Realtime immersive - Militar simulator cryengine

Personally, i LOVE ArmA. I LOVE IT. But at the same time i wouldnt be very pleased with ArmA 3 still being on the same engine as OFP, ArmA1 and ArmA 2. because i think there is a very high chance some of the problems that OFP had will still be here for ArmA 3.

Hell, they wouldnt have to use CryEngine at all but they do need a new engine at some point, if they write it themselves or not.

Yes, i really agree with this. The physical part of the RV engine hasn't be upgraded enough since OFP. I'm playing both OFP and ArmA2 on a daily basis (well, for various reasons), but while i don't care to see "a body playing the death anim then being rigid like a stone while falling down from a building roof, when not just floating in the air" in OFP, seing this in ArmA2 again is not good. And in ArmA3 just not possible anymore.

Surely you could allow for the synchronization of some things? ArmA 2 doesn't perfectly synchronize anything outside of the players view distance, go on to a MP server and the discrepancies between clients can be several hundred meters with fast moving vehicles.

Yes, you can, out side of what is able to be seen the need for immediate sync isnt needed, though there are some cases where I'd love it to be more sync'd (especially in ACRE, where things need to be tracked long distance, so much so we have thought about sending our own position updates).

arma just needs physics and all objects before they are realsed into game should have points of cover which ai will seek if it is the closest and best piece of cover, physics would actually reduce the game lags, but just better ai reactions are needed....

---------- Post added at 11:32 ---------- Previous post was at 10:35 ----------

I think BI should pull their fingers out and get to work on keeping their reputation for making great milsims. Arma2 is amazing, and it even has it's own scripting language completely made for it. I think instead of starting a new engine from scratch BI should take the good parts from their engine and start over on the not so good things. AI need improvments but starting again would take too long as the ai are pretty awesome, coupled with ai enhancing mods they act pretty realistically if you ask me, and although they are small, BI has proven they can sit down and make massive improvements, look at operation arrowhead, many people still complain about how the tree LODs are too high when you cant even see half the detail, making you lag unnecersarilly (please excuse my shocking spelling:D) but they forget the quality of the ENTERABLE buildings they have been bitching about since ArmA days....

AI is almost perfect with ai improvement mods made by the community, so BI could easilly take some of these features and improve the ai, and also take suggestions from the public. I dont care if the next arma/ other BI milsim doesnt come out for anotgher 5 years, if they work to their full potential, which I know they have I can wait and play arma2 because it is still a masterpiece how it is now.

Rag doll physics would be the next step, then full physics. But personally, I would just like to see balistic physics and penertration physics, instead of just loads of scripts to check loads of different factors like ammunition used, angle and speed at which it hit etc. If they can accuratly model this stuff without scripts (but only getting rid of the scripts once it is working completely and it is tested) then it wold be a much more realistic game. Since when should an arma 2 player care about graphics though, I think in most cases arma2 graphics are awesome and need no improvement atm because there is more important stuff to improve. .

Things to be improved: AI

Things to start over with: Vehicle armour, weapon ballistics and ammunition penertration done by physics instead of scripts.

Things to add: Physics.

The VR engine doesnt need to be replaced, only some of the majour parts of it do.

I say keep the awesome stuff that we all love, repair what's borken, and just redo the rest.

I have faith that BI will be able to do all of this at a top notch standard, it doesnt matter how long it takes. Go play xbox or playstaion for a week and then you will see how lucky the general public is to have access to the arma series. Keep up the good work bohemia.

---------- Post added at 11:45 ---------- Previous post was at 11:32 ----------

even then full physics are a bit much, and like someone posted earlier, they would just be impossible in large scale multiplayer games... I think balistics and penetration physics should be put in, but other then that, (asides maybe ragdoll physics) other physics should be left out, we dont play this game for the looks, but for the realism, and honestly, id take simulated ballistics and penetration and super ai over cool looking tank explosions and people NOT flying 100ft in the air when shot by a tank ANYDAY.

...Then again, technically, if all the random data is seeded the same on all clients, and the physics engine reacts exactly the same on all clients you can get away with assuming a lot of things are the same on all clients. In practice that is a lot harder.

