Quote:

Original post by Anonymous Poster wouldnt it be possible to make a twitch style game that just responds in a slower manner? I mean, people are constantly saying you cannot do it because of lag - but look at counter strike and other games. They are fairly twitch based. You move around and click a bunch. Lets say you have a ping of 200ms. Then your updates will be sent every 100 ms to the server. That will give you several clicks that are possible each second. For a standard punch (aka knife attack in CS) this should be possible cosidering that CS even uses TCP. The problem arises when you introduce combos. Youobviously have two choices: A) gather clicks on client side then send out "i did a combo" B) gather click on server then send out "player did a combo" Both obviously have thier pros and cons (if it would even be possible). The obvious flaw seems to be: in order to do a combo you must collect several peices of data in a very short amount of time and then conclude that combo is done (while determining that a basic move has not been done). This to me seems the real problem to me because of the introduced lag (er waiting) and determining if the combo was executed. Twitch play isnt the problem, twitch play with chained moves to determine combos is the real problem. With the MMO setup I would say you could not get basic twitch play (no combos). You would probably be able to pull it off with 20 or less clients. I would like to hear other peoples ideas for doing combos in fast paced games, ive always thought this would be a cool thing to play. As for the OP, Im pretty sure you were asking about reatime, because if it was not realtime the problem is easy - send all the packets time stamped, wait a little bit, reformulate them serverside, then conclude if there was a combo.

I REALLY would like to send the "I did a combo" command right from the client, but naturally I would run into the issue of hacked clients sending combo after combo.

combo cooldown =]

Quote:

Original post by TheyDontCallMeMatt combo cooldown =]

I've never liked the idea of a cooldown in games. Not to mention that the cooldown could be over by the time the player finishes inputting the combo anyways. I'm thinking more of a system where, since all combos drain player stamina, that once the stamina is fully depleted, it starts draining the other stats so that the combos are effectively less powerful than a conventional attack.

but the fight might be over way before all stamina is depleted

Why does everything have to be MMO? A fighting game could easily be made to have less than 30 players and I don't think anyone would mind, they can only be fighting so many players at once. Look at Soldat, that is loads of fun and yet not MMO.

Quote:

Original post by TheyDontCallMeMatt but the fight might be over way before all stamina is depleted

Well, obviously some numerical balance issues would have to be resolved, but thats no big thing to fix. It'll be trial and error for a while during the beta, but I can fix it.

Quote:

Original post by ussnewjersey4 Why does everything have to be MMO? A fighting game could easily be made to have less than 30 players and I don't think anyone would mind, they can only be fighting so many players at once. Look at Soldat, that is loads of fun and yet not MMO.

Yeah, heaven forbid I want to make an MMO for my own fun, and to expose people to something I think they might enjoy.

This is as much about the fun of myself and others as it is about me fixing some of the problems I see everyday in the rampant clickfest that makes up other MMOs as well.

So I've slightly reworked the combo execution system so that it will be a lot easier for me to balance. Basically, instead of just removing from stamina and arbitrarily adding to other stats, I'm having it so that stat value can be detracted from any battle stat, with a few other slight twists.

Enter "The Divine Fist of Crazy Punching Arm Flailing Madness"
Now, with the old stats, I would subtract an arbitrary amount from the stamina guage and let the player wail away, but last night I got to thinking. Now, I'll only add as much to another stat as I subtract from a stat.

So lets say that The DFOCPAFM has the following subtraction value:
Stamina - 5

Obviously, we subtract value 5 from stamina, and add 5 points to ATK, or add 3 to ATK and 2 to agility. We never take more than we give, so we don't get crazy high values for combos like I used to.

Expanding on that, we can also make it so that combos drain stats OTHER than just stamina. Lets say we have an attack called the "Blind Strongman". What this would do is drain everything from the players natural accuracy rating, and put it all into the attack power. So basically, we'll have a very low hit rate, but a high attack power.

Bump for more feedback.

