|
Posted By
|
Message
|
Pixelthief
Dedicated klik scientist
Registered
02/01/2002
Points
3419
       
|
24th March, 2011 at 18:59:41 -
I've helped build several RTS's from the ground up, so actually yeah I would have quite the idea of what I'm talking about.
And frankly, its the sort of thing you need to try to tackle to understand why its so important to understand every facet of it.
The support forums get flooded with requests by people trying to build RTS's once in a while, and they are mostly lost causes. To have anything on say the level of Starcraft, you'd need to construct a node map with traversals for the pathfinding, and 'fog of war' opens up a whole can of worms, and thats even before getting into the guts of the necessity of efficient unit grouping / finding algorithms.
Lets say something as basic as wanting your unit to attack anyone who comes within "600 range". How do you do this? You'd need to group all objects and compare their range, and over N objects shooting each other you'd have an O(N^2) algorithm? Or you could have tiles on the map store the positions of objects and iterate over them for an O algorithm with lesser results with huge memory constraints, or you could use clever tricks to sort objects by their X/Y positions and quickly pull them for AOE's, but you'd need to implement something like quicksort.
You can't do these sorts of things from widget style extensions. These are the core of an engine. You have to know how to do them and build them yourself to have any sort of respectable end product- you can't just plug a public pathfinding object into a public selection object and how for something that doesn't shrink down to 5 FPS when only 20 objects exist and doesn't do at all what you wanted.
We're talking about the fundamental data structures of a game. The need for pathfinding that explicitly maps into the level editor of your game- its not going to be the same from one game to another, and an extension based pathfinding would either have to be so incredibly general (ie a tree traversal algorithm) that it wouldn't save you any work, or it would have to be so exclusive that it would only work with a specific kind of structure that makes it worthless for unique projects- the paint-by-numbers problem.
You can't just hope for a one-size-fits-all solution. Thats naive, that doesn't work. You're better off using the Warcraft III World Editor and modding from there. If you want to know how to build an RTS from the ground up, I could explicitly describe all the pseudocode of what you'd need, and its not a short list.
What you're failing to identify is that movements such as "move in 8 directions based on keyboard" are so basic that the discrepancy between what you can accomplish with custom coding and built in movements is marginal. Yes, most serious projects will need everything coded from the ground up. Simple flash games might get by with them, especially when you're just rapid prototyping. But built-in RTS movements? Don't be too nutty now. Theres no way to build a 'small scale custom RTS' like you could poop out a flash bat&ball game. Its the difference between building a car out of legos and a car out of auto parts. And if you know the mechanics behind it, its not impossible.
I'm aware theres plenty of irrational fanboyism on each side of the debate, and clickteam diehards are just as prone as constructors, but you should both really tone it down a notch. You have to stop seeping so hard into one category and take a look at things from a broader view. Construct is a great little piece of coding, but its new and unpolished and lacks the community behind klik. Klik is being heavily updated enough to prevent it sinking as a legacy software, but still has quite gaping holes. But if you can't step back and see both sides of it, all you'll accomplish is moving a lot of wind, not informing anyone or swaying opinions.
Gridquest V2.00 is out!!
http://www.create-games.com/download.asp?id=7456
|
UrbanMonk
BRING BACK MITCH
Registered
07/07/2008
Points
49567
       
|
24th March, 2011 at 21:43:38 -
I just tried Construct 2.
I like it better than Construct 1, cause it can export to HTML5.
In fact I'm going to pay for it when it's decent enough to make a game with.
I think they should make it compatible with iphones and such, but at the moment there doesn't seem to be a way to intercept touches
A little pong game I made:
http://jsoftgames.info/pong/
EDIT: Just read this comment by Johnny
Originally Posted by Johnny Look
I don't want to sound offensive, but it seems to me you don't know much if anything about what you're talking about.
Johnny, if he doesn't know what he's talking about then you most certainly do not.
Out of everyone on the site I'd have to say that Pixeltheif has the most knowledge on this subject.
Please don't think of yourself more highly than you ought.
 Edited by UrbanMonk
n/a
|
![]()
Ski
TDC is my stress ball
Registered
13/03/2005
Points
10130
     
|
24th March, 2011 at 22:10:17 -
http://create-games.com/download.asp?id=8233
http://create-games.com/download.asp?id=8347
hmmmm 
n/a
|
Jacob!