Sorry NouberNou

Yes that is exactly what I am saying is impossible!

Your physics in this case are inherently random.

Even the same computer running the same complex physics test, of this kind, multiple times will create different results every time. It is inherent in the complexity of the physics calculations themselves! They are beyond the limit of predictability using Classical Newtonian Mechanics. It is why you cannot predict the weather beyond 3 to 5 days or when asteroids in the asteroid belt will leave the asteroid belt. Even if you new the exact position mass and vectors of every asteroid in the asteroid belt and fed in tidal forces from the planets and the sun. Because of the complexity of these particle based systems randomness is bult in.

Follow my links in the post above. It is an effect called "Sensitive Dependence on Initial Conditions" and is a well known limiting factor in complex physics based simulations.

Kind Regards walker

Sorry NouberNou

but that is exactly what I am saying is impossible! Even the same computer running the same test multiple times will create different results every time. It is inherent in the complexity of the physics calculations themselves!

Follow my links in the post above. It is an effect called "Sensitive Dependence on Initial Conditions" and is a well known limiting factor in complex physics based simulations.

Kind Regards walker

No, I understand that, a true physics engine would have those conditions, but for the sake of the problem you can make the physics "not" that way. If you provide the EXACT same initial input and the EXACT same parameters to all the calculations, then the results should be the same every time.

@ any BI devs, ai improvments is first step. Take a look at ace, zeus and GL4 and take some ideas, but also speak to military officials or if it is more convenient talk to someone with experience about how fights are fought. Maybe even go do a boot camp course and fire the weapons and do those live fire excersises where you crawl under razor wire with MGs firing over you or something:D

Then consider how a real section/ squad and fireteam/ team works. No more firefights with ai moving up by themselves. IRL people move in groups of atleast 2. MGs should attempt to find the most advantageous position and maintain base of fire and aim to achieve fire supperioty over the enemy to supress them while riflemen fire accurate rounds onto the enemy/ enemy position, and attempting to advance, flank, or fall back. Then think, what would I do? It looks pretty stupid seing enemy just stand up and run while you are landing rounds within a few feet and in some cases, inches from them. Also, ai should react to any incoming fire, whether it is friendly fire or not, because IRL you probably would not be able to tell, and even if you knew, bullets (and mortars lol) are bullets. When increasing ai skill they shoud not just have increased auto-detect ranges, but better weapon handling, fitness (maybe leave fitness and stuff to ace as they are already doing a kick ass job) etc etc. I wouldnt know if this is alreayd imlemented, but maybe give all ai a feild of veiw, and a noise detection range... That way loud or quiet sounds would have different values to whether you are detected and they can only 'see' you if you are in their field of vision. I belive the ai are already suburb and still are tough to beat in a balanced fight in a hardcore coop server, but there are some things that should be implimented or updated/improved to add realism to the game. My points are probably the most obvious and highest priority, but i bet there are probably a few I have missed:D I have heard rumours that a new dlc is coming out most likely within a year, im not going to mention who told me or the contents of the dlc beause it is just a rumour and no doubt it isnt alowed by the terms of use... But if the rumours are true BI, I know this is probably the first time you have ever heard this from a customer but: delay it's release and please add in ai improvments with it's release. It has been done by many amature mod makers, so experienced devs like you should have no trouble with this. All it takes is a bit of research to get the ai workign really well, and we all know you are capable of research. Even if the rumoured dlc is the size of BAF, if it includes ai improvments like the ones i just mentioned, which are aimed to increase the realism of the game (lets face it, most hardcore communities play coop custom missions, and it is hard to get every slot filled in by a player on one side, let alone both in these cases, so ai is a must). If these improvments are implimented, i will happily pay full price and even double that for just the dlc. Heck, i would pay just for these improvments in a patch.

If anyone has any opinions or feedback on what I just mentioned, especially the last part about ai improvments, please mention in next post. The more feedback and the more ideas and suggestions the more likely the engine will be updated.

...

