Session: Multiplayer Mastery: Key Considerations for a Successful Start - Robert Boehm, Photon.

[Music] multiplayer Mastery key considerations for a successful start we've got Robert from phonon so yeah please again round of applause and we'll we'll get cracking with a new session thank [Applause] you yeah hello good morning welcome to my speech today thank you all for coming I will talk today about some key things you should consider in order to make your multiplayer gameer start from the very beginning so things you should actually do before you start developing your game right um so I'm Robert I'm vice president D at

Photon I'm in the industry since about in the multiplayer Industries particularly uh since about 20 years I was uh also uh founded my own game studios I was working on some nice Brands like like bomber man or Giana sisters so I worked on both sides on the buyer side so looking for multiplayer technology because all my prod uh projects I did as a a game Studio were multiplayer but also on the vendor side so for Photon I'm working as a VP so selling basically uh multiplayer technology to game developers to make uh them successful so I know both sides of the story uh although I cannot predict the future I think I can give you some good insights of things you should definitely need to know before starting uh your multiplayer Journey so the first uh part of my uh talk will be about what you need to know so some basics of

multiplayer technology everyone should know before starting and the second part is when you have a game idea how do you actually choose the right multiplayer game topology uh when I talking about a topology I mean there are certain ways and we think there are four good ways to build multiplayer games back on track and how do you know what is the right way for you I will talk about that in uh the second part of my talk today so what do you think how many people actually plan their multiplayer before going into production and this was a little bit shocking so this is a number from the unity um gaming report 2024 so only 62% of Studios actually plan before they start their multiplayer production and I mean we on the master of the multiplayer track and I wonder how can they rest

there in this uh uh pie chart uh Master the multiplayer if they don't plan at all so for me it's really shocking that 38% of players uh of Studios sorry do not plan the multiplayer so you shouldn't do that because there's one very simple rule everyone should really understand and know and I call it the first law of multiplayer you do not simply add multiplayer so don't start with single player and saying oh I will add multiplayer at a later point in time so this always will be more complicated will take longer in total so it's not not a good way of doing things you should always start right at the beginning with multiplayer in mind because otherwise you have to reite lots of your code and you will go on a bumpy road if you if you do that so never do that so uh multiplayer can be a little bit intimidating for some developers I

know so because of the complexity of it so the complexity depends a little bit if you're doing a more casual style of game like like co-op games or more PVP competitive stuff the more competitive your game is the more complex it gets that's for sure and then there are tons of stuff around multiplayer you don't need to know everything you need to know a couple of things uh but it can be a lot and it looks very complex and many teams um have not maybe the right team size and everything to handle everything so like life Ops or what about cheating what about do I need voice moderation if I have voice in my my game AC crossplatform spectating matchmaking and then there are stuff I will talk about later a little bit snapshot interpolation client

side prediction lack compensation what is all these so it looks very intimidating very complex there's tons of stuff but the good news here is there's tons of not tons but some services at least out there that help you to reduce this complexity uh by a lot so you can concentrate on what you are good at on building a good game experience and this I can definitely recommend so don't so try to to reduce a complexity not only in terms by choosing the right services or products whether free or paid uh but also by really concentrating on what you think makes your multiplayer experience fun for the player so many people start have a rough idea and they always ask me about oh what about cheating and they think more about cheating than creating a great game experience cheating is an issue of course you shouldn't uh not think

don't do not don't think about it at all uh but it shouldn't be the first thing that comes into your mind right so try to strip out everything that's that's possible to reduce the complexity for for your team and concentrate on the things that are really important for your game play so whenever uh someone ask me hey Robert uh why shouldn't I use free technology why shouldn't I build myself or build your own or and why should I use paid services so I'm a kid of the '90s so not sure who knows this lady there but whenever someone ask me that uh there's a song coming into my head the best things in life are free by jenet Jackson uh back in the '90s so sorry I'm a little bit old but I like uh so I always think about this and then my answer is so unfortunately nothing is free so the German in me says no

