Community Why is DOA5 not getting awards?

[d3v]

If that were the case, only visible lag would be during the match, then the replays wouldn't show the complete lack of reaction, no? Moreover, if you used either method, the input lag wouldn't be noticeable, only the frame issues would be. That would allow people to hold as normal (actually more easily because the game would inevitably pause when "fair state" rollback is waiting to occur), yet the inability to do things like that is the greatest criticism of DoA's online. Juggles that do not rely on delay also should never drop.

As @CyberEvil as pointed out, you obviously have no idea about how netcod works.

Whether or not the system is using variable delay/slowdown or fixed delay/rollback netcode, there will still be input lag. This is due to the need to make sure that both games are in sync. The amount of delay that's needed can easily be measured using ping times in milliseconds. If we take that every 17 ms added to ping time results in 1 frame of delay, then even a decent connection with 80 to 100 ms ping results in 5 to 7 frames of delay. Now consider that most variable delay/slowdown netcodes (like DOA) usually require a confirmation from the other peer, resulting in double the delay, meaning that for 80 to 100 ms, you get 10 to 14 frames of delay in total. Then there's having to deal with changes in the connection quality, which forces the system to increase the delay. Juggles are still affected because the game is waiting for confirmation from the other peer and compensating by adding delay as necessary.

Fixed delay/rollback style code, which only a few games so far use (Skullgirls, KI, Darkstalkers Ressurection) fares better since there's no need for confirmation from the other peer, so for the same 80 to 100 ms, you only get the 5 to 7 frames of delay since lag spikes and desyncs are handled by rolling back to the last fair state. That said, 5 to 7 frames of delay is still quite noticeable.

For more on this, you really need to read:
http://mauve.mizuumi.net/2012/07/05/understanding-fighting-game-networking/

 

[Kohlrak]

As @CyberEvil as pointed out, you obviously have no idea about how netcod works.

Not experienced in it, but i'm not entirely oblivious to it. I know what i'm seeing, as well as what i should be seeing.

Whether or not the system is using variable delay/slowdown or fixed delay/rollback netcode, there will still be input lag. This is due to the need to make sure that both games are in sync. The amount of delay that's needed can easily be measured using ping times in milliseconds. If we take that every 17 ms added to ping time results in 1 frame of delay, then even a decent connection with 80 to 100 ms ping results in 5 to 7 frames of delay. Now consider that most variable delay/slowdown netcodes (like DOA) usually require a confirmation from the other peer, resulting in double the delay, meaning that for 80 to 100 ms, you get 10 to 14 frames of delay in total. Then there's having to deal with changes in the connection quality, which forces the system to increase the delay. Juggles are still affected because the game is waiting for confirmation from the other peer and compensating by adding delay as necessary.

Right, so the game actually should go slow. This then forces the question in a peer-to-peer world (if the game continues frames assuming it's going to get it's information when it expects it, then does not): "Alright, who should've won, here, the holder or the puncher?" If DoA is using this, it's little priorities system is a little messed up.

Fixed delay/rollback style code, which only a few games so far use (Skullgirls, KI, Darkstalkers Ressurection) fares better since there's no need for confirmation from the other peer, so for the same 80 to 100 ms, you only get the 5 to 7 frames of delay since lag spikes and desyncs are handled by rolling back to the last fair state. That said, 5 to 7 frames of delay is still quite noticeable.

All they're really doing here is adding frame count tag (4 to 8 bytes depending on system, not that it'd matter much since you won't see the overflow, anyway, unless you're out in the woods playing and you both die and you come back as a spook) to the input. This really shouldn't be novel (we should've been doing it all along).

For more on this, you really need to read:
http://mauve.mizuumi.net/2012/07/05/understanding-fighting-game-networking/

It's not that hard to understand. With method 1, you have to delay everything by half-ping (whole ping is round trip, not one way, as to prevent issues with clocks not being synced). With method 2, you're setting up keyframes (unless you like jumping, the keyframe is the frame-count for the shortest move+1 [6 frames for DoA, iirc, so 100 milliseconds, or 200 milliseconds for your typical round trip ping, is limit before visual slowdown]) then the game will have to pause on those keyframes. If you need more than that, the fastest reaction time recorded is 7-8 frames, average is like 30 frames, so you could negligibly add 7 frames, which is only noticeable if you guessed a move instead of reacting.