Those primes

•Lockheed Martin

•Thales

•General Dynamics

•Bechtel

•Kapl

•Cubic

•Meggitt

etc. are hurting bad because VBS destroyed the market for multi billion dollar bespoke sims that promise a lot but deliver nothing and replaced with a mass Commercial Off The Shelf (COTS) product that does what it says on the tin. ...

LMAO you really don't know what you are talking about at all do you. Just how many of these bespoke sims have you used? What makes you an expert on COTS products vs Bespoke?

Lockheed, General Dynamics have increased their market share in both high and low cost systems in the last 10 years by huge amounts. In fact they are even developing solutions that actually use VBS2! If anything - at least in Lockheed's case - VBS2 has opened a door to yet another market for them. A market that wasn't accessible to them with their multi million dollar sims. Its a win for both them and VBS. All you need do is wander around ITEC or similar trade shows. You will see Lockheed and other companies demonstrating their products and systems in VBS2!

Thales develop mostly for specific equipment. Again things like procedural trainers for very specific systems and kit. Something that is hard to simulate without significant effort and re coding in a COTS package. And they have taken steps into a "common platform" area which has seen their prospects leap ahead to the point where the sim division may even get split off from the main company to be an entity in its own right.

Bechtel & Kapl (Knolls Atomic Power Lab) are really just "end users". Not really developers. They do develop bespoke systems for very specific applications though. But not really anything that a COTS package could supply accurately so its hardly damaging their market share in their specialist fields.

Cubic Defense has seen a hit. But since it makes the hardware for the training rigs anyway they are actually expanding to support other platforms. So their business for integration and construction of training facilities is actually growing as a result.

Meggitt, well VBS2 and Lasershot products are encroaching on their market. But since over 90% of their sim product line is aimed at procedural vehicle/aircraft trainers (a lot of which are for products or systems they make) again COTS products aren't much use to them.

For many companies that previously developed their own software, VBS2 is just another common platform to become compatible with or build upon. It has displaced some of systems but as anyone in the practical/real world of military training will tell you it is not "the perfect tool" for every job. It has significant limits. Limits that bespoke software created specifically for that system just wont have.

But right now VBS its riding high because it is the chosen system of some pretty impressive customers. And a correspondingly impressive budget. But then only until those customers find something else that suits their needs better. I know BI Sim are working hard to keep up and adapt. But so are others. Its obvious that RTI have them in their sights...

For those of us actually in the industry watching on the sidelines its all rather exciting. Lots of new tech and platforms are appearing all the time. Its a fast growing industry. The market is only getting bigger so there is lots of space for several competing products.

Yea, having started into the VBS2 commercial field over the last year we are not too concerned about other platforms at the moment, at least for the products we are developing. VBS2 will reign supreme for quite sometime based on the number of ACTUAL installs.

You can look at DOZENS of sim companies that make their own platforms and see customers that VBS2 also has. Its very easy to claim people as clients/customers... ;) Its how much business you did and continue to do with that counts.

That is not to say BIS shouldn't be looking out though. :p

On another note, I'd be happy to work on multiple platforms, so I hope to see some competition, just for the sake of entertainment/stimulation. :p

They are beyond the limit of predictability using Classical Newtonian Mechanics.

I should say that, a system using a sufficiently robust Newtonian engine, could reproduce the same results time after time given the same parameters. That's what's so good about game engines, you can tailor them to suit. They don't need to be random, only pseudo-random, which is the same as saying reproducable.

However, I agree that a sufficiently complex game environment is beyond the sync abilities of current broadbands :)

Hi RKSL-Rock

Premise

The emergence of VBS into the market has fundamentally altered it.

Analogy

In much the same way as the printing press did in the book market. Before then books were hand written and consequently often subject to the whims, knowledge, inebriation and skills of the scribes copying them hence subject to bugs created by those human scribes, and even deliberate alterations to the texts and indeed removal of whole chapters and "Testaments" of bibles by the various sects of Christianity. Those old books might have been beautifully made of the finest materials and only available to the rich but the cheap paperback replaced them.

