Third Party Photon Quantum: Deterministic Multiplayer Engine for Unity

Photon Quantum, the technology powering Stumble Guys, is now FREE for development.

Five years ago, we started to build the world’s first fully deterministic game engine. Initially, we focused on dedicated and professional studios. Due to its intricate nature, Quantum has been exclusively accessible to curated Gaming Circle members.

Now we feel ready for the next step: Quantum is FREE for development. Thanks to its mature and robust feature set, we are confident that game developers around the world can now leverage this powerful deterministic engine without financial barriers.

Photon Quantum allows by design to create multiplayer games as simple as developing a local multiplayer title. You can focus on creating amazing gaming experiences without writing netcode, network synchronization, or platform limitations.

How Does it Work?

Quantum uses deterministic rollback which is a network model in which each client runs a fully determinstic simulation and only inputs are exchanged over the network.


Deterministic rollback has been the go to choice for 1 vs 1 games such as latency-sensitive fighting games and RTS games with thousands of units. Quantum improves on the model by allowing up to 64 players to play together by coordinating with our cloud relay. With that, Quantum brings the advantages of the determinstic rollback model to a wide variety of game genres such as sports games, fighting games, ARPG, RTS, MOBA, physics based games and many more.

You can learn more about how Quantum works here.

Key Features

No Netcode – Multiplayer by Default: With Quantum, you can create a multiplayer game without writing any netcode. Just create a single simulation with multiple connected players, similar to developing a local multiplayer experience.

Low Bandwidth with Input Only: Game clients only exchange input, distributed to the players at high frequencies through our Photon Cloud. Use webhooks to connect your backend or plugins to run server-side custom code.

Optimized for Extreme Performance: Each component of Quantum’s ECS architecture provides the necessary performance to run even physics-intensive multiplayer games on PCs, consoles, VR and on mobile phones.

Cheat Protection by Design: Deterministic games inherently resist cheating. Combat cheaters through methods like replay validation or server referee simulations which is the most effective anti-cheat measure.

Cross-Platform Determinism: Simulations coded in Quantum do not have any dependencies on Unity and can run anywhere.

Zero Lag E-Sport Grade: All local actions are executed with zero lag while remote inputs are predicted and rolled-back for a seamless multiplayer gaming experience.

Replays & Spectating: The engine supports in-game replays and built-in spectating, allowing you to implement features such as kill cams and live streaming.

Bot SDK Addon: Create engaging multiplayer experiences by using bots to fill rooms and replace departing players, ensuring a consistent player base for your game.

Is Quantum Right for my Game?
Quantum is an amazing choice if the determinstic network model fits your game. We have created the Quadrant to point you into the right direction in your development journey when picking between determinism (Quantum) and state sync (Fusion)


Games Made With Quantum

Photon Quantum already powers a huge variety of multiplayer hits such as Battlelands Royale, LEGO Brawls, LEGO Star Wars Battles, and the worldwide smash hit Stumble Guys!

Demo
You can explore a collection of our samples live on itch.io (with multiplayer ofcourse)

Getting Started

Get started with Quantum here and follow our Quantum 100 intro tutorial

Explore the free samples:
- Blueless
- Golf
- API Sample

 

I went to your link for the Q100 tutorial and it said that the sample was only available to members.

Do I have to be a member to go through the tutorial?

Also, I figure I will find it eventually, but what are the total costs involved in deploying with Quantum in my game? [edit: found it, $500/2kCCU/mo]

I also remember talking with someone via email about Quantum for a First Person Shooter MP, and he seemed to be recommending me look at another produce rather than Quantum. But I am really only interested in Quantum, or I will continue to roll my own due to ownership and costs. Could I use Quantum for a Halo or Apex style shooter? If not, what are the drawbacks of using Quantum for games like those?

 

I'm sorry about the Q100 permissions. Of course you should have access to the tutorial sample. Note you can already go through the tutorial as it starts with an empty SDK project. The download is for the finished result of the tutorial. We will get the download permissions for it fixed ASAP.

