Kayane interviews Katsuhiro Harada of Namco


darkfender: No graphics or physics stuff would ever be transmitted over the network, but they would still have to be considered as the netcode affects the gameplay state. For instance, if a character is moved from A to B as a consequence of the netcode trying to keep things in sync you wanna make sure it looks correct. I personally think the cloth physics would be the biggest potential problem as that's entirely dynamic and it'll look very funky if it suddenly snaps to a weird position.

I assume that's the kind of stuff windie is talking about.
@sectus forget this i read wrong..yet i'll leave the explanation...but i don t think cloth destruction will ever be dynamic in any game O__o of this generation of consoles

its not like that...how could a single xbox manage to serve a 8 player match >.> if graphic was involved (and halo has even a decent physic engine)

i'll try to explain in a simple way without recurring to examples of netcodes...

the problem is not bandwidth but SPEED....
And speed is a porblem than in a poor netcode is hindered by consistency checks for example and other things....

The problem is easy....
There is a base communication time that can be verified with ping command. (from point A to point B)
That is the MINIMUM lag possible.
latence value on the best games is really near the ping time.....and this is eve for games with great graphic as well as for older one.

But those systems often doesn t have the problem a fighting game has i.e. balance for both players.

An atual FPS uses hosts system where a player (or a server) is elected as MASTER....

This master has ZERO LAG, and have to manage communications from all the hosts machines.

All hosts instead have direct la acording to ping (distance)....so some player has far less lag than others.

A Fighting gameneed another system to have same lag for both:
So basically both game are servers and decide a fixed lag time for both to continually sync datas.

This result in something at least 2X the ping time but could be reduced with some tricks (see GGPO) but this has a price.....
And that price is a less reliable synchronization and various issues.

In sc4 is something like 3 or 4X >___>;

AS A PROOF...there are tons of custom fighting game for PC...
You can see examples of netcodes and grasp few concepts even without being a programmer....(basics helps though).
Again, I'm not saying graphics is directly involved, it's just something you'd have to consider. With a game like SC4 where you always see both characters very clearly, I think any graphical oddities would be extremely obvious. I think the developers would have to keep that in mind to reduce whatever graphical artifacts is shown as result of resyncs.
Sectus, I'm from Norway too, but haven't experienced the same laggyness that you seem too. I get 5 bars to most people in Sweden, Germany, Poland and Benelux, and 4 bars to France. It's hardly anything like offline, but it's playable. Of course, there is that unwritten rule to not use "lag tactics". Without it, the fun would be totally gone.

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.
I can see how making some moves slower only online will affect the game, so that's a bad idea. However, equally increasing blockstun and disadvantage on block will not affect anything. It will be like the slight pauses in Street Fighter when someone hits or blocks.

Let's take an example you're probably familiar with :) I do 66B+K which is fast, unsafe as hell, yet very hard to punish online. The framedata says GRD -20, which might mean the recovery time is 30 and the blockstun is 10. I'm just making that up, but for sure the difference is 20. By equally increasing both the recovery time and blockstun to 40/20 the framedata will still say GRD -20, but you'll have 167ms additional time to react.
Maybe I was just unlucky with the matchmaking then. 5 bars was a pretty rare sight for me.
it depends even from other factors....
Talking on a lower base, data encryption protocol heavily impact the speed:
on good connection its used a system called fastpath
on standard connections its used a system called interleaved.
exA to transmit 1234:
1234 1234 (same packet 2 times to prevent errors)
1111 2222 3333 4444 (each number many times to prevent errors)

in my country you can ask ISP to use fastpath if the connection is stable enough...
obviously it reduces lag a lot.
sorry again for my english :/
I doubt there's something wrong with my connection. I just did some pings to a few Norwegian servers now and I'm getting pings of around 20ms, which sounds low to me.

I honestly haven't played SC4 online in years now so I dunno how the matchmaking would be now. Even if it's a lot better now and I'm mainly finding people close to me, I still think it's super important to improve the netcode.
its not "something wrong" they are 2 standards for ISPs....
not many customers requires a fast ping over a stable line....obviously most swedish connections are so flawless and fast they are setup as fast by default but in most other EU countries interleaved is quite frequent...and the change of protocol is quite easy for any ISP if requested.

If you happen to contact your ISP just ask them wich you have....they will easily answer you.
I did google a bit about fastpath and from what I found people with interleaved would get higher pings. Even if I'm not on fastpath, I'm happy with the pings I'm getting.

We should probably go back to talking about netcodes than ISPs :-P

I'm starting to get curious how many people are satisfied with the current netcode in SC4, and how many people won't play online without a better netcode.
I did google a bit about fastpath and from what I found people with interleaved would get higher pings. Even if I'm not on fastpath, I'm happy with the pings I'm getting.

We should probably go back to talking about netcodes than ISPs :-P

I'm starting to get curious how many people are satisfied with the current netcode in SC4, and how many people won't play online without a better netcode.
The current netcode would be defined by most folks as 'notoriously bad' >>
I just asked my ISP whether I had fastpath or interleaved. They had no idea what I was talking about =) I got 13ms ping to uio.no (which is 600km away) so I'm not really concerned about it.
aside that thing....

There is a physical limit that can be surpassed and that is ping...

considering that you'd lose in the best scenario something like 4-5 frames it would still be a problem for a game like sc4

Nothing would change aside 50% of lagmoves being easier to react but most 1a would still be deadly....and lagtraps still viable (even if reduced in number).

So the point is:
-they change how the game works trying to adopt something good for both online AND offline.
-they reduce lag with a better netcode but the problem will be still there.
there is no GGPO or any stuff that can help to have a decent online experience even remotely similar to offline.

and as i said in 2012 no matters what trolls says but a game SHOULD really be thought with online in mind.
(that has nothing to do with adding lag offline....but for ex they could change grounded game completely and increase blockstun time that would solve lot of problems)
I think there's some solutions they could implement to make it fully playable. There will definitely be side effects, but hopefully they can make those as trivial as possible.

I suppose the trickiest problem is when both players are in their idle animation and one player does a super fast move. At that point you'd have to do something out of the ordinary to let the defending player react as quickly as if it's offline. I wouldn't mind if they used a solution which might affect the pacing of the game (ie, if the attacking client isn't aware of the outcome, freeze the game temporarily) as long as it lets the defending player block/parry/evade properly.

Heck, turn the game into freaking Tekken Chess if there's no other good solution: http://www.youtube.com/watch?v=WsePqSFAJJg (okay, I'm not being completely serious with this one :-P)