88 replies to this topic

[MeiSooHaityu]

Star Citizen is Client Side Hit Detection?!? Oh, that's bad. No no no no.

[KraftySOT]

That Dustbowl Vulture video makes me weep with joy.

Woods.

Woods man.

Makes me want to jump into them and make a PSR.

[KraftySOT]

MeiSooHaityu, on 11 March 2015 - 10:24 AM, said:

Star Citizen is Client Side Hit Detection?!? Oh, that's bad. No no no no.

But it has to be verified once it leaves your "instance".

Lets say me and my 5 buddies are in my ship. Were in two instances. The ship, inside, and the ship, outside in space. When the damage is done by another ship, and sent to me, in my inside the ship instance, its verified to make sure you didnt cheat.

Fighters can apparently cheat each other, or in ship fights. But thats so far down the line they think theyll have it figured out by then.

I have no idea. I thought the whole thing was a scam, and I occassionally check it out at my buddies house who dumped a ******** amount of money into it, and it looks...neat...but not nearly as big or cool as Elite or X, and not as functional as Eve.

Blows my mind Eve has its own engine....RSI does not?

They...made the engine....for....WC1-3. Wut are you thinking Chris?!

[MeiSooHaityu]

I know what your saying, but you have to wonder exactly what the threshold is. The server may check for ridiculous instances that shouldn't be possible in game, but small aim or assists that could theoretically be possible goes unnoticed by the server.

I'm just not a fan of any game that relies on the client(s) to tell the server what's going on.

In all honesty, if we had a half decent internet infrastructure (Their 2) in the US (and maybe Canada too) that wasn't old, out dated, and made for dialup, connections would allow a purely server side validation system to work with very little issue. Unfortunately we are stuck with unmotivated IP services that have monopolies within their region (stifling competition and keeping infrastructure upgrades to a bare minimum).

Edited by MeiSooHaityu, 11 March 2015 - 10:41 AM.

[Kristov Kerensky]

KraftySOT, on 11 March 2015 - 10:17 AM, said:

Here you go:

https://forums.rober...o-the-cryengine


Notice where people, and eventually a dev says that MWO is server side hit detection, and SC is client side. Also notice where its not an issues specifically to the Cryengine, it happens in Planetside 2 as well. With the Forgelight Engine (completely different)

A large part of the problem for clientside is that the US/Canada ISP service is AWFUL and drops packets like crazy.

Though CIG and RSI are working together to fix those issues if it is actually the Crytek engine.

Wheres CIG and PGI are...uh... not.

Heres some meat and potatoes discussion:

https://forums.rober...wyer/p1/#Item_0

First off, SC is using hybrid netcode, mixture of client and server auth, and it's custom made. They are also using CryEngine 4 currently, actually 4+, they started with 3+ and CryTek has actually incorporated some of the CIG updates into CryEngine itself.

CIG, you DO realize that's actually just the company Chris Roberts started that's behind everything at SC, right? RSI is just the PR side of CIG, they aren't actually different companies, it's one big team, and they currently hiring for a lot of positions, all for SC. They've done some amazing work, graphically at least, but outside of that, I'm not seeing anything innovative or impressive so far. Networking isn't all that awesome to date, probably the biggest complaint next to missed deadlines that my friends who've bought into SC have.

And Krafty, you are rather off about the 'instances' reference, it's not what you describe. You, your ship, and everyone inside your ship will all be in the same 'instance', so will any other ships, their pilots and possible crews. Thing is, the CryEngine netcode, which CIG is still using, doesn't do well with more than 16 individuals in a single 'instance'. An 'instance' is the game world you are in at any given time, it's the match you drop in for MWO for example, that's a single 'instance' also known as 'shards' for other games. CIG isn't have a lot of luck getting that netcode to work as they need, much less want, so far, but they are working on it. I personally wouldn't hold out any great expectations about the former CryTek dev working for CIG, since the netcode in CryTek has ALWAYS been horrible, that's simply a fact about the CryTek engine, it comes with less than good netcode, see ALL Crysis games for reference. That dev is probably working on other aspects of the game, I'd hate to think they'd hire someone who's proven track history with netcode is the Crysis games, really?