Similarity

The old bespoke sims were are expensive products developed and also totally run and controlled by programmers (scribes) and do not include user created content. Few soldiers ever got to use them and usually only once in their entire careers. Most were and are buggy as hell as in all honesty they are often made by smaller development teams even than BIS use. And they do not get a tenth of the beta testing that a game gets, and yes I include the beta testing that gets done by the community. In the case of a game Millions of people beta test it. Lockheed's CCTT is said to be so buggy only two thirds of the stations ever work.

Difference

VBS works out of the tin and is more or less issued with a soldiers boots in the various Lite versions given out as a recruitment tool. And much of the training content in VBS is being created by the work of soldiers in the field, something that never happened and never will happen with bespoke products. After all, what bespoke product developer wants the customer going off and making the product them selves?

The plane fact is soldiers use it and it has become the de facto simulation for NATO. It is becoming ubiquitous; even the three and four star generals are saying "Oh we can do that in VBS." and as you your self point out the Primes are licensing it.

Will VBS replace a multi million dollar helicopter pilot training simulator? Nah but the pilot may well use it to preview an approach to get a feel of terrain because it is on a laptop in a dusty tent in theatre rather than in an air-conditioned lab back home. Or they encourage their on the ground colleagues to use VBS to get feel for how it feels to be in the pilots seat flying into a particular location and what things the pilot needs to be aware of, so that they can communicate with each other better. A big ass bill board might be a good visual reference on the ground but it may be a just black line seen from the air.

The market has changed from expensive bespoke to cheap mass market. Instead of buying a limited number of expensive sims the customer is buying hundreds of thousands of cheap sims. From the family tailor to Tesco or Walmart. That is the market BIS know because that is what the game market is. The Bespoke Market players relied on using various legal requirements and barriers to entry to keep out cost based competitors. BIS went round this by doing core development using public sourced documents without need for expensive secure labs and allowing the customer to add or refine anything of a confidential nature them selves by the simple expedient of exposing variables in the config. If vehicles top speed is secret use the public data in development or even a similar vehicle then let the customer change the value them selves.

Will VBS be the only player in the market? Nah since VBS's success clearly other major game developers are looking at the market hence this thread!

Kind Regards walker

You do not seem to understand that if the tree stops bullets as you say.

Client {A}'s Computer says he is safe behind a tree. Client {B} shoots him because according to his computer he is out in the open with no cover anywhere near him and Client {C}'s co [blah blah blah]

Don't condescend me, Walker. I understand full-well what you're trying to say and do. I just choose not to pay attention to it. You always "win" the endless circulal arguments simply because no one can match your inhuman persistence to spew theoretical bullshit and wikipedia quotes that have nothing to do with the facts of the matter.

And forget about the trees, they had little to do with my actual point anyway:

You don't have any idea what the CryEngine3 is cabable of doing on a gigabit LAN MP. It has not even been released yet.

I'm sure whatever training scenarios this Realtime Immersive engine will be used, they probably don't actually need a tornado tearing some beach shacks to millions of pieces, or thousands of red barrels exploding simultaneously in realtime multiplayer.

Hi Pulverizer

I am sorry I upset you. I know realising that under the presently understood physics of the real world; the fact that an all singing all dancing physics simulation in MP is impossible; is not something people want to hear. I too will be happy to proved wrong, it may surprise you to know.

The reason for my persistence is that anyone saying this type of physics will work in MP does not understand real world physics, and they are perpetrating a myth on those reading what they write. Worse they are raising peoples hopes of rescue from this unpalatable reality only to have them dashed.

Chaos theory is not some silly thing for physics nerds; it is a fundamental part of your real life and effects every thing from the way your heart beats to, the way your mind works, to world economics and animal populations over time etc. etc. Every iterative process is fundamentally affected by chaos, the predictability of weather is set by it, as is any other iterative particle simulation type system such as "realistic" game physics and including rag-doll physics.

