s | into account like unknown within a |
| 6713s | particular piece of code like all of |
| 6715s | those can come in to kind of disrupt |
| 6718s | what an end goal is for for delivery on |
| 6722s | a milestone and 89% is pretty up |
| 6725s | there yeah yeah I was I was really quite |
| 6727s | proud of the team it was really good |
| 6729s | absolutely uh yeah and bugs bugs bugs |
| 6731s | bugs bugs right squash them um so yeah |
| 6734s | uh so that was uh kind of how far we got |
| 6737s | along in Milestone 9 um we're about to |
| 6739s | kick off our Milestone 10 here techn |
| 6743s | yeah we have we have a town hall in |
| 6744s | about two hours an hour and a half |
| 6747s | that's right uh so we're you know I |
| 6749s | think we've already identified some of |
| 6750s | this stuff too but we're you know we're |
| 6751s | going to continue looking at polishing |
| 6753s | for our V uh VFX and our UI elements uh |
| 6756s | we're going to continue looking at the |
| 6757s | golden path which we play test regularly |
| 6760s | or I think as last month you called it |
| 6762s | Steven the golden on fire Lavar ridden |
| 6764s | tree yeah the that's right H but we're |
| 6768s | uh continuing to play test uh on a |
| 6770s | regular basis with that stuff trying to |
| 6772s | make sure that uh it does have a good |
| 6774s | experience for the player we are really |
| 6776s | player driven uh in what we're trying to |
| 6777s | accomplish even though I know I know |
| 6779s | it's not it's not a game I get it yeah |
| 6782s | um but uh but it is something that we |
| 6783s | were trying very hard to keep uh keep |
| 6786s | going so uh yeah go ahead I was just |
| 6788s | goingon say another huge thing which I |
| 6790s | know you're you're probably about to get |
| 6791s | to is we had a |
| 6794s | 15% Staffing growth over the last |
| 6797s | quarter excited he's taking all of |
| 6801s | Brian's I am sorry |
| 6804s | I'm just excited I think you know I am |
| 6806s | too I think we've hired more this so far |
| 6809s | this year than we did in all of last |
| 6811s | like it's it's been an incredible growth |
| 6813s | Journey for this company um yeah I was |
| 6815s | going to mention a couple other things |
| 6816s | we're we're we're hammering on economy |
| 6818s | we've got our features and gameplay |
| 6820s | stuff are looking great we got more to |
| 6821s | go we're working on material components |
| 6824s | um and obviously we're fixing bugs in |
| 6826s | advance of the Sala two stuff so I think |
| 6827s | we're W we're wrapping it up pretty good |
| 6829s | and getting ready for uh letting you |
| 6831s | guys know what that date is mhm all |
| 6834s | right well with that I'm going to swap |
| 6837s | us over to our art |
| 6838s | update this is a long stream so thank |
| 6841s | you all for so much for for sticking |
| 6843s | with us had two long back toback streams |
| 6845s | I know I know so hopefully um you know |
| 6849s | you took your body breaks and this and |
| 6852s | this month we're also I think people are |
| 6853s | going to be really excited we can we can |
| 6855s | leak perhaps what's coming up this month |
| 6857s | because it's in stone isn't |
| 6859s | it Margaret just gave me the side you |
| 6863s | mean like what the content is that we're |
| 6864s | going to we've already talked about it |
| 6866s | but go good okay yeah we're this month |
| 6869s | you guys are going to be seeing for the |
| 6871s | first time at the end of July the barred |
| 6874s | archetype and it is thought talking |
| 6876s | about something else oh shoot my bad no |
| 6879s | no Bard is yeah that b is up and coming |
| 6883s | I did tease that a little bit I did |
| 6884s | tease that yeah and I talked about a |
| 6886s | little earlier in the Stream too but yes |
| 6888s | what maybe we can get Steph to play a |
| 6890s | guitar for us on stream I'm down get a |
| 6893s | loot loot ready all right we'll just go |
| 6898s | through some art here really quickly um |
| 6901s | we've got these amazing silken web hand |
| 6903s | wraps that were made by Chris and a |
| 6907s | studded leather belt you know the or St |
| 6910s | studded silver belt sorry not leather |
| 6912s | it's the upgrade from the the um leather |
| 6915s | belt um this is still a work in progress |
