MMO using the unreal engine..

[kisk]

Me and a group of guys are toying with the idea of creating an mmo using the unreal engine and we wanted to get some feedback on the idea. It wouldn't work? Tell me why. Will it be a faster paced mmorpg... hell yes.

[WickedPenguin]

The basic engine is certainly capable of handling an MMORPG. As far as pro-level MMORPG's that used the Unreal Engine, I know at least one off the top of my head: Lineage II, which used a very highly modified UE2 or UE2.5. Also, a quick Google search revealed a game called Huxley being built off of UE3: http://www.webzengames.com/Game/Huxley/default.asp

However, I'm pretty certain that there will be limits to how far modders can go with regards to the code. If you recall, UT2004 gave a lot of free reign for modders, but there were certain things that were just hardcoded into UT2004's "deep code" that were either impossible to overcome or required some very complicated workarounds. I think it's safe to assume that UT3 will be the same way. They can only give away so much without risking their intellectual property.

[Kronos]

Me and a friend thought of the same thing(creating a MMO mod for UT3).

Technically its possible if you can some how exceed 100 players on one server and have multiple servers linked together.

[[GU]Elmur_fud]

mmo's for UE3 other then huxley that I can think of off the top of my head.

Fury

Lineage 3

Spellborn

http://www.endlesssaga.com/

[Kronos]

SpellBorn runs on Unreal Engine 2.5.

[TotalChaos]

Elmur_fud;25005060']Lineage 3

Yes Lineage 3 was originally going to be made with Unreal Engine 3, however the project was canceled and is going to be restarted in 2008. This time around, NCsoft (creators of Lineage, Lineage 2, and what will be Lineage 3) stated they were going to be developing their own engine for the game.


Article here if you dont trust my word: http://lineage3.wordpress.com/

[[GU]Elmur_fud]

SpellBorn runs on Unreal Engine 2.5.

This I know, I have been watching the site since the first news posting about it on beyondunreal a couple years ago.

However from what I read on unreal wiki (i think) they must intend to migrate to UE3.0

Perhaps though whomever posted it was wrong.

in the list of licensee's is spellborn internation

bottom of page for a nice big list.

http://wiki.beyondunreal.com/wiki?action=browse&id=Unreal_Engine_Versions/3&oldid=Unreal_Engine_3

[[GU]Elmur_fud]

Yes Lineage 3 was originally going to be made with Unreal Engine 3, however the project was canceled and is going to be restarted in 2008. This time around, NCsoft (creators of Lineage, Lineage 2, and what will be Lineage 3) stated they were going to be developing their own engine for the game.


Article here if you dont trust my word: http://lineage3.wordpress.com/

Well i have no problem trusting you. But I find such sites to be highly shakey evidence at best. I would want to see/hear it from ncsoft themselves. For now I will take your word for it.

[TotalChaos]

Elmur_fud;25005124']But I find such sites to be highly shakey evidence at best. I would want to see/hear it from ncsoft themselves.

Yeah, understood, I will do a little searching for an official NCsoft announcement sometime, but since its a heavily Korean based company the info is probably on some Korean site that will require translating... But if I find solid evidence I'll PM you the info. :)

[re8]

I've seen so many big projects in UT & NWN never get anywhere since they are too large for a volunteer force. The best bet, is to plan it out in parts, & allow each part to be used outside the MMO.

For example, the only way to do a MMO in UT3 is to have separate-loading zones (maps). You'll need to have the maps be able to gate people in & out via doorways, paths, portals, etc, to simulate the illusion of concurrently moving from one area to another. Also, you'll need to build routines to handle carried-items from one map to another, handle global chat, NPCs (bots) that may travel zones, etc. If you build this part, other types of UT3 games may be able to use such technology. So, even if the only completed part is the zoning aspect, then you'll atleast accomplished something usable. A multi-map campaign would be fun, even with out the MMO in full effect.

IMHO :)

[l-_-l]

I would LOVE to see this.....


IMO - nice little magic mmorpg but thats just me.

however, you should make the movement awesome! allow people to jump, climb, jump a gap and grab onto the other side... stuff like that. there is NO mmorpg that has that, if you guys implelmented it.. you would be awesome.!!!

ive actually started writing out an idea for a mmorpg that was fantasy... if you want to hear it just say so.... its not done not by a long shot.!


good luck and please do it!

[WickedPenguin]

I've seen so many big projects in UT & NWN never get anywhere since they are too large for a volunteer force. The best bet, is to plan it out in parts, & allow each part to be used outside the MMO.

That's a very, very true statement. How many Total Conversion mods were there for all the previous UT's that promised tons of advanced features, delivered none of them, and then died out when one or two key members left, taking all of the assets with them?

Quite frankly, it's a sh!tload of work to make a game mod. People don't realize it when they get into it. Inspiration is nothing without execution. Everyone wants to crack open Maya and Photoshop and crank models out, but they don't realize how much work is truly involved. UV mapping, code bug squashing, version control, asset management, and - most difficult of all - team wrangling can be miserable work.

