I’m looking to put together a history or time line of sorts for netplay when it comes to fighting games specifically and was wondering if I could get some help.
So far I’ve got it set up as XBAND being the first which was in 1994 which used the 28k modem. I also wanted to include emulators that had netplay such as Znes around 98-99. Then I got SegaNet being next in line in 2000 but of course this gets tricky because researching it so far I only know about Capcom Vs Snk 2 being available in japan only. The Kaillera client of course came around 2001 which I believe came before Xbox Live. Then you have 2002 with the PS2 which had some titles as well that were compatible.
I figure I’m pretty much covered after that but what do you think should be included? I have a feeling there may be some holes especially with PC and even Sega Saturn. There were a lot of JPN titles that got netplay we didn’t get over here in the US.
Any suggestions or direction you could point me in would be appreciated. I like to know about the experiences as well. I believe something was done here on shoryuken about Xband but I could not find the article.
Personally, I’d subdivide them into “generations” based on certain criteria.
Basically:
1st gen - everything before netplay became standard, so everything from XBand to stuff on PS2 and XBox. Basically the genre’s first experiments with network play.
2nd gen - games with netplay when it became a standard, but still on old, variable delay netcode. So just about everything during the previous console generation with variable delay netcode. We have network play, but the netcode still isn’t optimized to give a near offline experience (i.e. variable input delay). Also includes newer games released recently that don’t use rollback netcode (i.e. GGXrd, MKX).
3rd gen - games with rollback netcode (HDR, all of Capcom’s recent re-releases, SFxTekken, Skullgirls, Killer Instinct, etc.). In other words, netcode has been refined and we’ve figured out a better way.
Thanks for feed back guys I really appreciate it! I like that idea d3v definitely will something to consider especially when you think about the progression of netplay.
Something about ggpo works great on old fighters arcade cab releases . But not on new games or 3d according to harada tekken. Sf x tekken i think they tried to build there own. Some one needs to make it happen so mofoz can license it like unrealengine n shit… snk needs help fucking bad.
KI shows that modern consoles can handle GGPO netcode (and before anyone says anything, KI is running on what is GGPO in everything but name since Ponder was actually involved).
Yeah, we’ve been over it in other threads but 2D/3D doesn’t affect the feasability of rollback netcode. The factors are how fast you can run the update loop (ie one frame, one “turn” of the game system, graphics and effects/fluff not included) and whether said update is deterministic (ie “can you replay it”). For the speed, fighting games are really on the simple end of the game complexity scale and for that matter so are most arcade games (you could still use GGPO rollbacks with a bullet hell shmup), you’d need something like complex physical simulation affecting gameplay (and even then all modern FPS are based on a more flexible form of rollback, though one that induces more latency since it has a server in the middle). As for determinism, you need it for delay-based netplay too so that’s that.
The main factor is know-how, really, since it does require the devs to step up their game somewhat, it’s a lot less intuitive to program than delay-based, and though multiple articles exist you kinda have to be part of specific english dev communities to know they exist (especially for how to handle sound), so that’s the main obstacle for japanese developers. And since they’re still pretty insular and they don’t actually need to change their tech for the netplay to function in japan they’re naturally reluctant, out of convenience and/or ignorance.
And once you have a GGPO-like netcode it’s the best of both worlds since if you like input delay better you can just set it in the rollback options and it’ll work just like your pre-rollback games.
Harada says there is no such thing as netcode. His belief is that in order to have good netplay online, the game has to be developed from the ground up for netplay. He used the Tekken game as an example. Essentially, they tweaked the frames taking into consideration the input lag that happens online amongst other things. Funny thing is they did something similar with Weapon Lord but I haven’t found anyone who talked about their online experience with the game.
He’s right that the word netcode does get thrown out a lot in these types of discussions but its just a general term loosely referring to the quality of online play, when compared to offline.
This can be many things such as input delay (most notable one), slower gamespeed, visual distortion, sound problems etc. which is caused by how the developers built(coded) that part of the game. So in that context, yes netcode is something that exists.
Harada’s approach toward online play is unfortunately not the best solution to emulate offline-like gameplay.
Seriously he should just consult Ponder (like how Double Helix did with Killer Instinct) and save all that extra work tweaking the frames from ground-up or whatever to conpensate network latency.
Tweaking a game so that it takes into consideration online delay is fine, and in fact should now be the defacto standard.
However, the fact that something like rollback netcode does help reduce online input delay (by about half, considering that the simulation only has to account for one way data) is something that shouldn’t be ignored.
There’s a contradiction there, though: if the game has to be developed for netplay, then the type of netcode you’re gonna use fundamentally determines and constrains the kind of game you’ll be able to make. That’s why netcode is impotant even then. All those input delay considerations become a lot less important game-wise when your netcode is already designed to deal with it, giving you more freedom in designing your game.
Even better, GGPO-like netcode can be combined with input delay, in which case it’ll behave like standard input delay but instead the rollback frames will contribute towards hiding lag spikes (ie a 4f delay game will freeze if there’s suddenly a 6f lag but a 4f delay + 4f rollback game will go on as if nothing was happening).
*
Battle Arena Toshinden’s one saving grace is that it’s completely networkable, either through directly-networked computers or modem, so you can beat down people around the world. While the graphics and music to Battle Arena are great, actual play could be vastly improved.*
I vaguely recall weaponlord.
It was different in that I believe all the moves had you hold a button and then press the directions instead of directions then buttons.
It played slow in general so when you were on Xband it didnt feel as bad.
I remember when they added Doom for 1v1. It was so bad because there wasn’t an actual model for the player so they used the sprite from the end game credits, which was only ever facing you. Whether the opponent was facing you, away, to the left, etc, he was always staring at you.
In one way that was fairly in depth to me since it had absolutely no multiplayer aspect to the game by default and they patched it in.
@d3v help me out…
Why is it that, at least where I live games like Tekken and Guilty Gear which dont’ use rollback are so much smoother than the Iron Galaxy ports, for example? SkullGirls is nearly unplayable for me, but meanwhile I’ve had friends that have played from coast to coast. GGXrd gives me like 2 frame input delay with Chicago (I was playing ElvenShadow the other night, so I know and I also frequently play BlackSnake). Third Strike? I can’t finish a match.
That all said… GGPO client with ST seems to be fine.
It wasn’t teleports.
It was disconnects, not being able to parry anything, and rollback that would bring me full combos back. Once, I did a c.mk, lk tatsu, DP combo. I demon flipped for hte oki and it brought us back to before the combo even landed.
But typically, I’d have it as 1 or 2. I know better than to have it at zero.