Registered
17/06/2011
Points
153
|
28th March, 2011 at 00:47:08 -
After reading this entire thread, I have come to a conclusion. MMF2 is more powerful, more customisable, and requires you to know what you're doing in order to create your own product. On the other hand, Construct has better built in features that make you not have to do anything on your own. I quote, "Better built in movements, built in physics, etc.". Physics in MMF2 are easy once you know what you're doing. And since Construct 2 is just coming out, MMF2 had them first 
Good day to you all
Have you even been far as decided to use even go want to do look more like?
|
Cecilectomy
noPE
Registered
19/03/2005
Points
305
      
|
28th March, 2011 at 02:37:02 -
We'll see jacob. Construct 2 is not Construct. It is an infant in comparison to both Construct 0.x and MMF2. The only selling point for me right now over a new mmf is that contrary to what you said, "MMF2 is more powerful, more customisable" Construct 2 is actually potentially FAR more powerful and customisable (emphasis on 'potentially'), with its supposed development around addon runtimes. Not like mmf2 where they keep having to code proprietary runtimes. or maybe the mmf2 runtimes ARE easily made as addons, but they arent letting the users in on that, Construct 2 is. The vibe Scirra is sending out is that their focus is on extreme extensibility for the forseeable future, to the point where there really won't need to be a Construct 3 for a long time.
Edit: And considering all the time I've been spending doing actual work in Java, C/C++, and C#, i would rather code extensions and runtimes FOR these products than actually using them myself.
Edited by Cecilectomy
n/a
|
zedd777
Registered
01/11/2009
Points
9
|
28th March, 2011 at 09:44:57 -
They say that the HTML5 runtime will be open source but that is not exactly the term that should be used because it's useless for anything other than C2 so it would be more like a runtime SDK with HTML5 as a default example. The HTML5 runtime source code is currently not available to anyone yet either.
The thing is though C0.x is buggy and they basically gave up on it, the open source code is also not so "open" as it requires commercial library's before you can build it. Even though it's still a really early for C2 they now have put some time trial on it plus it's now closed source so it seems like they have totally dropped the whole open source thing other than the plugin and runtime SDK's. I like the way the current javascript plugin SDK works for C2 work though so it's not all bad but since it's just javascript i could just code this easily anyway in notepad if i wanted so i am hoping they will have a C++ runtime soon.
For Java and C/C++ you can already code extensions for these in MMF2 and there is the .net extension which allows coding with C# and i am also guessing there will be some sort of C# based SDK when the XNA runtime is released. I am sure C2 would have these at some point also but currently there is not much and i guess it could be a long time before they are done.
People forget that MMF3 could be released at some point also, since they already have the source code for many runtimes and extension even if they make it 100x better and totally re-made they could do this much faster since they have a load of code to work with. I have been using Construct 0.x for about 2 year now but found it to be totally buggy as i have had many crashes and found bugs. Construct has a nicer editor, good for HWA effects and nicer for built in movements but even with a GUI that seems a bit more retro now i still find MMF2/MMF2 HWA much more stable and powerful with what it can do and think it's the best game/app maker to use other than coding.
n/a
|
Cecilectomy
noPE
Registered
19/03/2005
Points
305
|
28th March, 2011 at 10:17:05 -
when i mentioned java and c/c++ i meant using it to code runtimes, not extensions. I am already aware of mmf2's SDKs and have already worked in them.
n/a
|
Jacob!
Registered
17/06/2011
Points
153
|
29th March, 2011 at 04:14:00 -
MMF2 runtimes are just addons. Mathias is coding the Anaconda (Python) runtime and he doesn't work for Clickteam
Have you even been far as decided to use even go want to do look more like?
|
Codemonster
Administrator
Klik & Play Expert
Registered
03/08/2008
Points
133
|
30th March, 2011 at 00:37:41 -
Guys honestly just forget about the whole MMF&Clickteam vs. CS/Scirra thing.
Hands down MMF wins and anyone that says Construct will give it a run for its money is forgetting the fact that in 2+ years they were unable to achieve stability with an even more simple version of construct, let alone finish it. The only worry for you clickteam fans is the software being pirated.
CS can output to HTML5, and you can make your own runtimes... Still got to take into consideration that these guys lose steam quickly.
As for the excuse of a quarter million lines of code being worth the extra money (even though proof of a WORKING CONCEPT has never been delivered in all these years with Scirra)...
The entire code base (including scripts) for Khan Adventures nears half a million lines, and on top of that I've completed all the graphics, sounds, music, scripts and storyline myself - so, the amount of work you have done is not a legitimate reason to ask for money. Producing something FINAL, that WORKS AS IT SHOULD is worth money. People buy results, not half-finished ideas.
--
Jesse J
|
Codemonster
Administrator
Klik & Play Expert
Registered
03/08/2008
Points
133
|
30th March, 2011 at 00:41:30 -
And that is not to mention they wouldnt even have a community to sell to if it werent for those of us that pushed it so much through word of mouth.
They could have at least had the audacity to finish the product we promoted for them for free, before deciding to abandon it thus wasting 2+ years of our time helping to report bugs and help with ideas on how to improve the product. Asking for money doesn't sit well for many of us - not until there is something to buy.