My advice is: start small. Not think small, but work within your means. Take things one step at a time. And - very important - make each phase actually releasable to the public so that you can build community interest. Your community is the only thing that will make your work worthwhile, so listen to them and give them something to play with as soon as possible.

Look at the path that Red Orchestra (winner of the first MSUC) took:

* The "grand" idea was to make a realistic WWII shooter that featured accurate weapons and settings.
* It began with a few basic maps and some basic weapons - bolt-action rifles and submachine guns.
* Machine guns were added.
* Semi-auto rifles were added
* Artillery was added.
* A couple vehicles were added, with one or two maps.
* Vehicles were revised and many new features were added.
* Wins the MSUC.
* Revises all content and releases the game for sale.
* Continues to add maps, features, and content.
* Has sold a ton of copies, enough for the developers to be working on a Unreal Engine 3 game.

Each phase was not perfect, but it was releasable and playable. This built considerable interest over time, increasing player counts and creating momentum. The more people play your mod, the more inspired and committed your team is. By the time Red Orchestra hit Version 3.1 (the version that won the MSUC) it was as flawless as it was going to be as a Total Conversion mod.

[DGUnreal]

Depending on the number of simultaneous players you are intending to support, you'll have to write your own netcode.

[XenonG]

Major server side modification?

[DGUnreal]

Major engine modification.

Not only would all of the maps have to be designed to run this player load, but the current engine netcode is only designed to handle the Unreal Universe style of twitch FPS.

A "mod team" would require at best multiple professional-level programmers familiar with coding the entire client-server networking framework for MMO games.

To quote Epic on this subject:

"Client-server model supporting up to 64 players as provided"

"Please note that we don't provide a server or networking framework suitable for massively multiplayer games."

" ... such a task is a multi man-year engineering effort ..."


I don't mean to discourage the community at large, but a lot of community mappers and gamers are going to be in for a surprise when it comes to the new engine, or any Next-Gen engine for that matter.
Creating assets, maps, and mods are significantly more complex and more work with the new technology.
I recommend that all mappers, mod'ers or TC-hopefuls wait until UT3 ships before they start sharing what they would like to accomplish with UT3, as it is going to be substantially more work than what they assume, if they are basing their goals on what it required for UT2004.

[re8]

Major engine modification.
Not only would all of the maps have to be designed to run this player load, but the current engine netcode is only designed to handle the Unreal Universe style of twitch FPS.
A "mod team" would require at best multiple professional-level programmers familiar with coding the entire client-server networking framework for MMO games.
To quote Epic on this subject:
"Client-server model supporting up to 64 players as provided"
"Please note that we don't provide a server or networking framework suitable for massively multiplayer games."
" ... such a task is a multi man-year engineering effort ..."If you want a more or less seamless world, yes alot of engineering. But you make it multi-zone like NWN1, once you get past that part it won't be too too hard.

UT2004 had "persistent data", where you would "level" & can change your "stats". Some mods had "money" & "stores". I don't see anywhere that UnrealScript is 100% re-written, but please correct me if I'm wrong. Changed & upgraded yes, but I think you'll start seeing new mods within a month of the Unreal editor release, not years :)

Take WickedPenguin's advice; playable phases. Or mine, make one recyclable segment at a time. Or both (preferred) But please don't bite too much to chew :)

[Melekai]

for a project like this.. the only way i can see it working is to open source it.

i have a ton of ideas. but the truth is just like someone else said.. just wait and see what the engine can handle first. i'm not sure im a big fan of MMO's but i wouldn't mind being part of an open source project for up to 64 players. at once.
that way you could create a base core Midrange multiplayer system that could work on a local level accross local servers. the server maintainers could perhaps modify their own rulesets ( xp gain for rpg mods etc).

what im really waiting for is to see if Unreal 3 can handle melee. oh. and some way to tie in a database to the whole thing. as well as seeing what can be done with physics.

but im just one dreamer :) we will see

[DGUnreal]

If you want a more or less seamless world, yes alot of engineering. But you make it multi-zone like NWN1, once you get past that part it won't be too too hard.


I don't mean to sound discouraging, the chances of a community game pulling this off is not good. 64 players is not MMO, MMO's are hundreds to thousands of players on the same server. This requires re-writing a portion of the engine, re-writing the engine requires the source code, which requires an engine license, which requires very deep pockets (~$500k).

If you have ever tried even a 32 player server under UE2.5 such as UT2004 Titan servers, you would know that attempting to support more than 32 is going to tank on most systems. We attempted a 128 player hack of UT2004 a number of years ago as a test, and it was unplayable. The chances of creating even a mini non-MMO 64 player game is going to require some people with serious knowledge, and it will probably only support UT99 level of polygon counts or less.

Assuming you can limit servers to 32 simultaneous players with streaming won't work well, you'll still require a master server that supports the entire world, and how many people are going to play with the limitation that they cannot go to world servers that are full. That severely restricts buddy playing, interaction and clan warfare, etc.

