Kayane interviews Katsuhiro Harada of Namco

  • Moderator
Last edited by a moderator:
If MK vs DCU is any indication, I don't want UE3 going anywhere near SoulCalibur.

MK9 was UE3 too, but a lot of the look of both those games was based on design decisions, rather than limitations of the engine.

I like bringing it up as a choice because there's no way in hell they're remaking the engine from scratch and a license is the next best thing :p

ASura's Wrath is in UE3 tho', and looks about right, for animation quality.

(gameplay vid, sorry for the terrible VO)
 
They understand, its just kind of a really difficult problem to solve technically. From the sound of it, they need to rebuild the Soulcalibur engine from the ground up (or do my version and license Unreal 3) to seriously fix these problems, and that's still a pretty big investment to make (altho one I'd like to see done).

no lag has nothing to do with graphic engine...

To keep in mind online is more a matter of balancing the game...
Example: LONG blockstun is less impacted by lag
Faster pace is also to a certain extent.

Removing grounded disadvantage would help drastically also...

Those may be not the best things to do for sulcalibur but are just easy examples of how they should face the problem.

For example do you know why Cod seemed almost lagless and halo seemed more afflicted by lag?
Because in cod people stand still campering for single shot kills, in halo mobility is predoinant and there is a shield so LAG impact the game more on halo.

Same for soulcalibur ...a 2K is less lag related than a 1a...that's because 2K is not reactable offline..
So they simply have to change moves like 1As for example...or give to grounded opponents less disadvantage giving them a way to counterattack.

Its just a matter of balancing the game keeping in mind the lag.

Its probably very hard to do but not expensive at all.

They instead already reduced blockstun for example and that is a thing we will pay with lag.

P.S: a graphic engine has really nothing to do with lag.
the network code can be adjusted just because in sc4 it was far superior to geographic latence...but even reaching the minimum value equal to geo latence it would impact too much a game like sc4 that is slow paced and centered on defense/steps etc.

P.P.S. i am in a hurry i hope someone can translate what i ve written in a decent english XD
 
The argument I've gotten is that there's a crapton of legacy code (some of it from Soul Edge even) and weird piled up stuff in Soulcalibur 4. Presuming that's true (and there's no reason for me to think they were lying), then its got to be a nightmare of spaghetti code anyways, with network coding thrown on top of it.

That's why they need to rebuild it, there's too much trash, and the network stuff can't really be properly integrated. Part of the reason I chose UE3 was because it already has fairly good netcode included in the package (also, its relatively cheap and very easy to use).

Lets see what else. I think designing the moveset based on expecting online lag is a terrible idea becasue it utterly rapes options in offline play (which is and should be more fun anyways).

Oh and you're wrong, the graphics (or physics) engine can definitely effect lag, as many require more information to be sent than just inputs. They also matter because they are related to how much of a consoles resources are used, which effects things like being able to use rollback for lag mitigation.
 
I think designing the moveset based on expecting online lag is a terrible idea becasue it utterly rapes options in offline play (which is and should be more fun anyways).

Excuse my radical thinking, but why should a game be balanced for something a huge majority will never experience?

How about balancing it for both? 1A could simply be slower online than offline, making blocking it on reaction possible. Fast moves could also have higher disadvantage on block and equally longer blockstun to make punishing them easier online.
 
Excuse my radical thinking, but why should a game be balanced for something a huge majority will never experience?

Well first of all, it effects the single-player experience as well. It also has a negative effect on gameplay for people working at relativly low levels of lag.

If you're designing movesets around lag, what ping do you aim for? 70? 150? It's a lot harder to hit a moving target.

Also, to be honest, you're messing with move properties (in general making reaction impossible across the board) for a relatively minor gain in gameplay at *just the right* level of lag. What the guy was originally saying was 'lets make Mitsu's 1A the same as his 2K so one isn't given a bigger advantage in a high-lag environment!' That just seems backwards and wrong minded to me in so many ways.
 
So you're basically saying: Ignore the majority. From a business perspective, that doesn't make sense you know :)

And who cares about balance in singleplayer?
 
So you're basically saying: Ignore the majority. From a business perspective, that doesn't make sense you know :)

And who cares about balance in singleplayer?

I didn't say 'balance' I said 'gameplay' :p They're slightly different things.

It's not about 'ignoring the majority' but about 'gain vs. losses'. You gain a very small amount for the very high lag players (or people that blame high lag) and make the game worse for low-lag players, offline players, and 1p players. To reiterate, there really isn't even that much gain... or much gain for that many people.
 
The argument I've gotten is that there's a crapton of legacy code (some of it from Soul Edge even) and weird piled up stuff in Soulcalibur 4. Presuming that's true (and there's no reason for me to think they were lying), then its got to be a nightmare of spaghetti code anyways, with network coding thrown on top of it.