Regarding the price, you are correct, you just straightforward pay for CCU as described here. No additional costs. Keep in mind that Quantum is a serverless solution so you do not need to operate dedicated game servers making it a very cheap option as dedicated servers on are generally much more expensive per CCU.

For an FPS shooter we indeed rather recommend Fusion (see the Quadrant maxtrix above). While you can make a shooter in Quantum and some studios have already sucessfully done so, we recommend Fusion because:

1. Quantum has no built in lag compensation (lag compensation meaning the FPS term; server side rewind for raycast weapons). While it is technically to implement this it is much more advanced than with a server authoritative solution.
2. Quantum can not hide information from clients. Fusion can use AOI to hide enemies completely from the client if they're not visible to prevent wall hacks. Quantum always has the full game in memory so cheaters can extract that data.
3. Due to lack of AOI, Quantum is not well fitted to support large scale maps with a lot of players (BR style games).
4. Some systems are missing in Quantum and would need to be built from scratch such as the option of having tick accurate animated hitboxes for characters. With Fusion and our animation samples it is much easier to achieve this.

So TLDR; while you can build an FPS with Quantum, to achieve the quality of competitive shooters a lot of work is needed. For more casual experiences it can work fine though especially mobile shooters. Fusion already comes with everything out of the box for making competitive FPS games and we have plenty of shooter samples to get started.

 

I appreciate the detailed answers, thank you.

edit:

So one of your answers about animation raised another question.

Are animations of characters or any asset done by Unity even for the entities managed by Quantum? I saw the video tutorial that showed how to setup a character, but it wasn't animated. I was assuming the animations would not be managed by Quantum

I would implement the animation to be driven by character state (previous animation state, velocity, aiming vector, etc.). I would think this would be very accurate.

 

Most multiplayer games animate the way that you explained. E.g. in the gameplay simulation animation does not exist and then the view drives animation simply from the gameplay state. This works well for most types of games but it only works if animation does not affect gameplay.
- In Quantum this means animation is outside the gameplay simulation fully on the Unity side.
- In Fusion this means animation is not networked but instead played locally and driven by networked state.

In competitive shooter games animation is part of the gameplay since characters have hitboxes that follow their arm / leg / head animations and these colliders need to match exactly what the player is seeing on the screen to provide a good gameplay experience. So you need a solution for that. So if you want accurate animation in Quantum it needs to be inside the deterministic simulation. What this means is you need to bake the movement of the hitboxes into a deterministic format and animate them yourself using a custom system.

 

I appreciate your answering my questions. Here are a couple more thoughts and more questions.

I understand what you are saying about why Quantum is not the best choice for competitive PvP game mode. And I agree it wouldn't work for what I want to do for my PvP game mode.

But I was thinking it would be a great fit for a coop campaign. I was thinking of selling a copy of the game (campaign) for $30 a copy. If I took 25 cents per player per month to pay for the service, that would come out to be $30/10 years, which is my expected service life of each title. Of course, no one will play the campaign every month for 10 years, and many will not play coop, so perhaps this is a non issue, but I want to understand the costs and risks related with the decision to go with Quantum.

Then I thought, what if Quantum servers were discontinued or the pricing later became unreasonable?

So my first question is can I get an agreement for a set price ahead of development so that I know what my costs will be over a 10 year period prior to development?

My second question is can I get an agreement to acquire the rights to host my own servers in the event that Quantum's server support is discontinued for any reason?

My third question is can a game that I develop with Quantum run without actually using the Quantum servers if the player wanted to solo and not use coop? In other words, can I play a campaign designed with Quantum for coop in a stand alone mode without internet access? Or does the solo player require internet access to the Quantum servers just to play solo by himself?

 

First let's talk monetization. In your calculation you are assuming that a majority of your players keep playing the game actively over 10 years. In practice this is never the case. Only a very small core community of all your buyers will do that. Most people will open the game maybe a few times in total or have some short periods where they play actively and then put it aside for a while. To put this idea into more concrete terms, we usually use the following when tracking user counts:
Peak CCU: Peak concurrent user count of the day
DAU: Total amount of active user accounts during the day.
MAU: Total ammount of active user accounts during the month.