That said, best of luck to you.

[[GU]Elmur_fud]

...and it will probably only support UT99 level of polygon counts or less.

Assuming you can limit servers to 32 simultaneous players with streaming won't work well, you'll still require a master server that supports the entire world, and how many people are going to play with the limitation that they cannot go to world servers that are full. That severely restricts buddy playing, interaction and clan warfare, etc.

That said, best of luck to you.

I don't mean to be rude by being contradictory but you may have some misunderstandings about how mmo's generally function. I have first hand knowledge about this as I have made or altered many a private server for many a mmo. B4 any1 asks, no I am not distributing them, so don't ask.

The the server runs little to no content from the game itself in an mmo, what it does run is tags that tell it what stuff is and where it is.

Most mmo systems are broken down into 3 parts; you have a login server, a character server, and a world server. All of which generally can be run from different machines (IE have different IP's).

This is just a generalization of common functions.

The login server validates a user against a mysql database (usually) and determines that they have the needed files and there appropriate versions b4 passing them on to the char server.

The char server takes them to the user selection screen for their account(again usually) and passes them on after the appropriate info. The world server is where the game is (yeah that should be a DUH) but it itself communicates with the char server to maintain data (to prevent cheating, item loss, etc).

The world server maintains a coordinate grid in x,y,z format where ID#@x,y,z;facing;does something. This is why wall hacks and such work so well in mmo's, that info is client side.

Still it takes a mofo of a comp to maintain all that info for every player. And further yet UE3 isn't designed to run that way so yeah I think you would need a license also and some damn good coders.

RPIS (Relative Position In Space) is what it's all about in mmo's

[WickedPenguin]

Red Orchestra: Ostfront was able to modify their Unreal 2.5 Engine to handle 64 players at a time, but it took a tremendous amount of troubleshooting and testing, a very powerful server (by today's standards), and extensive modification of their licensed copy of the UE2.5 engine.

Like DGUnreal I don't want to sound discouraging, but without engine-level code access a UT3-based MMO simply isn't going to happen. I would rather see artists and coders devote their time to something more reasonable than waste their valuable efforts on something that is technically impossible with the available assets. Maybe someone from Epic can weigh in here and give us an in-house perspective on this, but right now I really don't think that the MMO idea will work out.

[DGUnreal]

Elmur_fud;25008558']I don't mean to be rude by being contradictory but you may have some misunderstandings about how mmo's generally function.


uhh.. no I understand, i am referring to attempting to hack UE3's streaming system into an MMO. :p

Standard MMO's are still hundreds to thousands of people per server (look at Eve), not 64 like UE3 supports. You still require a master server that contains persistent player info, tracks who's currently online and where they are, etc., plus their interactivity with the world if that is being supported which requires server-side control (otherwise it's just a simple online chat room), along with one or more servers to encompass the entire world if you are attempting to do UE3 map streaming. Attempting to split this up too small with limited players per server is no longer an MMO.

[WickedPenguin]

One thing that may be possible is a microcosm of an MMO world. Instead of creating a massive online world that you can traverse from one end to another, why not focus on the "quest" aspect?

The Unreal Engine 3 allows for map-to-map jumping. While you're playing one map, another can be precaching, so you can jump to the second map when the first map is over. I researched this a long long time ago so it's kind of hazy how it works.

However... in theory you could create a string of five or six maps to create one single "quest".

For the sake of common example, I'll use Lord of the Rings.

Map 1: Hobbiton (laying out the quest)
Map 2: The Mines of Moria
Map 3: Helms Deep
Map 4: Pelennor Fields
Map 5: Minas Tirith
Map 6: Mount Doom

While it would no longer be an MMO, it would still have a lot of what you play MMO's for: cooperative teamplay, character-building, etc.. You could start up a server with a group of friends and play one quest all the way through. As you progress, you level up within the maps. The matter of sides would need to be explored as well - do you all play on one side and fight A.I. NPC's, or can people join the enemy and fight against you?

The main question, though, is what happens when the level ends? How do you retain your status? Is there a way to create a persistent client-or-server-side persona for your character? (Client-side would be way too easy to hack and create an "uber" character.) Or are your character's randomly created every time you join a server? I would think it would still rely too much on a master server and database.

It seems the biggest technical challenge - aside from player count - is keeping that persistent character. After all, isn't that virtual persona the whole point of playing MMO's?

[[GU]Elmur_fud]

uhh.. no I understand, i am referring to attempting to hack UE3's streaming system into an MMO. :p

Standard MMO's are still hundreds to thousands of people per server (look at Eve), not 64 like UE3 supports. You still require a master server that contains persistent player info, tracks who's currently online and where they are, etc., plus their interactivity with the world if that is being supported which requires server-side control (otherwise it's just a simple online chat room), along with one or more servers to encompass the entire world if you are attempting to do UE3 map streaming. Attempting to split this up too small with limited players per server is no longer an MMO.