| 6918s | I believe but you kind of get to see |
| 6919s | some of it and then this is the grow |
| 6923s | guard |
| 6924s | helmet love that yeah it's looking cool |
| 6927s | it's |
| 6928s | creepy and then we have The Robe with |
| 6930s | the |
| 6931s | financier financier financier I don't |
| 6934s | know if you're financier |
| 6937s | finer very rich yes they got the gold in |
| 6942s | there Blues love loving the folds and |
| 6945s | the cloth looking good yeah those were |
| 6947s | by Danny the helmet and the robe and |
| 6949s | then jiny worked on |
| 6953s | this cloak of ill |
| 6955s | Omens Jacob our lead producer was saying |
| 6958s | he wants a version of this with the hel |
| 6961s | with the hood up the hood up |
| 6963s | yeah but it's very that's cool and then |
| 6967s | we've got this awesome |
| 6969s | bat I love it it's like a bullfrog and a |
| 6972s | bat together I'm so into it so cute it |
| 6976s | is cool I like it's Little Neck a big |
| 6978s | neck it's not a Little Neck um and then |
| 6981s | the last one here is the rosarium |
| 6985s | guard I think I have a a turntable for |
| 6988s | this one as well that looks |
| 6992s | uncomfortable it looks |
| 6994s | uncomfortable yeah I've been watching uh |
| 6996s | I I was watching House of dragons and um |
| 6999s | they had some really good armor in this |
| 7001s | last episode and it moved really well |
| 7004s | with the shoulders but I'm looking at |
| 7005s | this thing and I'm like oh I don't know |
| 7006s | if I can move that well that it's got |
| 7008s | the built-in beard scratcher right here |
| 7010s | I know the beard scratcher armor chafing |
| 7012s | is awful as a cosplayer I have made so |
| 7015s | many pieces of armor one of the things |
| 7016s | that I've learned is like you always |
| 7018s | want to put like fur underneath there or |
| 7020s | cushions because like anywhere where |
| 7022s | you're like moving your joints it's like |
| 7025s | really awful yeah those edges yeah |
| 7028s | leather and like if you're actually |
| 7029s | wearing like heavy armor it's um you |
| 7032s | know if you're make it out of foam and |
| 7033s | stuff it's not as bad and that is our |
| 7036s | art update so beautiful into the last |
| 7039s | segment here and I'll just play some of |
| 7041s | the slides in the back ground so you |
| 7043s | have something visual to look at whilst |
| 7045s | we uh go through our Q&A these are the |
| 7047s | questions from the forums and the first |
| 7049s | one here is from Phantom X and they |
| 7052s | would like to know about war commissions |
| 7054s | I thought it was fantmx |
| 7056s | Phantom I don't know phant I'm just |
| 7059s | kidding I called him I called him Fant |
| 7061s | MX one time and and he didn't hear the |
| 7063s | end of it okay sorry go ahead uh vnar |
| 7066s | put it in quotes Phantom X to make it |
| 7068s | clear to make sure okay but Phantom you |
| 7071s | can let us know you prefer it or maybe |
| 7074s | St definely I'm just trolling him maybe |
| 7078s | Steven will keep trolling you uh the |
| 7081s | this question is about war commissions |
| 7083s | we know that non-citizens can complete |
| 7085s | mayoral commissions but does that |
| 7087s | include those that will start a node war |
| 7090s | in other words could non-citizens do the |
| 7093s | commission required to start a war even |
| 7095s | if citizens Del liberally don't complete |
| 7097s | them that's something that um I think we |
| 7100s | are still leaving undetermined on a |
| 7103s | design side it can go either way it's a |
| 7104s | flag that a designer can set that either |
| 7106s | requires it to be citizen only or |
| 7108s | non-citizen as well it's something we're |
| 7110s | going to test my inclination is that for |
| 7112s | certain types of node actions that Grant |
| 7115s | the mayor certain node powers that we |
| 7117s | will expect only citizen participation |
| 7120s | at a certain threshold but it's |
| 7121s | something that obviously we're going to |
| 7123s | be testing in A2 and then casty wants to |
| 7126s | know about threat generation of taunts |
| 7128s | heals and Damage Done each of equal |
| 7131s | output perform formed against enemy mobs |
| 7134s | will there will they all generate a one |