What you pay us is for peak CCU. Based on our data on thousands of titles the translation of CCU to MAU is roughly.
1 CCU > 20 DAU
1 DAU > 20 MAU.
So a factor of 400. So for your game to have 1000 CCU that would mean on average 400k MAU (and obviously at least that many copies sold as not everyone is active each month). The factor might vary a lot depending on the game or the community etc. so you need to account for that ofc. but the overall pattern is the same for most games as you can also see on Steam Charts.

So will you run into monetary troubles when selling a game for $30 that has a lifetime of 10 years? The answer is honestly not at all. Our costs will be barely noticable. Single digit percents of revenue at most. We have free to play mobile games that can easily sustain our costs while making much less ARPU (average revenue per user) than your 30$ price point. Our guideline is usually 1-5% of revenue you can expect to pay us and that has been the case for almost all titles so far.

Finally in regards to this. 10 years lifetime is a very long time for a game. Most multiplayer games can sustain themselves for less than 2 years before they die out. Some games can sustain themselves and build a community but this is rather rare so you should not expect this. But if you do everything right and end up there you should consider monetizing in some way. This does'nt have to be any form of toxic monetization like agressive micro transactions. It can be merch, expansions, skins etc. it's fair to charge something, because your players are getting thousands of hours of entertainment for 30 bucks which isn't really fair to you. And therer are far more expensive costs than our CCU costs involved in running a sucessful multiplayer game. You will need:

• Customer Support, and you want to do this well.
• Operations team for keeping all your other servers running such as your database, authentications etc.
• Development and maitenance
• Marketing
• Legal and regulatory
• Many more......

You can dodge most of this while you are small but if your game gets moderaretly sucessful they will stack up quickly. In the end you can hardly expect profit margins that are higher than 30% for a multiplayer game. So 70%+ of your revenue will be spent on various expenses. 1-5% of that on us.

I'm not sure if I understood the question. Are you asking for a one time purchase that gives you the capabilites of running Quantum with any amount of users for 10 years? Or just a fixed rate for CCU? That is probably not possible but anyways would need to be discussed with the business team (info@photonengine.com)

Needs to be discussed via email. Generally, no. We have a very good track record with Realtime, PUN, Bolt etc. for keeping services running so this hasn't been an issue for our customers so far. Your concern is very valid though and always a great risk when using third party services.

Yes offline solo play is available, works out of the box. Does not require an internet connection and does not cost any CCU. You can also do split screen. You cannot do local co-op play using multiple devices without an internet connection.

 

Thanks again for your answers. I have trouble believing that my profit margin will be less than 50% TBH. Time will tell I suppose. But I can see your point about how often or not people will play the game over 10 years. I agree and i thought I mentioned that I realized that as well in my earlier post.

If I cannot guarantee a service life of 10 years, then I don't see how I would commit. I'm surprised you said that most games are far less in service life. Halo 3 had a service life of 14 years before their XBOX 360 servers shutdown. H3 is one of my favorite games, and I still play it from time to time, though I had to purchase MCC (a second purchase), but the cost was a lot less per title overall, something like $5/title I believe.

I'm rolling my own netcode, it is fun as hell, and I know it's going to be there for me to continue to evolve and it provides a framework that makes sense to me. I can also ensure that I can do things that I feel are no-brainers, such as picking up assets and attaching them to the right transform of the character with no special or complicated coding. I mean, this isn't difficult, it's just a lot to do.

What I really like about the pricing of Quantum's servers is that it is really low for the service provided, to maintain the servers for me, etc. I just need to know that I can recover from an unbelievable event.

 

Service life is more or less a success and financial topic.
No service will give you such guarantees for a low number of players and unless you can guarantee to pay for the servers running the game all the time.
Service Level Agreements can be defined for bigger titles. It's just not feasible to start with that expectation while also paying less than $100 for 12 months of service (e.g.).

And we are still running age old PUN 1 games, too.
Just wanted to add those details. If doing netcode is fun for you, by all means do that

 

Quantum sounds like an amazing tool ! Thanks for making if freely available for development.