I certainly agree it would be impossible if not almost impossible without a license, engine-wise I agree with you, but you seem a little off on server side function, most mmo's can be run off a fairly low-level machine. The catch is you wont be doing it for free because the bandwidth requirements. As you say the server needs to maintain often hundreds of thousands of players and the server must have the bandwidth for that many connections.

Packet-wise the amount of info to maintain connection is small on the individual basis, but when you are talking even a thousand people on the server the load is much greater.

Mostly I was referring to... "and it will probably only support UT99 level of polygon counts or less."

My point was that, polygon count is fairly irrelevant if you code your server correctly (but again we come back to the need for a license), even in UT3 if the server is not loading anything but rudimentary data and isn't loading but the needed information it is very possible to make a mmorpg.

Without a license, best you could probably do without badass coders is maybe a multiplayer squad based online rpg.

If you wish to do world checking, some servers offload world data to a server that maintains a table of coordinates you can be at. Some mmorpg's that do this react to coordinates your not allowed in like blocking zones. Others determine that the client side game should be sufficient to keep where they should be and just move you to disciplinary areas if the server detects you out of range.

I don't know of any however that load much game content at all.

If I remember right UT2K4 only offloaded terxtures and staticmeshes (keeping the collision) and that put a greater load on the server. But made cheats more difficult...unless you left holes in your map.

I solved the getting outside the map issue in my maps simply by putting kill volumes outside the map. :p Boy did that piss some cheaters off.

[Wolferey]

After all, isn't that virtual persona the whole point of playing MMO's?

Exploring? Meeting new people online? Maintaining a clan/guild/ledership? doing quests/missions? Social aspects? experiencing the lore? O,o
There are alot more aspects to MMO's than just your online avatar, he/she one of the things that makes MMO's enjoyable.

As for making an MMO for UT3, perhaps this will spawn some ideas?:

https://research.microsoft.com/~lorch/papers/iptps2007.pdf
http://channel9.msdn.com/ShowPost.aspx?PostID=290801

[WickedPenguin]

Exploring? Meeting new people online? Maintaining a clan/guild/ledership? doing quests/missions? Social aspects? experiencing the lore? O,o
There are alot more aspects to MMO's than just your online avatar, he/she one of the things that makes MMO's enjoyable.

Well, that's what I meant. You don't play WoW just to make a pretty Dark Elf avatar. You play it to create an online identity (a.k.a. virtual persona) and to live an alternate life, interacting with people and achieving goals within that world.

The problem with making a UT3-based MMO is keeping track of your persona's history, accomplishments, friends, acquisitions, etc. A person is a product of their past, and if you can't save your past...?

[[GU]Elmur_fud]

Well, that's what I meant. You don't play WoW just to make a pretty Dark Elf avatar. You play it to create an online identity (a.k.a. virtual persona) and to live an alternate life, interacting with people and achieving goals within that world.

The problem with making a UT3-based MMO is keeping track of your persona's history, accomplishments, friends, acquisitions, etc. A person is a product of their past, and if you can't save your past...?

UTrpg did that in 2k3&2k4 why not again.

[WickedPenguin]

Elmur_fud;25009100']UTrpg did that in 2k3&2k4 why not again.

Cool. I was not aware someone had done it successfully.

[DGUnreal]

Elmur_fud;25008881']
Mostly I was referring to... "and it will probably only support UT99 level of polygon counts or less."


Last time I looked, this:

http://www.eq2center.com/gifs/July/1.jpg

sure did not equal this: :)

http://www.beyondunreal.com/images/e3_2007/ut3_e307_04.jpg

and this:

http://www.gameogre.com/everquest2.jpg

did not equal this: :)

http://www.unrealtechnology.com/screens/character_creation3.jpg

My point was, the big issue is going to be what you can run client-side for poly count. 128 x UT3-quality players is a massive amount of polys to be pushing on one map (FYI: ~850,000), then add onto that your scene poly count, effects, all of the physics, etc.

UE3 and UT3 are designed for relatively high-poly characters, vehicles and scene trangle counts, plus all of the effects, which is one reason why player counts are going to be limited in-game.
Even if you assume that only 1/10th of the players may be in the frustum at any one time, that is still a lot of polys when you are running hundreds or perhaps thousands of players on one map. And if you want decent view distances, you can't count on that for a lot of culling. Or if you attempt to solve the issue with larger maps sizes (512k world and 1024 heightmaps) in the hopes that people will be spread out more and cull out, the issue becomes terrain and deco rendering overhead, etc.
If you don't believe me, launch a recent common UT2004 outdoor map and spawn 32 players or hop online one of the Titan servers (not the best test as they throttle back distance fog to 15k or less on their maps), or better yet, hack the scripts and spawn 64 or more bots in.
Unless you are planning on limiting the total number of characters and scene triangles within a given map area to 8 thru 16 or your distance fog is 8k, your client framerate will tank, unless you have very low total poly counts and you change the engine to either optimize the scripts or better yet drop the interpreted script system which has a lot of overhead compared to pure code. The engine isn't designed and optimized specifically for this style of game.
I believe I already mentioned previously, we tried UE2.5 using UT2004 detail level scene setups with 128 players a few years ago here and the resulting fps was below 10 on average hardware of that time, which is unplayable. So we tossed out the game idea we were working on.