--
Jesse J
|
GamesterXIII
Registered
04/12/2008
Points
1110
|
30th March, 2011 at 15:03:26 -
So are you going to try to sell Khan adventures? 
Edited by GamesterXIII
n/a
|
Hagar
Administrator
Old klik fart
Registered
20/02/2002
Points
1692
|
30th March, 2011 at 15:15:24 -
As an engineer whom does software and hardware for a living I agree with Codemonster in a way.
Construct was buggy and unstable. Instead of fixing it, they have instead decided to make another version from the ground up, which you may think is a good idea presuming a process of simplification would be carried out to help increase the odds of producing something commercially viable. Instead of this, it seems they have decided to re-write an unfinished product, and in the process make it more complex.
I would love to see Construct become a real finished product, but I think in terms of planning they are going in the wrong direction.
Edited by an Administrator
n/a
|
Codemonster
Administrator
Klik & Play Expert
Registered
03/08/2008
Points
133
|
30th March, 2011 at 21:38:03 -
Gamester - Absolutely not, I'm assuming you didn't understand my point. With that being said, even if I did wish to sell such a product, it would clearly be that way from the beginning, worth the money and without bugs that affect the core functionality of the product. Excellent critism though, from a guy whos claim to fame are two projects (with either minimal or stolen artwork) that, much like Scirra, are obviously buggy and not updated for huge amounts of time.
Hagar - Yes, that is precisely what I mean.
--
Jesse J
|
GamesterXIII
Registered
04/12/2008
Points
1110
|
31st March, 2011 at 07:19:24 -
Originally Posted by Codemonster
Gamester - Absolutely not, I'm assuming you didn't understand my point. With that being said, even if I did wish to sell such a product, it would clearly be that way from the beginning, worth the money and without bugs that affect the core functionality of the product. Excellent critism though, from a guy whos claim to fame are two projects (with either minimal or stolen artwork) that, much like Scirra, are obviously buggy and not updated for huge amounts of time.
Hagar - Yes, that is precisely what I mean.
I was just messing with you, and i'd give you props if khan is completed whether I like it or not, but wow you're an idiot. I never "claimed fame" over my unfinished projects, or anything for that matter. The projects aren't being worked on, you're right about that, but the art is far from stolen. The art in the game with the frog was 100% done by me in very little time. The first screenshots of Eden (the really bright colored one) contained my art (everything but the character/bird) , a friends art, and ONE sprite taken from zelda which was mentioned and replaced as soon as my friend got around to it. I just needed something with flying animations to see if the movement looked alright - that is probably the first time I've ripped a sprite since I was 9. The screenshot with the large sprites was 100% my friends artwork.
I hate to pull the jealous card, but why would you go so far as to claim that some half-assed mediocre artwork is stolen? Not to mention neither engine had any bugs in it as I refuse to release or show anything with bugs. The only thing I remember ever being "wrong" with Eden that wasn't really a bug was a misaligned hotspot that made it look like you were floating off the side of a platform. Something that I, frankly, dgaf about as that can be fixed at any time.
Consider suicide
Edited by GamesterXIII
Admin Note
Do not abuse other members (Rule 4). You have been warned.
n/a
|
Johnny Look
One Happy Dude
Registered
14/05/2006
Points
2942
|
1st April, 2011 at 15:51:55 -
Pixelthief: Sorry for the late reply, here it is:
"To have anything on say the level of Starcraft, you'd need to construct a node map with traversals for the pathfinding, and 'fog of war' opens up a whole can of worms, and thats even before getting into the guts of the necessity of efficient unit grouping / finding algorithms."
Fog of war can be done rather easily using the active overlay object for example, as for unit grouping and finding it's relatively easy when you know how it's done in any other language.
"Lets say something as basic as wanting your unit to attack anyone who comes within "600 range". How do you do this? You'd need to group all objects and compare their range, and over N objects shooting each other you'd have an O(N^2) algorithm? Or you could have tiles on the map store the positions of objects and iterate over them for an O algorithm with lesser results with huge memory constraints, or you could use clever tricks to sort objects by their X/Y positions and quickly pull them for AOE's, but you'd need to implement something like quicksort."
Or you could just have a circular sprite with 600 pixels of diameter attached to the unit sprite and then check for collisions. If it's colliding, attack. Pretty simple I think. Alternatively you could just check for distance with the closest enemy unit much like you would if it would have been done in c or c++.
Anyway I think it's pointless to discuss if it can be done or not and why, this discussion will only end when someone shows if it can be done or not. I know it can be done in construct at least because I did a basic rts skeleton with it before. I'll try to find the source and post it here, if not I'll recode it again.
"Johnny, if he doesn't know what he's talking about then you most certainly do not.
Out of everyone on the site I'd have to say that Pixeltheif has the most knowledge on this subject."
Ahah sure. On what grounds do you base your opinion ?
n/a
|
|
|
|
Advertisement
Post Reply