And the speed of light unless it is a got round by something like quantum entanglement is an absolute. You only calculate a frame of simulation at a time, you could then theoretically have a pipe wide enough to send every client that whole frame in chunk but it would still move at 299,792,458 metres per second in vaccum; though in glass, due to its refractive index, it is slowed to about 200,000 km/s, important to note for Fibre Optics. Either way it takes time to send a message from A to B. So Desync is, under currently understood physics, as inescapable as the third law of thermodynamics. So throwing bandwidth at it while it can improve things to a small degree in the short term is fundamentally a dead end. I should also point out that the requirements of the 6 layer model are a bigger influence on speed of trasmission.

In the end you bang up against the limits.

Kind Regards walker

Yeah but if the average distance the bits need to travel is reduced to mere meters on a LAN then what the hell does it matter if they only move at 200,000 km/s? You'd be looking at less than a thousandth of a millisecond delay, which of course rounds to zero point zero milliseconds of latency caused by the speed of light for any practical application.

I think the lag is more likely to be a function of faster and slower GPU's.

Part of your "ping" is not just the length of time it takes for the signal to travel up and down the wires, but also the amount of time it takes your computer to task it's response to the request for a reply to any transmission.

Hence in a game such as Unreal, I can ping the other computer on my LAN of 0.01, but inside the game engine, the ping goes up to .4 because with the game software running my CPU has other tasks to manage at the same time.

So with a master server calculating all the physics and relaying it to the dumb terminals, we are limited not just by the bandwidth but also by the amount of RAM and CPU GHz the server has.

And for clientside prediction such as we would get in an RTS game, where our falling tree's might be calculated clientside, we have to worry about the relative calcualtion speeds of differing hardware and also packetloss which can cause synchronisation errors.

@ Walker

Increasing bandwidth is perfectly respectable way to increase data load.

The reason people don't do it more is because laying new cable is expensive.

On a LAN bandwidth is almost never an issue.

The amount of data that can be transfered on a 100 MB link, let alone a 1,000 link is more than enough to transmit some decent ingame physics.

In Eve Online for example, a game with very basic physics, upto 30,000 players were being simulataneously served on a 100MB link. So clearly this kind of link is capable of transmitting the co-ords of 30,000 comparatively static tree's.

...@ Walker

Increasing bandwidth is perfectly respectable way to increase data load.

The reason people don't do it more is because laying new cable is expensive.

On a LAN bandwidth is almost never an issue.

The amount of data that can be transfered on a 100 MB link, let alone a 1,000 link is more than enough to transmit some decent ingame physics.

In Eve Online for example, a game with very basic physics, upto 30,000 players were being simulataneously served on a 100MB link. So clearly this kind of link is capable of transmitting the co-ords of 30,000 comparatively static tree's.

Hi Baff1

I agree what you said about CPU load.

The thing is most people do not know the difference between client side and server side computing and why some things such as graphics can all be carried out on the client where as the position of the things that graphics applies to may have to be carried out on the server or passed to the server for dissemination.

As to your Eve Online data I think you will find that as with any game including ArmA that the concept of if a tree falls in the forrest and no one sees it applies in games.

Essentialy if Client {A} is in a place outside the view of client {B} then client {B} only gets occasional updates about client {A}

So I think you will find the data from a client interacting with another is updated more often than one who is not in the same star system. Messaging is interacting too and would cause data to be updated more often.

These are all using up a limited resource. So much of what netcode does is balance the needs for information to reduce the information from entities and objects outside "Viewdistance" as it were while increasing it from those close to each other.

You also have to consider what information should be broadcast to many without confirming reciept versus what has to be confirmed as recieved. Both use different packet methods.

You also have to consider with broadcast to many the need for sanity checking. eg client {A} recieves data that says client {B} is moving between location x and location y but client {A} has it recorded that client {B} is dead because client {A} shot him. This has to negotiated between all clients and the server is client {B} dead or not dead?

If on the other hand everything is confirmed as recieved before being enacted you can see the obvious potential for lag increasing as number of clients increases and lag spikes happen.

For the record I think you will find ArmA deals in far more than 30,000 entities. You are forgeting projectiles which have real ballistics. In a big fire fight they quickly become the majority of entities in a game.

The key factor is that a lot of netcode is purely checking and correcting.

