AGS Kart

Started by SinSin, Sun 22/04/2012 21:33:22

Previous topic - Next topic

Monsieur OUXX

Quote from: dkh on Wed 02/05/2012 12:15:57
Please, once again, don't even attempt multiplayer for a game like this unless you know exactly what you're doing. It's really not feasible at all. Focus on a good singleplayer, it'll be more than enough to worry about it! :)

Awesome to see this getting started though!

Not only it is feasible, but it's overall rather easy (not VERY easy, but rather easy).
 

Monsieur OUXX

#141
Quote from: Kweepa on Thu 03/05/2012 01:32:14
Here's a new version with improved physics, extracted out to Racer.asc.
http://www.kweepa.com/step/ags/tech/AGS_Kart_2012_05_02.zip

SEE, that's EXACTLY what I was talking about.
as soon as someone said "I'll work on it", then someone released a parallel branch.  >:(
Don't get me wrong: I'm not saying Kweepa is silly, I mean the situation of coders "competing" because of the lack of sawrm development tools is silly.
Thank god, though, Kweepa, you uploaded it as an asc file :-) Thanks.
 

DoorKnobHandle

#142
Quote from: Monsieur OUXX on Thu 03/05/2012 09:09:53
Quote from: dkh on Wed 02/05/2012 12:15:57
Please, once again, don't even attempt multiplayer for a game like this unless you know exactly what you're doing. It's really not feasible at all. Focus on a good singleplayer, it'll be more than enough to worry about it! :)

Awesome to see this getting started though!

Not only it is feasible, but it's overall rather easy (not VERY easy, but rather easy).

So, since you say this with such confidence I assume you're an experienced netcoder then? Have you implemented a multiplayer mode for a comparable game before?

Tell me if I'm totally off-track here but I think you might have messed around (or seen other people play around) with the IRC stuff that works within AGS, you can chat and special text commands are translated into "Move cEgo here: x/y". If that is indeed the case and what you're basing your opinion about netcoding on, then you're unfortunately very, very much mistaken. A glorified chat room is one thing, a multiplayer action game another! We need movement prediction/lag compensation for one thing, that alone is highly complex to implement and requires the entire game engine to be designed around it.

In case I'm wrong, though, I'd be more than happy to see you write the multiplayer game mode for AGS Kart (since it's 'rather easy')!


Quote from: Monsieur OUXX on Thu 03/05/2012 09:11:12
Quote from: Kweepa on Thu 03/05/2012 01:32:14
Here's a new version with improved physics, extracted out to Racer.asc.
http://www.kweepa.com/step/ags/tech/AGS_Kart_2012_05_02.zip

SEE, that's EXACTLY what I was talking about.
as soon as someone said "I'll work on it", then someone released a parallel branch.  >:(
Not blaming Kweepa, just saying it's stupid. Thank god, though, Kweepa, you uploaded it as an asc file :-) Thanks.

I don't even understand what your problem is here, really? What's stupid? There are no parallel branches at all and as long as that's the case, everybody is welcome to chip in. That's what swarm development means!

Monsieur OUXX

 

Calin Leafshade

Lag compensation is not particularly difficult. Essentially you just allow objects to continue along their current path until told otherwise.

Also, a game like this would have *tiny* bandwidth needs so it might not even be necessary with modern networks.

I'm with frenchy on it being relatively easy in this case.

DoorKnobHandle

Well, if you guys are competent and willing to do Multiplayer then go ahead by all means. I'd be absolutely awesome to have it in AGS Kart of course, don't get me wrong! I'm just cautious because I've tried this before several times and it always ended in a huge nightmare, at the very least turning a simple 1-2 month development time into half a year or longer. But I'm no netcoder so if you guys know what you're doing it's all for the best! As an addition to my warning, I also don't think that Multiplayer code in a swarm developed game specifically works well either as this will make it extremely hard for anybody else to jump into the (net-) code and make some quick changes here or there so we're kind of limited to really having one guy code it all from beginning to end. But, again, if you guys think it's a good idea and feasible then go ahead and do it!

@Monsieur OUXX: I don't think I've said anything that implies that I need to relax in any way. I'm trying to give my input as a programmer and game designer to help this project out as much as possible because I'd really love to see it actually get made!

Monsieur OUXX

Quote from: dkh on Thu 03/05/2012 15:40:03
(happy ending)

I'm experimenting stuff in every direction at the moment, I'll re-surface when I've produced something interesting.
 

tzachs

As one who made a failed attempt to create an online real-time multiplayer game in the past, I agree with dkh that it's really not that simple, it was a nightmare for me too, and I've used a dedicated library for it (Player.IO).