EVE has their own engine because no one make anything that did what they do, simple as that. It's why DiCE makes their own engines, no one else does what they did originally, still doesn't. DiCE has never sold or licensed any of their engines either, wish they would, Frostbite 2 would be awesome for MWO...and SC for that matter, since it already does what both of them want and need to do.

[MauttyKoray]

FupDup, on 10 March 2015 - 08:55 PM, said:

It's also just a somewhat pointless mech in general, being not much more than a laser Locust clone. Even if the engine issues got sorted out there would be more worthwhile lights to add instead of the Flea.

We're getting an Urbanmech...this argument of yours is paper thin. Honestly though, I'd rather have as many mechs as possible even if some are similar or overlap in design/roles. The more mechs we get, the more varied a battlefield will be.

[Fate 6]

stjobe, on 10 March 2015 - 10:57 PM, said:

FTFY.

It's not CryEngine's fault, CryEngine doesn't have a speed limit of 170 kph.

MWO perhaps does, but if so it's due to PGI adding server-side authorization and HSR to the engine.

In short, it's PGI's crappy code that's to blame.

No, the engine is definitely limited. Sure, PGI hasn't handled it super well, but unless you have a crazy budget like Star Citizen (which uses the same engine) it's very difficult to fix the shortcomings of the CryEngine. Also, with the way Star Citizen plays it's nearly impossible to tell if you are missing due to a real miss or due to netcode.

[FupDup]

MauttyKoray, on 11 March 2015 - 12:46 PM, said:

We're getting an Urbanmech...this argument of yours is paper thin. Honestly though, I'd rather have as many mechs as possible even if some are similar or overlap in design/roles. The more mechs we get, the more varied a battlefield will be.

A slow light is at least somewhat more distinct than a nearly exact Locust copy.

Also, the Urbie will typically have more firepower than the current IS 30 ton Spider even without quirks, which can be used to set it apart. Also higher weapon mounts than the Spider and many other lights. And it's just more iconic in general.


I'd rather fill the game with mechs that are actually useful in some capacity than just throw in a bunch of ghettomechs. Adding ghettomechs is only a short-term variety increase, not long-term. Once people finish grinding out their preorder robots, only the truly good ones keep getting played afterwards.

As an example, let's look at the Kintaro. Did you remember that such a mech even exists in MWO in the first place? I actually forget about them entirely quite often. They're like unicorns. So are Gargoyles, Commandos, Adders, Vindicators, etc...

Those mechs don't add variety, because almost no one uses them. A mech can't add variety if it doesn't get used. Mechs need to be at least somewhat good to get used.

Edited by FupDup, 11 March 2015 - 01:05 PM.

[Mirumoto Izanami]

FupDup, on 11 March 2015 - 12:59 PM, said:

As an example, let's look at the Kintaro. Did you remember that such a mech even exists in MWO in the first place? I actually forget about them entirely quite often. They're like unicorns.

My Kintaro weeps...

[Deathlike]

Repeat after me:

The Lolcust lied, the Flea died!

[Mawai]

Koniving, on 11 March 2015 - 09:53 AM, said:

Try MW:LL. Same yet older engine, far better code.


I believe that's what they were implying. I also think that opinion sucks balls because that basically means "We don't want 20 ton mechs because they aren't meta in PGI's bad design." What that clearly means is PGI needs to redesign their game to allow such things to be viable. FLD and pinpoint convergence seem like perfect places to start. Small target + slow convergence = viable 20 tonners -- until they make a mistake.

But that's exactly how it should roll.

"Try MW:LL. Same yet older engine, far better code."

How many times do people have to say the following before folks actually read/listen?