And as a level designer who has been working with UE3 for over a year now, I highly doubt that the current common PC hardware is going to run fully equipped 64 players plus vehicles (or horses or whatever is desired) on a map with Epic UT3 quality scene complexity and detail.

That being said, the community is more than free to pursue this avenue of game design. :p
An RPG is do-able (single, co-op or limited player counts), and would be great to see, but an MMO/MMORPG is going to take a game-studio level of expertise and serious engine modification.

Like I said, a 64 player mod can probably be accomplished, and you may get UT2004-ish quality if you used a lot of tricks, but 64 does not an MMO make... ;)

[[GU]Elmur_fud]

Dude mmo servers don't load the polygons. It don't matter if the game uses 10 poly models or 10 billion poly models. :cool:

Those screens show the difference of artisan skill more then anything else.

Lineage 2 is the same engine as UT2K4 it has thousands of players... and the models have higher polygon counts then UT2k4 at times. Also massive outdoor environments.

I run my wow and lineage 2 servers off a P1 laptop. That thing could never hope to actually play those games it freaks out at DN3D.

mmo's are generally less quality more quantity also they make there money by supporting as many people as possible. Most mmo's offer their client for free and either charge to connect and/or for special in game (cash shop) items. So it behoves them to have as seemless of game play for the client as possible, because server lag + client lag = dead. Which if it happens to much your customers leave.

[DGUnreal]

Elmur_fud;25009204']Dude mmo servers don't load the polygons. It don't matter if the game uses 10 poly models or 10 billion poly models. :cool:

Ok, well I'm out of this thread, it is clear you do not know what you are talking about, and I don't want to waste my time on this.
Read what I post before you reply, I said client side, I didn't say the server renders the scene. And if your PC can actually render 10 billion polys client-side, then you must be running hardware from the future. Poly count client-side does matter in the game, especially when you have MMORPG player counts in-scene and in-zone.

Any MMO using the Unreal Engine will as I have stated repeatedly, been using a modified engine code base and low poly counts client-side. If you think you can get UT3 level of detail with EQ1/EQ2 player counts of 200+ players visible / 400+ in-zone, then good luck to you.

[[GU]Elmur_fud]

Ok, well I'm out of this thread, it is clear you do not know what you are talking about, and I don't want to waste my time on this.
Read what I post before you reply, I said client side, I didn't say the server renders the scene. And if your PC can actually render 10 billion polys client-side, then you must be running hardware from the future. Poly count client-side does matter in the game, especially when you have MMORPG player counts in-scene and in-zone.

Any MMO using the Unreal Engine will as I have stated repeatedly, been using a modified engine code base and low poly counts client-side. If you think you can get UT3 level of detail with EQ1/EQ2 player counts of 200+ players visible / 400+ in-zone, then good luck to you.



Ok I see nowhere other then one line in your last post that speaks about it being client side that would have a issue with the number of polygons. Admittedly I did miss that. But the tone of all your posts have came across as if you were saying that it was the server that had to load the poly's.

I know fully of what speak, I have been researching the process and procedures etc etc for MMORPG's for 5+ years now. I have everything from business structure to 3d assets... in fact the only 3 things lacking in any large shape or form to get the game finished are financial backing, score, and code (that last one is quite the monkey). I have Concept art, story boarding, 300+ pages of storyline, hundreds of sounds, a couple thousand textures, around 1700 models, detailed layout and planning of all gameplay fundamentals (economic system, political system, equipables, pickups, spells, quests, etc) As well as having the details thereof of much fleshed out.
That is neither here nor there I suppose, but I do know a fair bit of what I speak.

If indeed you spoke all the time of client side then this was mostly one colossal miscommunication. Because the only thing about client side I said was that is where the assets are rendered. I was speaking strictly about server side as it looked (and looks to a degree) that you spoke of server side also. So it would seem that you also didn't read what I wrote well either.

I also said at least twice that it would take a license to pull off a mmo, agreeing with you that it isn't possible without a extreme rewrite.

I agree that that sort of load would kill a client, but a server couldn't give a rats ass. I don't know that there is anyway to make the client behave with current hardware under those conditions. I do know in the UE3 game Gothic3 I have encountered close to 50 orcs at one time. but that is nowhere near the level of mmo still.

And are you saying everquest2 is a UE based game? Cause I hadn't heard about that 1 that I remember.