 

[CyberEvil]

Somebody post a fucking mind blown GIF please.

 

[d3v]

Right, so the game actually should go slow. This then forces the question in a peer-to-peer world (if the game continues frames assuming it's going to get it's information when it expects it, then does not): "Alright, who should've won, here, the holder or the puncher?" If DoA is using this, it's little priorities system is a little messed up.

All fighting games are simply frame/state machines. Both peers simply compare inputs from specific frames. As long as the inputs line up, then the gamestate on both ends should be the same.

EDIT: Here's a fun fact that I just remembered. One way that certain games deal with lag, such as Tekken Tag 2 (considered by many to have good netcode), is by cheating and adding input delay on the offline mode. This means that players don't feel a difference between offline and online because even offline has lag.

 

[Argentus]

All fighting games are simply frame/state machines. Both peers simply compare inputs from specific frames. As long as the inputs line up, then the gamestate on both ends should be the same.

EDIT: Here's a fun fact that I just remembered. One way that certain games deal with lag, such as Tekken Tag 2 (considered by many to have good netcode), is by cheating and adding input delay on the offline mode. This means that players don't feel a difference between offline and online because even offline has lag.

that....sounds stupid, but its actually a really good idea, especially nowadays where pretty much all fighters are expected to be played online.

 

[crapoZK]

that....sounds stupid, but its actually a really good idea, especially nowadays where pretty much all fighters are expected to be played online.

I know right? It is also very cool way, they delay like what, 3 frames? Something like that. But still, it feels almost lagless, even when there is a 1/2 bar connection.

 

[d3v]

I know right? It is also very cool way, they delay like what, 3 frames? Something like that. But still, it feels almost lagless, even when there is a 1/2 bar connection.

The thing is, we don't exactly know what a 1/2 bar connection means in terms of ping/miliseconds. It could be only 80 ms for all we know.

 

[Kohlrak]

All fighting games are simply frame/state machines. Both peers simply compare inputs from specific frames. As long as the inputs line up, then the gamestate on both ends should be the same.

EDIT: Here's a fun fact that I just remembered. One way that certain games deal with lag, such as Tekken Tag 2 (considered by many to have good netcode), is by cheating and adding input delay on the offline mode. This means that players don't feel a difference between offline and online because even offline has lag.

I noticed. It makes tekken easier, IMO, because you feel like you're driving a gundam instead of playing a fighter.

The thing is, we don't exactly know what a 1/2 bar connection means in terms of ping/miliseconds. It could be only 80 ms for all we know.

Exactly. You can't tell people packet loss vs average round trip time (i'm really annoyed when connections look bad for 5 minutes then are suddenly green because the developer of that particular game thought that 1 packet was enough). As a result, you get a bar representing one or the other or an average. Then you gotta worry whether the game is taking that into account as opposed to the actual numbers.

 

[Javert]

Egos are invested on the notion of the game being bad. Fanboys from other fighting games can feel good about themselves bashing DoA whenever they may be having doubts about their game: “Taki fanservice? No way, I’m not playing here for the boobs. If I were doing that I would be playing DoA”. “Bad graphics you say. I’m here for the gameplay, if I were all about aesthethics I would be one of those DoA players”, etc. Game journalists and sites would never jeopardize their self perceived reputation by praising DoA: much easier to score rating by telling their readers what they want to believe whether it is truth or not and DoA is an ideal target since it is mainstream enough while not straight porn or visual novels that would make themselves look like dumb puritans.

So no, don’t expect awards any time soon. At this point the community should be keeping a grassroot growth and having fun while playing a competitive game: the “I’m a fighter” campaign is nice: they should be promoting the game making it clear that the gameplay is on par with the aesthetics and that both bases are covered, which a lot of fighters can’t say.

I agree that in a barren land of bad netcodes, a top level rollback netcode would be a very straightforward way of setting themselves apart.

 

[lotr9690]

Somebody post a fucking mind blown GIF please.