MWO is server side authoritative

MW:LL is CLIENT side authoritative

The two games could be on the exact SAME release of the exact same engine ... and MWO would STILL have potential hit registration/HSR/netcode issues BECAUSE of the limitations of server authoritative implementations. Server authoritative prevents MOST hacks and cheating at the expense of more complex network interactions. If you trust the clients then the games get MUCH easier from a server perspective ... every client has PERFECT hit registration ... it has no need for host state rewind. On the other hand, folks can easily hack the clients to get perfect aim, register hits through walls, have perfect intel on where every other mech is on the map ... the more information the client has the more easy it is to hack and the more authoritative it is the easier it is to cheat.

You just end up with a whole list of different problems and complaints in the two cases. However, there is hope that hit registration and HSR can be improved at least for reasonable mech speeds ... in a client authoritative world there is no way to prevent cheating.

[CDLord HHGD]

Mawai, on 11 March 2015 - 07:23 AM, said:

Huh?

So you are suggesting slowing every mech down and then just giving it a different number? If the mechs cover the same distance relative to their size in a given time frame then you run into this issue no matter what number you assign to the speed..

Mechs have a specific size. Terrain has a specific size. In meters. Speed is how fast the object moves relative to its size and the terrain size. If a mech is 10 meters tall and the mech travels its height in one second then it is moving 10m/s,

I agree that they would have fewer issues with latency if they slowed down every mech in the game ... but then every mech would move slower. A Jenner would not do 150 .. it would be doing 130 ... even if you decide to call it 150.

This is exactly what I suggest. Scale is so borked anyways, who is to say otherwise?

The big issue is our speed is the only true counter we have to our ability to aim (versus TT's dice rolling). Slow everything down and it'll be easier to hit. Though I have seen here where crappy HSR can be a good thing...

[Koniving]

Mawai, on 12 March 2015 - 04:06 AM, said:

"Try MW:LL. Same yet older engine, far better code."

How many times do people have to say the following before folks actually read/listen?

MWO is server side authoritative

MW:LL is CLIENT side authoritative

The two games could be on the exact SAME release of the exact same engine ... and MWO would STILL have potential hit registration/HSR/netcode issues BECAUSE of the limitations of server authoritative implementations. Server authoritative prevents MOST hacks and cheating at the expense of more complex network interactions. If you trust the clients then the games get MUCH easier from a server perspective ... every client has PERFECT hit registration ... it has no need for host state rewind. On the other hand, folks can easily hack the clients to get perfect aim, register hits through walls, have perfect intel on where every other mech is on the map ... the more information the client has the more easy it is to hack and the more authoritative it is the easier it is to cheat.

You just end up with a whole list of different problems and complaints in the two cases. However, there is hope that hit registration and HSR can be improved at least for reasonable mech speeds ... in a client authoritative world there is no way to prevent cheating.

Edit: I can't even English today.

I am well aware. I used to explain the limitations of a server side authority to new players. It has limitations and only a fool would think otherwise.

However many of these limitations are those that we would not need to face had PGI designed the game more in line with the source material.

• We'd only ever have two mechs capable of going fast enough to break 200 kph until beyond 3060 and that's almost exclusively because of the effects of Speed Tweak,
  
• the limitation of lack of movement speed adjustment while in water is NOT due to the server side authority but because PGI has not included that into the code,
  
• the limitation of jumpjet-related hit registration is because the server does not have enough time to anticipate and correct for changes in Z-axis -- which it would if PGI would correct the root problems (sudden Z-jump in animation, spammable jumpjet controls registered in milliseconds; far too quick for the HSR to register,
  
• EXTREME gravity. Game's gravity is 2.8 to 3.2 times greater than the engine is meant to use which means rapid changes in upward to downward Z-axis movement).
  
• Adjustments (reduction) to threshold would both make single heatsinks viable (as 10 SHS could cool a mech from 50% in 15 seconds instead of 22 as it does currently), as well as make VAST improvements to hit detection due to a significant reduction in weapons fire ray-trace calls for attacks (as well as allow 20 ton mechs to be viable due to not being slaughtered in single shots).