uh nothing is free all comes at a cost because there are development cost and their cost when your game is live and when you have to operate it and development cost I mean your team cost money operation cost servers might cost money or Services cost money so look at the stuff which is out there check are they bringing updates regularly how do they handle buck fixing uh what about dogs and everything is a good uh documentation there can I find good samples are there great tutorials out which I can look at and learn from so there's a lot of things uh and when I get in trouble is there support can I reach out to someone is there someone helping me and when the game gets life what happens then can I scale is there a proof that I can scale with that technology ology I have

chosen right so while some things might be free in the beginning there's a cost Associated at it all the time so multiplayer uh understanding the basics of course multiplayer is about time and sometimes again uh is it a '90s movie Back to Future so it's about time travel and multiplayer development sometimes reminds me a little bit on on these movies where you have different time streams so like players seeing a little bit different things on their uh game clients and uh a good multiplayer game handles these different time streams in a way that it's fun for everyone so it sounds simple you have multiple devices people playing on multiple devices platform but you have one game world and you want to keep it in sync across all the devices

but unfortunately there's a thing called latency and this is inevitable so no one can avoid that no technology can make that go away completely but good technology can give you a way how to handle it so there's always latency and what what is even worse is Jitter so you have jittery data coming in and now your job or the solution of the net codee is that is responsible for timekeeping of all these different time streams and a smooth representation of this data for the players so they have a good experience so there are some fundamental techniques for it they are most commonly known from uh shooter games but I cannot give you examples for for all type of John

so when thinking about specifically shooter games not only but then there are three fundamental techniques and everyone of you should know them it's snapshot interpolation it's client side prediction and it's lack compensation these are super critical for many many games so snapshot interpolation what does it do it's taking this jittery data and um interpolates it so your game client so you have a smooth representation so this red dot is a jittery data and snapshot interpolation makes it just smooth so why is this needed uh think about trying to aim in a shooter game so it's of course much better if it's a smooth uh movement right then there's client side prediction that means that you take these uh

uh snapshot interpolated data and extrapolate extrapolates it into the future so you don't see the past but a prediction a prediction can be wrong so your net code needs to roll back reset and res simulate with the confirmed data from a server and then there's La compensation because these both things are great but still when you don't have lack compensation it's a little bit tricky to to explain just in a couple of words so there are tons of YouTube tutorials where you can learn the exact how it works but without lack compensation it's simply uh not a much fun of playing a shooter game because you won't be able to really hit at what you see so when you start uh shoot the game for instance and you don't know about these things uh you might have fun in

the beginning when you play in your own network with a constant stream of of uh latency not so much Jitter but when your game goes out in the wild you will experience that many people will complain because they don't have fun they cannot simply aim well at other people they cannot actually hit at what they see and this is super frustrating so if you start with the wrong technology if you don't know about these techniques uh you will end up having a game which is not a lot of fun so it's likely it won't be a success so okay now how do I choose the right topology and by topology I've mentioned it before I'm talking about the four ways we think uh you should build your multiplayer game there are more ways but we strongly think that these four are the ways you

should go so this is a dedicated server which many of you I guess know so there's a central server which is the authority of the game and a central referee or judge so he decides the central server decides what's wrong what's right who what's hitting who what's the truth so it's a single source of truth then you have client host where one of the players is the host this is great for especially co-op games when it's not so competitive and you have H shared Authority that means that all clients have full Authority so there's no Central instance often there's a concept used called Master client so one client is the master and decides and this acts as a kind of authority this is also good for certain

type of games um I will come to that later and then there's a old but pretty new uh way of doing things it's a deterministic way so determinism is there since since a long time um but there's not really uh or there has not really been a product out that you can use off the shelf so hint we developed one uh and this is a great way for doing things for certain type of games and game genres and for certain type of platforms all these blues uh all the blue uh squares are uh are working with State synchron uh State syn Technology that means you synchronize all the objects across all the clients so speed rotation of all the objects you synchronize them across clients um this is limited to some extent if you

think about RTS games where you have thousands of units and synchronizing thousands of units across the network leads to issues with the bandwidth right it's too much data so uh uh RTS games have been built forever with a deterministic approach so at first idea of what you should use for your game is whether your game is rather competitive then you likely end up uh doing a dedicated server game or a deterministic game if uh your game is more casual you'll likely use rather client hosted or shared Authority so I'm again a kid of the 19s whenever someone asked me how can I choose the right topology for my game G