| 7136s | to one to one ratio of threat generation |
| 7139s | or will one of those three be slightly |
| 7141s | better or worse than the other two at |
| 7143s | producing hate right now the formula |
| 7146s | includes modifiers so we can apply a |
| 7148s | modifier as we wish to either healing |
| 7151s | values versus uh uh um uh hate values um |
| 7156s | those things are going to just kind of |
| 7158s | be balanced as we kind of see play |
| 7160s | testing continue uh to see you know |
| 7162s | what's the right blend between the tank |
| 7165s | holding aggro the Healer not pulling it |
| 7167s | too often and the DPS not grabbing it as |
| 7169s | well so we we want to have maximum |
| 7171s | flexibility in the formula and that's |
| 7173s | why modifiers exist there all right and |
| 7176s | then kadak wants to know about Open Sea |
| 7178s | content will there be points of interest |
| 7180s | such as underwater dungeons that will |
| 7183s | also remain outside of the flagging |
| 7184s | system in the open sea or will those be |
| 7188s | contested that's a very good question um |
| 7192s | right now the expectation should be that |
| 7194s | those will remain |
| 7195s | contested and and in the event where we |
| 7198s | do not want to have you know let's say |
| 7202s | certain content contested we would |
| 7205s | likely have some type of doorway or |
| 7209s | shared instance opening that players |
| 7211s | then can move into where it is an |
| 7212s | uncontested underwater |
| 7215s | bubble exactly uh chronics wants to know |
| 7218s | about UI and uxix tools what tools can |
| 7221s | we expect for shot callers regarding |
| 7223s | marks and pings this is a big one that |
| 7226s | we've gotten asked a lot about maximum |
| 7229s | tools because the people making this |
| 7231s | game want the exact same thing you do I |
| 7234s | think I brought this up in pult the |
| 7235s | other day I was like just I know we have |
| 7237s | a lot on our plate for ala 2 but could |
| 7238s | we just make like one thing where I can |
| 7240s | put an i on one person's head and they |
| 7242s | were like okay we'll try we'll see what |
| 7244s | we can do that is in their backlog for |
| 7247s | pre to entrance so it's still |
| 7249s | outstanding work I know they have a long |
| 7250s | list of them I was just like just give |
| 7251s | me one like now absolutely um our next |
| 7256s | comment here is from Top wombat which I |
| 7259s | love that name uh server they want to |
| 7261s | know about server Tech which is very |
| 7263s | fitting for today's stream will mobs |
| 7265s | respond faster if there are a lot of |
| 7266s | players in an |
| 7269s | area so yeah that's that's that's the |
| 7271s | spawner system uh the spawner system has |
| 7274s | a number of ways that it can respawn |
| 7276s | this is a it can be a set value that's |
| 7278s | time based it can be predicated on um on |
| 7281s | Monster count so as Monster count drops |
| 7283s | beneath a threshold the spawner will |
| 7285s | spawn additional Monsters the designers |
| 7287s | the world designers have all that |
| 7288s | flexibility at their disposal it just |
| 7290s | kind of depends on the area the quest |
| 7292s | type uh that the mob is is servicing we |
| 7295s | have all of that functionality and it |
| 7296s | just you know something we're going to |
| 7298s | be testing all right and then zil far |
| 7300s | wanted about character creator will we |
| 7302s | be able to upload an image into the |
| 7304s | Character Creator and have it render |
| 7306s | that character for us we've had a lot of |
| 7308s | people asking this actually yeah not not |
| 7310s | for Alpha 2 but it is something that I |
| 7312s | know um Clayton is excited about trying |
| 7315s | um and it is something that we're going |
| 7316s | to discuss as a feature for the |
| 7318s | Character Creator post A2 launch um but |
| 7322s | not for the entrance of A2 um and since |
| 7325s | we're talking about character creator we |
| 7327s | did have a lot of people asking like |
| 7328s | when will we get get access to uh |
| 7330s | character critics we have said close the |
| 7333s | goal is to put that out before Alpha 2 |
| 7335s | so people can get their hands on it it |