Is support still limited to gaming circle members though ?

 

You will get the same support with Quantum as with our other Photon libraries, you can ask us questions on the Photon Discord and Unity Forums and we will do our best to answer them as time permits and if you have an active paid plan with us you can get email support as well.

Circle is extra on top of that and provides you with premium support directly live from our team and access to Circle content such as our samples.

 

We have two new Samples available for Quantum Free!

Bomber
Bomber shows how to build a bomberman style game in Quantum. It features a tile based collision and movement system, a procedurally generated map and many more features.

https://doc.photonengine.com/quantum/current/game-samples/bomber

Joints
Learning a new physics engine can be hard, especially complex topics such as joints. This sample shows you how to setup a variety of physics joints with Quantum.

https://doc.photonengine.com/quantum/current/technical-samples/joints

 

Does Photon Quantum offer a matchmaking system so players can simply click a 'find match' button and have Photon search for other players to group into a match?

 

Yes Quantum uses our default matchmaking API that most our SDKs use that has a quick match feature like that.

 

Can you share real quick how your match making system determines team matching? is it simple SBMM, or just random, or something else?

 

Is there a steep learning curve to Quantum when coming from Photon Pun?

 

You can do a simple form of SBMM by grouping players into skill brackets using SQL lobbies but this is fairly limited: https://doc.photonengine.com/realtime/current/lobby-and-matchmaking/matchmaking-and-lobby

Our matchmaking works for simple cases for anything advanced you will need a separate matchmaking solution.

 

Yes there's quite a learning curve with Quantum. It is completely different than PUN and also a new gameplay engine.

 

Been following Quantum for a while and glad to see it's finally out. Deterministic multiplayer sounds awesome.

 

Is Photon Quantum compatible with the Universal render pipeline?

 

Yes you can use Quantum with URP / HDRP and the built in renderer.

 

Any idea when photon quantum is coming out for Unreal? As I am very sceptical about moving forward with unity due to the sudden pricing changes.

 

Is it necessary to use pointers when working with Quantum?

 

We do not have an ETA on our Unreal version yet. It will be first released as a preview to members of our Gaming Circle and later on as a full release.

No it is not necessary. You can use the safe APIs which don't use pointers. E.g. instead of `frame.unsafe.get` you can use `frame.get`. Keep in mind that the safe APIs return a struct which is a value type. So modifying it will not apply that automatically to the component and you will have to use `frame.set` after you are done modifying a component.

 

I see this is available for VR as well? Is there any instruction on using it with Quest2/3?

 

VR is possible but we generally don't advise it as the complexity of VR development + learning Quantum from scratch is a bit too much for most people. Because of that we only have VR samples for Fusion which is our state sync based networking library.

 

Hello , what is best for a multiplayer top-down shooter between Fusion and Quantum please?

 

Both of them can be fine solutions. It depends on the details, in most cases Quantum is the better choice especially on mobile platforms. However if you want to support huge players counts on very large maps (more than 32 players) then Fusion with a dedicated server can be the better choice to reduce CPU load on the clients.

For Quantum we do have a top-down shooter sample: https://doc.photonengine.com/quantum/current/game-samples/battle-royale-topdown

 

For an FPS game, we're planning to start quickly with Quantum to prove the concept and get product market fit while not focusing so much on orchestration and matchmaking. How easy is it to transition later on to Fusion from Quantum?

 

For a casual / mobile FPS Quantum is fine (no lag comp, and accurate raycast hit detection). If you are building a competitive FPS that needs dual timing, sub-tick accurate lag compensation etc. then go with Fusion. Switching from Quantum to Fusion is a full rewrite of all gameplay code as they are completely different.

If you want to create a bunch of games quickly to figure out product market fit, Quantum is amazing for that. Just be prepared to learn a different solution and restart production from scratch if you want to take a competitive shooter to production.

 

Hello,
Is there any link or thread I can see all the games made with Quantum + Unity?
I want to play them as a reference

 

would Fusion/Quantum be able to perform machine-local (singleplayer offline) per-object timescale physics?

for example, the player will move at X timescale, and the world will move at Y timescale, and somehow correctly interact with each-other ?