Kind Regards walker

This looks amazing! But will it be? I don't know.

Hi Hellfire257

As I pointed out some of the videos are prerendered at real FPS of 1 to 2 Frames Per Second. So esentialy fake.

If they turn off most of the physics I think it could be a real competitor to VBS2 to VBS3 not so sure.

Kind Regards walker

Surely you could allow for the synchronization of some things? ArmA 2 doesn't perfectly synchronize anything outside of the players view distance, go on to a MP server and the discrepancies between clients can be several hundred meters with fast moving vehicles.

It's already visible on a lightwave physics engine (RV), I can't imagine desync issues after 5 hours of random destruction on Warfare when a client JiP while at the same time a tank is flattening a forest, being chased by a chopper also flattening the area with rockets (ie, concurrent sources of status changes on a uber real environment where event precedence will change calculation output... on PC A, tree 1 was 1st taken out by tank, on B, it was a rocket blast ....), all this needing resync over resync while randomPlayer150 is JiPing ....

Again, it's hell on engine with light physics....

With CryEngine? I'm not sure I want to see that :D

The problem is always the same : BI has done something around scale (huge maps, huge number of objects, huge number of entities interacting, huge hours of play possible) originally. That's why RV is somewhat resistant, despite its numerous faults. They took the extreme limits into account.

Take an uber engine, push it to BI's limits.... my bet is that result won't be what you expected :)

Again, it's hell on engine with light physics....

With CryEngine? I'm not sure I want to see that :D

Have you ever played a CryEngine game in MP? The physics is disabled :) it's a SP feature only. In the current CryEngine builds anyway.

Hi Hellfire257

As I pointed out some of the videos are prerendered at real FPS of 1 to 2 Frames Per Second. So esentialy fake.

If they turn off most of the physics I think it could be a real competitor to VBS2 to VBS3 not so sure.

Kind Regards walker

I think the videos were prerendered just for smoothness. The stuff they are displaying inside the video is not hardware intensive. Any computer can definitely display that at an acceptable FPS. Look at ArmA II for example. If it was prerendered, why does the FPS fluctuate in these videos?

SYUrbOvzlsQ

6gjoRuOEMv4

(notice the FPS drop when there are explosions)

If it was prerendered, it would be at a stable FPS at all times. These videos show otherwise. You can clearly see the FPS drops.

---------- Post added at 04:44 PM ---------- Previous post was at 04:08 PM ----------

Hi Pulverizer.

Statement of Fact

If the same physics based simulation calculations is carried out on those 20 clients each of those clients will see those trees in a different place.

Explanation

Even assuming each of the clients was running on the exact same console design, Heck even if you run your physics based calculation on the same computer 20 times this is so! Inconsistencies in the chip, timing, heat, the simple fact that client {A} turns left but it takes time to pass that information to all other clients, and indeed the very fundament internal workings of standard Newtonian physics make for microsecond differences, that a physics based calculation then multiplies thousands of time per second with each interation of the calculations.

http://en.wikipedia.org/wiki/Classical_mechanics

Result

So in case {A} you will have tree trunk in front of you to hide behind but in case {B} there is no tree trunk to hide behind, in case {C} the client sees client {B} squashed by the same tree trunk etc. multiply by number of clients, but worse each error builds on the last in a cascading set that means by the smallest of differences a storm of differences between each client results. That is why it is also called the Butterfly Effect.

http://en.wikipedia.org/wiki/Butterfly_effect

Effects

It looks pretty but the internal inconsistancies destroy it as an MP method.

Like I said look up "Sensitive Dependence on Initial Conditions"

Kind Regards walker

That is why there is something called serverside and clientside.

A client is an ArmA II player who is playing on a server in this situation. A server is a computer running an ArmA II server that allows people to connect to it.

Clientside is something that is simulated on every computer independent of other compuers connected. Your view distance, sound settings, your reloading animations(Other players cannot see you reloading in ArmA II!).