| 7338s | is getting close I just I just saw an |
| 7340s | update last week from the character |
| 7342s | Creator uh for the stable environment |
| 7345s | which is basically as you create your |
| 7346s | characters it's looking pretty good um |
| 7349s | we need to get it over the Finish Line |
| 7351s | on on on the outstanding polish work |
| 7354s | that's necessary there you can make some |
| 7355s | whack characters right |
| 7357s | now yes it's getting close I think you |
| 7360s | could like at one point we were able to |
| 7361s | like combine some of the raises it |
| 7364s | looked like a little bit of a mess but |
| 7365s | um they're fixing it they're working on |
| 7367s | it uh snow Al wants to about server |
| 7370s | meshing Tech uh well server mhing have |
| 7373s | any impact on the world bosses or events |
| 7376s | and is there any potential chance for |
| 7378s | exploitation from the player base in |
| 7379s | bypassing the spawn timers via this |
| 7382s | concept no there's I I wouldn't say |
| 7384s | there's really any correlation between |
| 7387s | you know exploiting or um uh respawn |
| 7391s | timers will it had have any implications |
| 7393s | for bostech absolutely I mean as as we |
| 7395s | just observed right when you talk about |
| 7398s | certain emergent behaviors whether it be |
| 7400s | raids coming together to contest a |
| 7402s | particular boss fight the server meshing |
| 7404s | Tech along with our Dynamic griding is |
| 7407s | now going to be capable of of of |
| 7411s | essentially you know reducing the scope |
| 7415s | of a particular server's Authority and |
| 7417s | spinning up new servers to add to the |
| 7419s | grid to facilitate that high player |
| 7422s | density right that's one of the benefits |
| 7424s | that's that server meshing teken and and |
| 7427s | what the networking team has been |
| 7428s | working on helps to facilitate as our |
| 7431s | design goals which is High player |
| 7432s | density so yes it benefit bring Absol |
| 7436s | absolutely so so so to answer that |
| 7439s | question yes there are going to be |
| 7440s | benefits there that's what A2 entrance |
| 7442s | will contain as part of our networking |
| 7444s | technology that's what our networking |
| 7446s | teams have been working on for the past |
| 7448s | several years but no to the exploitation |
| 7451s | part of that no yeah no there's not |
| 7453s | really any there's not really any |
| 7454s | considerations from a server meshing |
| 7456s | technology standpoint that would assist |
| 7458s | in exploitation or problems there |
| 7461s | obviously ly you know like any |
| 7463s | networking layer there are bugs that we |
| 7465s | need to hammer out there's going to be |
| 7468s | you know issues that that testing will |
| 7470s | will give us insight to but that that |
| 7472s | would be the same it's already started |
| 7473s | giving us data on it so it's been great |
| 7476s | having testers uh Tyran tyranthraxus |
| 7479s | wants to know about underr content will |
| 7482s | there be underwater content in the under |
| 7484s | realm parts of the game yeah uh under |
| 7487s | realm will have water uh associated with |
| 7490s | it and there may be areas where we um we |
| 7494s | leverage underwater content as either a |
| 7496s | means of traversal to reach certain |
| 7498s | locations um or uh require use of |
| 7501s | certain vehicle types in order to |
| 7503s | Traverse and our penultimate question is |
| 7506s | from Ron sconi I always love saying |
| 7508s | their name uh profession choices will |
| 7510s | your own profession have any effect |
| 7513s | besides a social one on which tradable |
| 7515s | recipes and blueprints you obtain um I |
| 7517s | talked to vnar a little bit more about |
| 7519s | this and I uh I think the example here |
| 7520s | is will I still Ain and loot recipes |
| 7523s | that are not my own profession and can I |
| 7525s | trade these with others yes absolutely |
| 7527s | it is an open market economy um the |
| 7530s | concept here is that you know you don't |
| 7533s | need to do everything yourself per se |
| 7535s | other people can do them and you can |
| 7538s | excel in a different area in order to |
| 7539s | make trades with them in order to |
| 7541s | purchase from them you know that is the |