uh in my hand there's always Mr Bill Clinton who once said it's the economy stupid of course it's not the economy for games for games it's a game play so I'm sorry but I had to bring this because it always pops up in my mind whenever someone ask me that so and we have developed a tool then uh and this tool oops should be a little bit lower there but anyhow uh this tool was used by this guy Fabian Z which is uh Unity uh employee a senior net code engineer and he was using it for his uh speech at unite 2024 so I thought y uh seems to be a good tool we have developed it because we have many people uh many customers who started with the wrong topology and then have trouble or uh luckily lots of customers are asking us before they start hey Robert can you help me what is the right topology for your game what

should I choose because there's not one silver bullet for all type of games that's why we have different products that's why we have our Quantum which which is our deterministic engine or Fusion which is our state sync engine and um so and but Fusion supports three ways of doing things server host and shared and you have to decide for the right one but how do you do it with our quadrant so you take your game play and you look for a game which is similar to what you plan to build but there please be true to yourself uh if you want to build a battle royale FPS TPS game like pubg fortnite or Apex Legend you would be in the dedicated server area if you would be build a game like Clash Royale or Age of Empires you would be in the determin IC area if you would

um build a sports game like soccer you should use a deterministic engine a fighting game like Street Fighter is definitely deterministic right but if your game is more social then you can use shared mode or client hosted and uh so you just use our so we have uh these uh brushes here so you can have them at my booth in the hall and we have also uh has a tool on our website of course and um yeah you look for your game genre and then you know what type of topology uh you can use or you should use but please also look at these uh small uh remarks there at at the uh edge of each of the uh squares because uh client hosted is great it's it's super cheap it's uh pretty easy to do but it's not uh made for mobile for instance or

dedicated server is is great but it's very costly that's why I meant stay true to yourself do you actually want to build something like pubg which is a very competitive uh uh shooter game or do you want to build something like our Pixel Gun 3D which was a very very successful and popular title which is using this approach which is much cheaper to operate so cost wise this is way better because dedicated server games are super expensive right and that's why you should be really careful with deciding which way you should go and if you're not sure just uh send us an email and ask for it please do so we are really happy uh to to tell you what's the best topology even if you don't use our our stuff then in the end but it will save you so much trouble so in essence if you know don't plan you have no part right you have to learn the basics you have to

understand that there's not one technology please don't take the first thing labeled multiplayer and go with it uh you have to understand these topologies you should focus on on features that are critical and you should strip out everything else that's so reduce complexity please but ever launch with Bots if you do a multiplayer game you should definitely launch with Bots because especially in the beginning Bots are simply super critic and lastly I said nothing comes for free so multiplayer is more costly a little bit but if you take the right tools it's absolutely doable even smallest teams can build fantastic uh stuff like um One customer stumble guys a small team from Finland just four guys one developer and they created a Global Mobile World hit uh within a very short time frame actually it was just just six weeks uh from beginning uh to uh the launch in

the Google Play Store yeah thanks any questions thank you very much um maybe one quick question if we've got a burning question we do have one quick question yeah hello thanks for presentation um I noticed that you didn't have in the like the Gat sheet they're like racing Sims you had like vehicle combat but not racing Sims where you would put those um yeah H we have arcade racing there it's in the deterministic thing because if you are think about networked physics these are very tricky and for racing you usually need uh networked physics if you

have a more arcade racer which is not so much physics driven uh then you can also uh go with with a fusion and um so then depending on on the platform so I would go with the deterministic then then you're super safe you can have uh damper physics and everything and everything just will work very smoothly with with Quantum there's no net code so you really just build your local racing game until the Racing Car it feels good it steers great it has the physics that that you like how it slides and everything and uh multiplayer will just work out of the box and you can launch on all platforms without any problems if you do uh you could also do client hosted uh but you know that wouldn't work on mobile so I would definitely do uh uh deterministic all right that's all we got time for thank you so much I'm sure

you'll be ready for questions sorry available for questions after so yeah big round of applause that was fantastic thank you so much [Music] he [Music]