Though, I've used Flash which is known to be slow, so maybe AGS with dedicated sockets would be more suited?

Anyways, if anybody wants to have a go at it, this tutorial helped me a lot (it's for flash & Player.IO, but most of the information there should probably be true to any real-time muliplayer engine):
http://playerio.com/documentation/tutorials/building-flash-multiplayer-games-tutorial/flashversion

SinSin

Things seem all over the place here. People not agreeing and people insisting that the KART racing game is now not one.
This was meant to be a happy project with a great team and now its chaos..... 8-0

I suggest that a team be built NOW!! (thats twice I think I've said this) before this thing loses all credibility.
I realize that a lot of people have put a lot of time into various pieces but I feel that this deserves better treatment than people just suggesting more and more.
Please do not be offended but without a team this will be lost and all that has been created so far will be gone due to zero direction.

Currently working on a project!

Hudders


Monsieur OUXX

hey guys don't let yourself be stunned by the passionate tone of the discussion: there is great stimulation at the moment, and several people ar eproducing great material for this game even as we speak.   ;-D

I'm sure it will all fall into place very quickly!  ;)
 

Sephiroth

#151
Sinsin: Why don't you take the lead (if you have time to dedicate to this), seeing how much motivation has been demonstrated, this could turn into another fine swarm project!
You would give the basic outlines and then call for help and organize the swarm :)

I'd be willing to work on a multiplayer feature, even if I fail in the end. I coded a working nbio tcp plugin for testing purpose, I can have a try after that week end and tell you.
The only problem I (fore)see with a multiplayer feat is the lack of bandwidth/cpu for clients (players computers), but it can be solved.

Anyways, good luck!

Icey

We could just add the basic parts of the multiplayer in the game and after the single player part of the game we could just add the rest of online stuff. You can have in the main menu Single player and Multiplayer. Clicking multiplayer can bring up the login textbox. Because you don't won't to make the player sign in right at load up cause it could interfere with the single player mode.

cat

I think the most important part is to find a way to split work. Is there a decision on what the graphics should look like? What about the cars, drivers, tracks, the menu screens etc. So one part of the team could focus on the mechanics and one on the graphics (have you thought about a two player on the same PC mode with split screen? Might be easier then network)

Could someone please specify the size of the cars and the drivers (would be awesome if you can mix'n'match them)? I'd help to draw those sprites.

DoorKnobHandle

Quote from: cat on Fri 04/05/2012 06:41:58
Have you thought about a two player on the same PC mode with split screen? Might be easier then network

It's indeed a metric shit ton easier to implement but due to the nature of the renderer in this kind of game it would have to render the track twice - and that's a huge performance issue. It was discussed earlier in the thread and dismissed as not being possible unfortunately.

Quote from: cat on Fri 04/05/2012 06:41:58
Could someone please specify the size of the cars and the drivers (would be awesome if you can mix'n'match them)? I'd help to draw those sprites.

For vehicles, there is no need for a strict maximum size but as a guideline I think 40x30 should work fine (we're talking about a 320x200 game resolution) - this was the size we were throwing around on the last page, if some cars end up being slightly smaller or larger it shouldn't be an issue.

Icey

I can still do the main menu.

m0ds

Personally I fail to see how this or things like previous swarm project can work effectively through a single thread. I'm not saying they won't, I'd just be interested in your opinions - would it be worth having a forum dedicated for specific swarm projects (perhaps under the Hosted Sites) board at the bottom. That way at least things can be contained and maintained a touch more coherently than pages upon pages of a single thread.

SinSin

This game is no longer to be considered a swarm. It was a silly idea as the project isnt big enough to warrant multitudes of helpers. a dedicated team is being created thanks for the support and keep an eye on the GIP thread for updates

@ M0ds  Definately set up a swarm thread maybe with its own GIP and Ideas area..    :)
Currently working on a project!

Monsieur OUXX

#158
I'm so upset right now  ;) I spent the afternoon trying to enhance Kweepa's track rendering code --  to boost the performance.
I did it by removing every possible real-time calculation (e.g. I calculated the cosinuses and sinuses beforehand) , by removing every possible nested function call (especially IntToFloat and FloatToInt), by transforming everything into integers, etc. Serious shit.

The glorious result of this is... a framerate divided by two compared to Kweepa's code  ;-D
I don't know how it's possible. for example, Kweepa is procedurally re-drawing the sky AND creating a dynamic sprite of the track every loop! Yet, changing these didn't save even one little FPS.

AGS optimization is still a mystery to me!  :-[
 

Monsieur OUXX

#159
EDIT: double post, I don't know how that happened
 

SMF spam blocked by CleanTalk