Serverside is something that is synced across ALL clients from the server. Examples would be locations of players, locations of vehicles, location of trees, health of players, what weapons they are carrying, if they're wearing night vision(Players can see other players wearing night vision) in ArmA II. These things are very important that should be synced across all clients. The problem though is that since it's serverside, the dedicated server has to do the work, putting load on the server and potentially lagging everybody.

In your tree scenario, this can easily be remedied if physics were server side. This means that the server is calculating the physics. If a tree falls down, the server will constantly send the information of the tree falling down to ALL players. So all players should see the same position of where the tree is at. The only downside to synchronizing physics is that it causes strain on the server calculating it. Physics in general require a lot of processing power to calculate.

KEEP IN MIND THOUGH

ArmA II synchronizes vehicle physics, which means that the player can see where the vehicle is at the exact coordinates of where all the other players can see it. ArmA II already simulates physics. Same with Cryengine, same with Source engine, same with most online games out there that have physics. It is perfectly doable to simulate a tree and have players see it at the exact location as everybody else. The problem is though, how will a server handle many trees blown up at once without lagging? ArmA II remedies this by having trees not being simulated by physics but by a simple animation. Crysis can definitely replace physics on trees with animations falling down like ArmA II.

I just wanted to get this point across because you do not seem to understand server networking. Any server can process physics and distribute the data across all players. You just need a strong server to do that though. It is VERY POSSIBLE to have physics on a server that is synchronized with every player.

Heres some more info about it:

http://en.wikipedia.org/wiki/Client%E2%80%93server_model

An example of a game with serverside physics

Counter Strike Source:

rROewTU9zvg

Notice how the physics of the vending machines and couches are synchronized across all players. The zombie player cannot walk in because it is blocked by the vending machine and couch.

Hi Cookieeater

Most of my argument was against expecting client-side physics to work in multiplayer.

If you look back through my previous posts you will see that I was explaining the reason's for server-side and client-side operations and why much of the physics will work well in SP but is impossible in MP unless the physics is kept simple. Highly intensive and complex physics such as ragdoll is near impossible to synchronise across more than say 16 to 20 players in an Internet based game. The bandwidth does not exist. While a computer passing it around on its main bus can cope the Internet cannot.

Any game that relies on client-side physics is not going to work.

From reading your post I think we are on the same page.

What I was also pointing out in that post you are quoting was that some of the videos with complex physics in them had been achieved by slowing down the game to 1 to 2 frames per second then speed-ed up to look realistic.

The developers were themselves honest enough to say so in the video but some people in this thread were not reading the fine print and implying it was what players would see while in game play. The Developers themselves were careful not to claim that.

It is I agree a trick also done by ArmA Machinima makers to create smooth detailed videos as well.

Incidentally the tornado was first done in OFP nearly a decade ago.

As to your theoretical server-side physics, it could be advanced by having either several cores or a separate computer in close proximity devoted to the physics while other cores deal with net-code, but the bandwidth requirements would be enormous. Something like the actual bus speed of a computer so say 3200MB/s. Most people are considered lucky to have 10MB/s and only those places like scandinavia have speeds of 100MB/s. While much could in theory be multi-cast I think in reality a lot would be uni-cast so bandwidth would have to be multiplied by the number of clients in order to parallel uni-cast or resort to some kludge of micro serial casting perhaps.

Such a system using server-side physics would be beneficial but I think the net infrastructure would need to be improved beyond what is physically possible today.

Kind Regards walker

A scenario where I can see Cryengine's physics being essential in a military simulator environment are the vehicle physics. ArmA II's vehicle physics are absolute garbage. They're dampened heavily, collisions are garbage. Crysis was a game that had jeeps that felt like jeeps instead of jeeps feeling like super weak tanks that halted everytime it collided with something.

This isn't going to be going up against Arma 2. This is going up against VBS. And according to there linkedin page they only have 6 employees so I'm guessing they wont be going comercial at least not for sometime or without someone else doing it for them like BIS and BIA

Truly immersive :D

Have you ever played a CryEngine game in MP? The physics is disabled :) it's a SP feature only. In the current CryEngine builds anyway.

EDIT : nvm, I wrote too fast :)