 

Hi,

I'm planning to build an online multi-player combat racing game. Which one do you suggest? Fusion or Quantum?

 

Different timescale per objects per physics objects are tricky in general. You always have to run a physics time step for all objects at the same rate else you will not get accurate collision resolution. If you want to artificially slow down an object and have it collide with other's what you need to do is:
- Disable gravity > instead apply gravity manually based on timescale
- When an object changes timescale adjust it's mass and velocity accordingly to slow it down / speed it up.
- Whenever your code applies a force to an object multiply it by the timescale for that object.

This can be done both in Fusion and Quantum but generally speaking it is easier to get good results with Quantum for physics driven use-cases like this.

 

For racing games we generally recommend Quantum. There is even a vehicular combat sample for Quantum: https://doc.photonengine.com/quantum/current/game-samples/keo-vehicular-combat

 

Today, Erick will stream a "Quantum 3 complete course for beginners". In about 3h. Exact time shown on YT.

 

Hello, we are working on a sports football game from past 1 year. However, new we have planned to create an online multiplayer of 1 vs 1 football game. Is it advisable to use Quantum in ongoing project or should we go with Fusion? I can see as per Quadrant, it is recommended to use Quantum for sports game.
For current game physics, we are using Unity's inbuilt physics engine. What is your recommendation

 

Hello, I'm trying to evaluate Quantum but it seems that the sample projects are still only accessible to circle members (which I am not). Is this intentionnal? I was under the impression that since Quantum is now free for developers, samples would be accessible to everyone.

 

Some samples are available in the Circle only. They are marked as that on our website. Free samples are available as well. You can find a few of them linked in the initial post here.

 

Hi! This may be a silly question - how difficult/time consuming would it be to integrate Quantum into an already built game (made in Unity/C#) that is further being developed for multiplayer? Is that something that is the intention, or should one begin their development already using Quantum? Thanks!

 

> how difficult/time consuming would it be to integrate Quantum into an already built game
That won't work well, sorry.
Quantum bascially a game engine within Unity and requires your game logic to be deterministic and decoupled from rendering. Refactoring a game to do this is possible but can be a lot of work.

Then again: In most cases, refactoring any existing game to add multiplayer support is almost never easy. I'd say it depends on the game's genre, complexity and platform(s) to guess if Quantum might make sense (and if it could bring enough benefit to look into it).

 

Is it possible to use Photon Fusion for mobile networks in the above case where Quantum is not feasible to integrate? I tried to integrate Fusion to existing code and it seems to fit well. Do you suggest to move ahead in this case with Fusion?

 

Well. It comes with a few drawbacks but overall, it will work.
Stick to the "Shared Mode", so you don't have to worry about hosts leaving.

 

With Quantum, how does matchmaking work? I want players of a similar player level in a match, instead of a level 50 player smashing a level 2.

How does Quantum do stuff that isn't inputs? For example monster respawn or random events.

How would bots work? Who is spawning and controlling them? Stumble Guys has bots and you're advertising their Quantum use.

 

Matchmaking for Quantum relies on the Realtime API. It happens before Quantum is being used.
So, you can refer to the Realtime API docs for plenty of details. This Quantum doc page gives you an overview of matchmaking. It links to the Realtime docs to dive deeper.

How instantiation/spawning works in Quantum, how to synchronize state and other events, is mostly covered in the Quantum 100 doc series.

All our demos have some simple form of matchmaking, spawning and so on. If you learn best by dissecting a sample, check out whichever vaguely fits your interest or seems simple (always good to begin simple).

There is even a Bot SDK sample

 

Ah yes, there are some free examples.

Shame that most examples and all the community help features are paywalled behind a subscription. I have to pay anyway to use Photon, why not make it easier to become a customer? $500 a month for Discord?

 

We have a public community Discord that you can access. We have to charge for paid support because otherwise we'd have an amount of support requests that we can't handle. This is what many companies do including Unity itself. Nonetheless many of us still try to be active in the public community Discord as much as possible when time permits.