That's why they need to rebuild it, there's too much trash, and the network stuff can't really be properly integrated. Part of the reason I chose UE3 was because it already has fairly good netcode included in the package (also, its relatively cheap and very easy to use).
That's a lot of speculation though. For all we know, one reason why they've been able to utilize the same code basis is because it's good. Or that they've made sure to keep expanding on the code in sensible ways without making it messy.

I did a quick check, and the lead programmers for the series has worked on Soul Blade all the way up to Soul Calibur 4. That bodes well for the code as there's less chance of people writing new/changed code in different styles.

I personally think they'd be shooting themselves in the foot by licensing UE3. It'd increase the budget greatly, they'd have to rewrite a ton of code instead of rewriting specific parts in SC4 code which would have to interact with the netcode, UE3 definitely doesn't ship with a netcode optimized for fighters, and out of all the billion UE3 games out there there's only 2 which are 60fps on consoles.

I'm gonna go with the simplest of reasons why we don't have a good netcode yet: laziness. Reviewers almost never complain about netcodes so there's no indication from the press that a better netcode is needed. Japanese developers mainly care about Japan, and the current bad netcode suffice over there. And there's little inspiration regarding netcodes from other fighting games. They can't go all "Hey, that Urban Melee IV has totally awesome netcode which makes everything lagfree! We'll just do the same as them!". Their only inspiration is GGPO, which is used primarily with emulators and is apparently hard to implement.

Although a better netcode is also required, I wouldn't mind if they tried to design the game as darkfender suggests. The game won't automatically become worse if they decide certain aspects of the game is slower. As long as they balance the game properly, it won't make competitive play any less viable. But well... that's definitely not the case as they've already confirmed SC5 will be a faster game than SC4. That'll inevitably make the lag feel even worse.
 
Well, I'd rephrase 'laziness' as 'cost-benefit analysis' but there is certainly a point to what you're saying. The problem is that there's only so much more advantage they can get before they hit a really brutal point of limited returns.

As to designing moves to reduce the effects of latency, I think Soulcalibur is actually already pretty good in that regard, not being that much involved in tight combo timing or multiple links. To be honest I think I just didn't like that guys example, which seems to be a very very backwards and even (as much as I hate to use this term, its loathsome) dumbed down way to do it.

In my experience its all the same, if you find people with reasonable connections and avoid certain things (large lobbies for instance) its perfectly fun and playable online (and I'm not in a huge city either), and I also enjoy playing with a few friends at my house. Unless you're simply playing against people you shouldn't be, the lag situation simply isn't that bad that you need to redesign move properties to make up for it (in the last ~2 years anyways).
 
In my experience its all the same, if you find people with reasonable connections and avoid certain things (large lobbies for instance) its perfectly fun and playable online (and I'm not in a huge city either), and I also enjoy playing with a few friends at my house. Unless you're simply playing against people you shouldn't be, the lag situation simply isn't that bad that you need to redesign move properties to make up for it (in the last ~2 years anyways).
This depends drastically on where you live. I'm in Norway, and finding a match against someone who also lives in Norway is practically impossible. I played SC4 a fair amount online when it was first released, and I think I only met one Norwegian guy. I played against a fair amount of UK people, which was technically playable but it was still laggy enough that I kept mistiming moves and the lag was bad enough that it was literally impossible to react to really fast moves.

So yeah, for me personally the online mode in SC4 might as well have not existed. It started giving me really bad habits for offline play too, like literally holding the guard button all the time when not attacking (since blocking on reaction is damn hard online).
 
This depends drastically on where you live. I'm in Norway, and finding a match against someone who also lives in Norway is practically impossible. I played SC4 a fair amount online when it was first released, and I think I only met one Norwegian guy. I played against a fair amount of UK people, which was technically playable but it was still laggy enough that I kept mistiming moves and the lag was bad enough that it was literally impossible to react to really fast moves.

So yeah, for me personally the online mode in SC4 might as well have not existed. It started giving me really bad habits for offline play too, like literally holding the guard button all the time when not attacking (since blocking on reaction is damn hard online).

See that's brutal. Don't know how I'd deal with that. (probably by not playing much online lol)

It kind of highlights the problem of designing to lag tho', do you design the game to Norway/Australia lag, or average US lag, or East Asia Lag?
 
Well, I don't think they'd have to think too carefully about different regions. Just make the best netcode they can. If they're able to make matches between US and Europe feel nearly flawless, then matches with lower ping (like any matches within US) should feel even better.

Maybe some insane scenarios might be nearly impossible to make playable though, like Europe to Australia. I think you're dealing with pings close to 400 at that point.
 
The argument I've gotten is that there's a crapton of legacy code (some of it from Soul Edge even) and weird piled up stuff in Soulcalibur 4. Presuming that's true (and there's no reason for me to think they were lying), then its got to be a nightmare of spaghetti code anyways, with network coding thrown on top of it.