Quote:

Original post by Steadtler kinda belong in networking, no? I dont think this system is feasable for a networked beat-them up. Im having trouble explaining why in words, altough it seems obvious to me. In networked games, only the server knows the real state of the game, every client gets a varied slightly delayed view. If I hit *punch*, the client should not be able to know if the punch landed. It send a 'punch command' message to the server, which updates the game's state and send new info to the clients, who gets to see if the punch landed or not. So lets say I have a ping of 100ms, and that I need half a second to enter a 5 key combo. That means I will have a least a 600ms delay between hitting the first punch and seeing if it lands or not. That means every combo will have to be done "blindly". That also means that combos can't really be interrupted by the opponent. So not practicable in a real-time game. If its Turn-based or pseudo turn-bases, then ok, altough a TB beat-em-up sounds wierd :)

Client side prediction means that you'll likely see you punch hit - even though it might miss. It works most of the time. In any case, what most games do is favor the client for those things - e.g, the server checks to see if you shot where YOU saw the player, not where the server sees the player. The latency issues are what cripple gameplay in MMO's though still - but only because of the security needed to prevent cheating. If you could give the client the benefit of the doubt, you'd be good to go. Unfortunately, you can't.

Nonetheless, I'd suggest trying a system where the server records it's last x number of states and then performs actions relative to the time that the client actually performs them. E.g., if you hit punch at 2:02:45:001, the server checks to see if your punch connected at the time that you hit (the update package sent from the client leaves with a timestamp). This way, the client behaves as if it's version of events is true and shows you hitting the enemy, and the server will ultimately ok this action because it checks your action against the state it has stored in memory. Barring huge amounts of server side slowdown, by the time the server checks your action, the simulation should not have advanced too far for it to still be plausible to update the results. E.g, unless you enemy teleports at the split second you attack him and the server lags for awhile, you should be able to hit punch and have the damage register before the opponent moves away.

There are lots of topics about CSP out there though along with networking in general. This is probably going to be moved, BTW.

Quote:

Original post by Nytehauq Quote: Original post by Steadtler kinda belong in networking, no? I dont think this system is feasable for a networked beat-them up. Im having trouble explaining why in words, altough it seems obvious to me. In networked games, only the server knows the real state of the game, every client gets a varied slightly delayed view. If I hit *punch*, the client should not be able to know if the punch landed. It send a 'punch command' message to the server, which updates the game's state and send new info to the clients, who gets to see if the punch landed or not. So lets say I have a ping of 100ms, and that I need half a second to enter a 5 key combo. That means I will have a least a 600ms delay between hitting the first punch and seeing if it lands or not. That means every combo will have to be done "blindly". That also means that combos can't really be interrupted by the opponent. So not practicable in a real-time game. If its Turn-based or pseudo turn-bases, then ok, altough a TB beat-em-up sounds wierd :) Client side prediction means that you'll likely see you punch hit - even though it might miss. It works most of the time. In any case, what most games do is favor the client for those things - e.g, the server checks to see if you shot where YOU saw the player, not where the server sees the player. The latency issues are what cripple gameplay in MMO's though still - but only because of the security needed to prevent cheating. If you could give the client the benefit of the doubt, you'd be good to go. Unfortunately, you can't. Nonetheless, I'd suggest trying a system where the server records it's last x number of states and then performs actions relative to the time that the client actually performs them. E.g., if you hit punch at 2:02:45:001, the server checks to see if your punch connected at the time that you hit (the update package sent from the client leaves with a timestamp). This way, the client behaves as if it's version of events is true and shows you hitting the enemy, and the server will ultimately ok this action because it checks your action against the state it has stored in memory. Barring huge amounts of server side slowdown, by the time the server checks your action, the simulation should not have advanced too far for it to still be plausible to update the results. E.g, unless you enemy teleports at the split second you attack him and the server lags for awhile, you should be able to hit punch and have the damage register before the opponent moves away. There are lots of topics about CSP out there though along with networking in general. This is probably going to be moved, BTW.

:P wouldnt bother me in the least, I was going to open a thread in networking two days ago, but my router wouldn't let me access gamedev.