The list can go on and on. In regards to environment hitboxes, this is an issue due to the extreme sizes used for the mechs (which many are far larger than canon -- the MWO Centurion [14.6 meters] is larger than the largest mech of 3055 [Executioner, 14.4 meters] and dwarfs the Behemoth and the BT Atlas, as well as the even taller BT 1987 Marauder in height!) Smaller mechs, smaller terrain, less 'issues' with making 'larger than life' terrain features. This isn't something PGI could fix from a business perspective; this particular ship has sailed in 2012.

There are issues with the setup, but the setup itself is only partly to blame. Many of the issues ARE possible to overcome but they require changes that have been frequently requested but have yet to be done because there are mixed feelings from the community and/or the devs about what those changes would bring or the cost of doing such changes.

Edited by Koniving, 12 March 2015 - 07:24 AM.

[NextGame]

by the way, im not reading 4 pages of fleaness, however, I hope that spaceships in star citizen go faster than 171 kph?

Edited by NextGame, 12 March 2015 - 07:19 AM.

[CDLord HHGD]

NextGame, on 12 March 2015 - 07:19 AM, said:

by the way, im not reading 4 pages of fleaness, however, I hope that spaceships in star citizen go faster than 171 kph?

Maybe 171 spacemiles per hour?

[Jetfire]

NextGame, on 12 March 2015 - 07:19 AM, said:

by the way, im not reading 4 pages of fleaness, however, I hope that spaceships in star citizen go faster than 171 kph?

Fastest ships hit 1080 kph or 671 mph. Mostly to keep the G forces while dogfighting reasonable.

[NextGame]

Jetfire, on 12 March 2015 - 07:53 AM, said:

Fastest ships hit 1080 kph or 671 mph. Mostly to keep the G forces while dogfighting reasonable.

So question for everyone regarding the flee and speed cap.

Either in Star Citizen, every ship will have an impenetrable lag shield wherever they go, and warp around dodging laser fire. Or is this engine limitation thing in MWO just an excuse that needs to go away?

I'm going to go out on a wild guess here and speculate that there are other cryengine titles where entities do indeed "gotta go fast" and don't suffer from the MWO problem.

Edited by NextGame, 12 March 2015 - 08:46 AM.

[Kristov Kerensky]

NextGame, on 12 March 2015 - 08:44 AM, said:

So question for everyone regarding the flee and speed cap.

Either in Star Citizen, every ship will have an impenetrable lag shield wherever they go, and warp around dodging laser fire. Or is this engine limitation thing in MWO just an excuse that needs to go away?

I'm going to go out on a wild guess here and speculate that there are other cryengine titles where entities do indeed "gotta go fast" and don't suffer from the MWO problem.

Yes, there are other CryEngine games that don't have the speed issue, and they all use clientside authorization netcode and are all open to the many CryEngine based hacks that rely on that clientside auth netcode, we even had them appear in MWO before they got the serverside auth into MWO.

SC, people will tell you, no telling if you are missing because you MISSED or due to lagshields. SC uses a hybrid of client and server auth, and it's got problems just the same as MWO due to that.

[Ghogiel]

Koniving, on 11 March 2015 - 10:22 AM, said:

Furthermore, MW:LL uses lower-level code for things like the HUDs and interfaces, MWO uses Scaleform which Russ has recently admitted gives a larger-than-anticipated performance impact.

MWLLs HUD had all the same performance issues as MWOs HUD did, maybe not as bad as it was from before MWOs recent fps fix, but it wasn't good imo. It's all flash and newer scaleform SDK probably would have helped cut down some over head on MWLLs fps ****** HUD. Just saying it wasn't roses for the MWLL HUD.

Edited by Ghogiel, 12 March 2015 - 09:19 AM.

[Stoned Prophet]

Whatzituyah, on 10 March 2015 - 08:53 PM, said:

R.I.P. Flea we never got a chance to pilot you just because of the Cry Engine. Curse you CryTek!

Just to explain Flea has MASC as a standard on the mech due to how Cry Engine runs it cannot handle speeds over 170 kph very well so were stuck with mechs below that limit and its Cry Engines fault.

"Commandos yes I know one can do 171.1 technically higher then the limit but even that 1.1 has trouble in the game and ends up rubber banding."

Flea got turned into the Locust. That is all.