| 7543s | basis of a healthy economy yeah we want |
| 7545s | you guys to be trading things and |
| 7547s | needing things from other people that's |
| 7549s | the whole point of an MMO absolutely um |
| 7551s | and then our final question here is from |
| 7553s | lashing wanting to know about stamina |
| 7554s | modification will there be specific |
| 7556s | stats on gear outside of set bonuses |
| 7559s | that modify or interact with stamina |
| 7561s | such as increased maximum stamina |
| 7563s | stamina regeneration or stamina gain on |
| 7566s | basic attack hits yeah so um first and |
| 7570s | foremost uh stamina is a waterfall uh |
| 7574s | which means it does get influenced by |
| 7576s | certain core ability stats however in |
| 7579s | addition to that trickle down benefit |
| 7581s | that will come from certain set bonuses |
| 7583s | that might modify Constitution um which |
| 7586s | could increase you know regeneration |
| 7587s | rate of stamina you will also have |
| 7590s | independent one-off stamina benefits |
| 7592s | like you know 10% stamina regen that's |
| 7595s | that is just a set bonus or you know 15% |
| 7598s | higher stamina value uh cap you know |
| 7601s | that gets applied to certain sets and or |
| 7604s | armors and Buffs as well and that wraps |
| 7608s | us up for our June development update uh |
| 7612s | again thank you all so much for uh |
| 7614s | sticking with us as F final reminders |
| 7617s | head on over to our forums leave us your |
| 7618s | feedback uh forums. ashesofcreation |
| 7625s | docomo to Alpha 2 so stay tuned for that |
| 7629s | and um we will be expanding our NDA |
| 7631s | testing so um for those of you who do |
| 7633s | get access in we will be emailing you |
| 7635s | sometime this month so stay tuned for |
| 7637s | that that will be under NDA testing um |
| 7640s | so you can't talk about it it's like you |
| 7641s | know Fight Club um but it will be |
| 7644s | exciting and we'll be expanding that to |
| 7646s | all of our A1 users as we move forward |
| 7649s | um thank you to the wonderful St server |
| 7651s | technology so big huge shout out to Zach |
| 7654s | Hunter Anton for putting together an |
| 7656s | amazing presentation and walking us |
| 7658s | through some big brain stuff but |
| 7661s | hopefully that was entertaining for you |
| 7662s | guys um I know it's a lot different than |
| 7664s | showcasing gameplay but don't worry |
| 7666s | there's going to be a lot of gameplay |
| 7667s | next month so or not next month this |
| 7670s | month later this month this month later |
| 7671s | this month so you won't miss out on that |
| 7674s | we'll definitely be sharing more of that |
| 7675s | and of course thank you to Brian and |
| 7677s | Stephen for joining us um and we'll have |
| 7680s | the video up over on YouTube really in a |
| 7682s | moment here so stay tuned for that we'll |
| 7684s | post it on all of our social channels |
| 7686s | follow us on all of our social channels |
| 7688s | we are ashes of Creation in all the |
| 7689s | places we post at least like once or |
| 7691s | twice a day almost um we try to keep you |
| 7693s | a breast to what's going on we ask a lot |
| 7695s | of questions of you sometimes they |
| 7697s | affect in-game I think one time we asked |
| 7699s | you if you prefer B for bag or I for and |
| 7702s | we actually based on the votes we |
| 7704s | changed it in game so there's a lot of |
| 7706s | things where like you can just be part |
| 7708s | of the development process and follow |
| 7710s | along with our journey um and of course |
| 7712s | note that the video the full development |
| 7714s | update video for this one won't be going |
| 7716s | out tomorrow it'll be going up July 5th |
| 7718s | because tomorrow's a holiday I would |
| 7720s | like to not have to work um so we'll be |
| 7722s | P we'll be pushing that to the 5th um |
| 7726s | and of course if you want to be |
| 7728s | spotlighted on our next one leave a |
| 7730s | comment there make sure you subscribe to |
| 7731s | our Channel whenever that goes up on the |
| 7733s | 5th and we will see you all later this |
| 7736s | month so thank you for joining us yeah |
| 7739s | that was fun awesome thanks everybody |
| 7741s | bye bye |
| 7743s | [Music] |