[mudd]

I've been looking into making an MMO like hundreds of other mod developers in the last month or so. MMOs in one form or another are clearly the future and popular, so I'll share how I analysed what was needed using the Unreal engine:

- Database + custom network code for holding character info, quest info, logins and so on. The schema for these are too complex - around 30 tables. But as someone has pointed out before me you can't customise the UT network code as far as I know to do this, and maps can only hold 64 players. How would you connect to (presumably) a mySQL database in UT? Most servers would have up to 100 people per zone too. Load balanced unreal server mod anyone? Didn't think so

- In game UI - menus for character screen, quests, so on. Easily done with UT2003 so UT3 should be doable

- In game IRC for the chat channels. I think they're ditching the ingame IRC in UT3, not even sure it would be usable for the mod anyway, but the idea was you have your own IRC server, with channels per realm/region, trade channels and so on. So if you had a realm called Server1, you would have channels like #Server1_RegionX_Trade. UT would connect to these.

- Huge maps. There's no map streaming in UT2004 but is coming in UT3 so this would be easily achieved. Probably the biggest hit and miss factor with an MMO is how interesting the zones are and load times of them. With UT3 obviously there would be no load time much like WoW (one of many reasons I think it's so popular), and no limits to the map (some system would need to stop you falling off the world though).

- Character graphic customisation. No idea if you can do this dynamically in UT2004/UT3 but it would be required for the game. You would need regular character models and weapons updates, and lots of items to start with. "Epics" as the WoW players know them, or green items, blue items and so on. People want their character to look impressive with rare items, and starting look like peasants.

- Classes - inbuilt into UT via the Class object so this wouldn't be a problem.

- Spells/talents for characters. Ontop of the usual run-and-gun you might want your different classes to have different "spells" they can perform. You would need to spend A LOT of time in my reckoning making shaders or particle effects for these. You could of course get around this by just having classes able to use different weapons, and some kind of talents that improve them.

- The mob AI. Not really a problem as UT has lots of nice AI built in at your disposal in unrealscript. It would make a nice change to have some clever AI in an MMO, instead of the usual attack at a certain range and follow you that features in WoW and so on. Tabula Rasa and Hellgate:London have tried to do this with their MMOFPS. I've played both in beta and in my view they haven't achieved it, both are dull after 10 levels.

I think the biggest hurdles would be the server and the character models. Hurdles that are way too big for a mod team in my view. It would definitely be fun to create the world however, as a single player game. Maybe in future versions of Unreal/HL2/id Tech/Crytek they will make their servers MMO capable.

Having pee'd on the fire, I'd be interested in joining an MMO mod project if one ever emerges. Just send me a pm :)

[DGUnreal]

Elmur_fud;25009270']
Ok I see nowhere other then one line in your last post that speaks about it being client side that would have a issue with the number of polygons. Admittedly I did miss that. ...


I talked about client side loads and design for most of my replies ever since my response to the "UT99-ish poly counts". Everything mentioned from then on relates to client-side, and is mentioned considerably more than once.

Don't worry about it, no harm no foul. :)

A UE3 Server technically should be able to handle a massive number of players, hundreds if not thousands, if the appropriate areas of code are upgraded and depending on the Server hardware, backbone, and how much other dynamic actor load is in the maps (UE runs a lot of stuff server-side regarding dynamic objects).

Client-side however, is going to be limited to current hardware rendering and CPU load abilities for total polys and map level of detail (physics, effects, etc.), so that is where player count is going to make a difference for anyone attempting to get UT3 next-gen quality to their RPG world.
Which is why I gave the comment regarding the fact that UT3 quality won't be possible if there are hundreds of visible players, and that a mod would have to drop back to lower poly standards.

EQ2 isn't UE, afaik it is a custom engine, although there are a few MMO's based on a modified UE.




MMOs in one form or another are clearly the future and popular ...


Large team warfare games are becoming very popular. I would like to see a 64 player multi-team battle game with UE3. I have a mod idea for this with some of the work already started on, but finding people to work on it is difficult, and game studio work tends to use up a lot of my time.



... How would you connect to (presumably) a mySQL database in UT?


I would tend to look into a simpler database system. Having to run a full SQL server to manage this seems overkill to me.



- In game UI - menus for character screen, quests, so on. Easily done with UT2003 so UT3 should be doable


UT3's 3D menu system is awesome.



Huge maps. There's no map streaming in UT2004 but is coming in UT3 so this would be easily achieved. Probably the biggest hit and miss factor with an MMO is how interesting the zones are and load times of them. With UT3 obviously there would be no load time much like WoW (one of many reasons I think it's so popular), and no limits to the map (some system would need to stop you falling off the world though).