That's why they need to rebuild it, there's too much trash, and the network stuff can't really be properly integrated. Part of the reason I chose UE3 was because it already has fairly good netcode included in the package (also, its relatively cheap and very easy to use).
i didn t read the thing, but its definitely a problem they had to face on sc4...
If the game calculates the events in a weird non optimized way, its not that expensive, nor dificult to fix it...
Done that there cannot be any problem to adapt the game to a netcode....


Lets see what else. I think designing the moveset based on expecting online lag is a terrible idea becasue it utterly rapes options in offline play (which is and should be more fun anyways).
This is a classic reply....
Well its not like that....its just more work to do to do a balanced gae that works both offline AND online, but avoiding that works means just reasoning on the old generation of console...

Believe it or not but ONLINE now is most important than offline.
Where would've been cod without online?
and Halo?
same for many games....


Oh and you're wrong, the graphics (or physics) engine can definitely effect lag, as many require more information to be sent than just inputs. They also matter because they are related to how much of a consoles resources are used, which effects things like being able to use rollback for lag mitigation.

no its definitely not how the netcode works...
networks completely ignore graphic....it may be slightly affected by things like physics that is completely inexistent in soulcalibur, but the graphic isn t involved at all....
The communication between the 2 hosts are made of parameters like:
Player A does move Y, request hit confirmation etc ......
Even with cas it just send the formula before the match begins and then is calculated on each client.

I had experience on setting up mmorpg servers and i know how netcode works...
I saw something also on a fighting game >.> and a fps but lot of time ago ..
 
How about balancing it for both? 1A could simply be slower online than offline, making blocking it on reaction possible. Fast moves could also have higher disadvantage on block and equally longer blockstun to make punishing them easier online.

It would be like playing two different games. Everything you learn from offline would be worthless online and vice versa. That wouldn't work very well I think.
 
Darkfender, seriously you're repeating a common myth, but I've spent a large amount of time researching this subject, and talking to honest-to-God experts, and you're mistaken.

The modern physics engines use random factors for a number of things including things like sparks and sometimes character animations/reactions (altho I'm feeling pretty safe to say that the SC engine isn't so advanced as to do the latter), and all those things have to sync. I'll admit the SC4 engine is pretty old, so it might not use some of those techniques, but saying it can't is just dead wrong.

I'm pretty sure that MMO servers work in an entirely different way and have a whole different set of concerns as compared to fighting games :p

Edit: And also I know its a classic reply, you don't know how much it pained me to have to make that argument. Nevertheless, it was the appropriate one to make :p
 
In my experience its all the same, if you find people with reasonable connections and avoid certain things (large lobbies for instance) its perfectly fun and playable online .

also lobbies shouldn t hinder the gameplay as long as players have a decent bandwidth...
another thing that makes clear how sc4 netcode sucks:

chat has a separate reserved bandwidth with lower priority....
but the game sometimes set up a huge latence when the lobby is large O___o something like 1500 msecs on a 200km distance that usually should ping for 30....

but in game lobbies doesn t change anything instead

This because lag can come from arious issues...but the ONLY that should has to be dealt with is dictated by geographic distance.....(see ping)

Anything other can be solved.

But sc4 is way from being perfect...with a ping of 20 ms (1,5 frame) with a friend near me, we lost about 5 frames O__o and that was the best connection i could find.
 
Darkfender, seriously you're repeating a common myth, but I've spent a large amount of time researching this subject, and talking to honest-to-God experts, and you're mistaken.

p

no really you miss knowledge bout those things, i worked with them...i am explaining how those works....obviously omitting lot of things...

But NO WAY a graphic engine has anything to do with netcode...and that is easy enough to understand even for people who doesn t know a lot about it...

The only chance is like when for example in game like crysis when you shoot down a tree and things like that....

but youy can have photoreality and the netcode would t be impacted as much as with an unreal 1 engine.

AGAIN excluding PHYSIC......

If you want we can move this discussion on forums liek b3d or anyone else
 
Heh, there's really no reason to bother. You're not interested in discussing anything. You're just trying to show off your 'I totally set up MMO servers!' knowledge.

It'll just go into a loop of you saying 'there's absolutely no way!' and me saying 'well actually there can be' and you saying "NO THERE ISN'T I SET UP SERVERS ONCE DAMMIT!'

So, have fun.
 
trust me you are denying a base concept of ANY netcode....of ANY game.....
You can research the thing for yourself if you want and you'll easily find the proofs...but only if you care to understand for yourself instead of saying "the brother of a friedn of my grandma sai me he hknew a guy who is a guru ....."

That thing is easy enough for anybody to understand with little research.
So they are facts.
 

Live streams

3 Viewers
Diabro420
Diabro420
FGC Fridays Warmup Session : SC6

Forum statistics

Threads
14,897
Messages
676,634
Members
17,201
Latest member
mrdaniyel
Back