Although I haven't worked with streaming yet, my understanding of it from UDN3 is that it isn't what a lot of people are perceiving it to be. I can't cut/paste from UDN3 yet (it isn't public at this time) so here is a bit of a paraphrase of information and my understanding of what Epic is doing.
For example, if you are essentially splitting a large world into multiple maps, the streaming will be utilized more for optimization reasons than the ability to create a massive planetary world. A 512kx512k single-map complex world is a massive amount of asset content to load at once, so streaming allows the engine to load only portions of world assets related to that stream level.
So split your world into smaller manageable chunks. This will also allow multiple level designers to work on the world simultaneously.
If you choose to go with Distance-based Streaming (there is also Kismet and Volume), you create a persistent level and then a set of streamed levels, the streamed levels use a property LevelStreamingDistance.Origin.X/Y/Z that essentially defines what distance the streamed map is located from the persistent origin and at what distance from the player it is to be loaded. Loading may cause hitching, so be aware of that. Essentially the LevelStreamingDistance is like a tile offset to where this map is located in the full world.
Since this is an origin offset, and there is a limitation to the distance from true map origin that you can go (512k in a standard map) before you exceed the coordinate data size internally and within the network packets, and before things in the engine start to go haywire (physics etc.), there will be a maximum limit to the size of the entire world. I haven't seen anything published yet, but I believe the streaming offset is limited to world space, meaning that the maximum you can still make a multi-map streamed world is 512kx512k or about 10 sq kms. So this would mean that streaming still limits you to a total world of 512kx512k, the same as a single map. The streaming only allows you to split up a complex world into manageable chunks, most likely usable more for consoles with their limited RAM resources than PCs. Whether there is a workaround or whether Epic has expanded this limit in the later builds, I cannot say (I would have to search).



Character graphic customisation. No idea if you can do this dynamically in UT2004/UT3 but it would be required for the game.


This should be possible. The issues are the time to create all of the additional character "upgrades", making sure they can attach to all players, and then tracking the upgrades that every player has (persisting the data between online sessions).



Spells/talents for characters. Ontop of the usual run-and-gun you might want your different classes to have different "spells" they can perform.


This is too DnD for me. :)
Personally, I would rather see a well done battle style game, where there is a better balance between players and their upgrades, so that while seasoned players can get cool upgrades, it doesn't shift the power so greatly that newbies are put off because they die within two seconds after every spawn, and end up leaving frustrated.
The battle scenario mod I designed tackles a lot of this issue.



I think the biggest hurdles would be the server and the character models. Hurdles that are way too big for a mod team in my view. It would definitely be fun to create the world however, as a single player game. Maybe in future versions of Unreal/HL2/id Tech/Crytek they will make their servers MMO capable.


Twitch FPS and MMO are different enough in design that I wouldn't expect to see it even in future UE versions. Currently it requires engine modifications, there is no reason to assume that will change. Even if/when hardware improves enough to allow UT3/UT4-LOQ 64 or 128 player UT matches, the game design is still different enough that an engine optimized for the game style is still required.

Since UT3 does support 64 players, I still feel that a massive battle scenario game would be a better candidate than an MMO. The main issue is still the fact that with that player count, you would not support UT3 level of quality.

[re8]

NeverWinter Nights 1 had a mod that allowed their player-ran servers to use SQL.
They also had large worlds that were a series of smaller maps.
UT3 can handle much larger maps, esp if there are not many NPCs.

[GeoffW]

I would tend to look into a simpler database system. Having to run a full SQL server to manage this seems overkill to me.


i would have thought MySQL was pretty simple. got any suggestions?
do you believe that it would be a big task managing a MySQL server?

[[GU]Elmur_fud]

mysql would be extremely easy, ...even easier is just go download mysterial's source code UTRPG and see how he did it there. If a non-programmer like myself can learn to admin/setup a mysql database anybody who can even code halfway can.

I don't think a MMO would be waste of time. I personally think that huxely and endless saga look awesome and sound awesome.

SS using the UE3
http://www.endlesssaga.com/scr_02.htm

http://www.endlesssaga.com/

Not having spells in a fantasy mmo is a sure way to not have people playing your mmo.

Since mmo's generally are PVE (player verses enviroment) mainly and online FPS's are PVP (player verses player), your not dealing with being overpowered. That and the higher level people are generally playing elsewhere where the monsters are there lvl means newbs don't generally get overwhelmed.

[DGUnreal]

i would have thought MySQL was pretty simple. got any suggestions?
do you believe that it would be a big task managing a MySQL server?


Using SQL is relatively simple, IMHO it is just overkill. For something as simple as tracking MMO states, even xml would probably work.



Elmur_fud;25010934']
I don't think a MMO would be waste of time. I personally think that huxely and endless saga look awesome and sound awesome.


If this was aimed at me, I'm not saying that it is a waste of time, only that there are many considerations and those considerations are a lot more work than the majority of people realize. :)


Elmur_fud;25010934']
Not having spells in a fantasy mmo is a sure way to not have people playing your mmo.


You assumed it had to be fantasy. ;)


Elmur_fud;25010934']
Since mmo's generally are PVE (player verses enviroment) mainly and online FPS's are PVP (player verses player), your not dealing with being overpowered. That and the higher level people are generally playing elsewhere where the monsters are there lvl means newbs don't generally get overwhelmed.


That depends entirely on the game design.

[MonsOlympus]

Well I didnt read this whole thread so I dunno if its mentioned! Anyways I would just like to say even if you can get a working MMO mod up and running you'll need massive server power and for me as a modder that would hold me back from attempting it. Its not because its out of reach but for something you give away for free I wouldnt invest that amount of money, for retail products it makes sense.

Youve also got realms in a MMO so not everyone connects to the same space because of ping concerns. Honestly I think massive online worlds are best left to people who can make money off it, then again secondlife is free to some extent but there are payment methods and a whole economy to go with.

I would consider this very very carefully, while the mod might be very popular you'll need to invest even more for upkeep as player numbers go up. It could lead to the possibility of going retail eventually but I wouldnt have that as the only driving force behind making a mod.

Sorry to sound alittle negative but I just believe Im being realistic, if you guys can pull it off more power to ya :cool:

[DGUnreal]

Its not because its out of reach but for something you give away for free I wouldnt invest that amount of money, for retail products it makes sense.


The biggest expense, other than long-term operational costs, is still going to be the UE3 license (~$500k).
You simply cannot do the level of MMO that most people are talking about on this thread without modifying the engine. And without high long-term costs. There are numerous issues and hard-coded engine setups that I have already mentioned, such as total player count limits (64), client-side polygon render limits, total streaming world size (512kx512k), plus the fact that every client connected to one server must have the same stream map set as the server at all times (you cannot preload the full stream set on one server, so one server cannot run the entire world), then if you are planning on tens-of-thousands of subscribers you will need a decent backend.
Which is why I have been mentioning things that are more small-scale, such as a 64 player limited world, smaller server requirements for backend, etc.

*edit*

IMHO, what should be done is:
- A large single 512kx512k map on a single server
- Fantasy or whatever style of design you want
- Low-poly (2k characters) and small 512x512 texture limits and use NMs and Shaders to pull up the quality
- MMO style game play
- Limited to 64 players

If this was popular, additional realms could be created as one map per realm, running on one or more servers, and hop between servers through some sort of map portal. Build up enough interest, and see if it goes somewhere, perhaps try out for MUC.

Anything else requires engine modification.

[[GU]Elmur_fud]

If you pay the 500k for an engine license sell the game.

You could get the server power you need for about $400.00/month (best price I have found).

I don't think it is impossible to do without a licence. Unless code is that much different then UT2k4. ...but you would need to be damn good to make a mediocre mmo that runs semi decent on good machines.


Using SQL is relatively simple, IMHO it is just overkill. For something as simple as tracking MMO states, even xml would probably work.


Well there are good reasons why all the big wigs (WOW, Everquest2, Lineage2, ROSE, Guild Wars, Ryzom, etc) that I am familiar with use mysql. The bigest reason is probably database size. Official WOW for instance probably has about a 1/2 gig database, that is just a geuss but I only ever got about half the content working and and mine was 238MB. The items table alone had almost 1 million keys, each with about 50 values, and I know I didn't have them all. I'm not saying it couldn't be done otherwise but to me mysql looks like the best route.

All that info is another reason why I think a license would be almost a essential.


Honestly I think massive online worlds are best left to people who can make money off it, then again secondlife is free to some extent but there are payment methods and a whole economy to go with.

Secondlife?....
http://gamerzunlimited.net/mtbpman/smilies/197.gif

Well that abomination aside. There are lots of free mmo's out there. Free to dl free to play.

Here are a couple I play or have played.

Flyff - http://flyff.gpotato.com/
Last Chaos - http://lastchaos.aeriagames.com/
Sword of the new world - http://www.swordofthenewworld.com/
Rappelz - http://Rappelz.gpotato.com/
Silkroad - http://www.silkroadonline.net/
Archlord - http://www.archlordgame.com/

Then there is this site which has a very partial out of date list but it is already there:
http://play-free-online-games.com/games/games_all.html

Edit: Those games make money off of cash shops. Where you buy stuff to make your ingame charecter better or to make the game easier to play.

[DGUnreal]

Elmur_fud;25011408']
You could get the server power you need for about $400.00/month (best price I have found).


For a mod project, subscriptions for playing would almost be a necessity.


Elmur_fud;25011408']
I don't think it is impossible to do without a licence. Unless code is that much different then UT2k4. ...but you would need to be damn good to make a mediocre mmo that runs semi decent on good machines.


I don't think it is impossible either. It would have to be simply scaled back.
It could be fun just to be able to roam around a 10km detailed world. The map design would have to be such that occlusion was well worked into it. Mountains, rivers, lake, small towns, etc. Throw in dynamic weather, some dynamic world objects, ...
If the scene was set back in medieval times, then the lack of high-speed vehicles will make the world seem larger.


Elmur_fud;25011408']
Well there are good reasons why all the big wigs (WOW, Everquest2, Lineage2, ROSE, Guild Wars, Ryzom, etc) that I am familiar with use mysql.

Yes, for a true MMO it would be necessary to do this. With what can be pulled off on a mod, I highly doubt it would be anywhere near as popular as EQ for example, so a SQL server, although the best way to go for database creation and maintenance, might be overkill.

Another big issue for a free MMO will be cleanup and maintenance. You will be guaranteed to have spammer signups, people that go in just to try to break the rules and mess with other players, etc.
This is one reason why if it was pay-to-play you'll have less of this. But most people won't pay for a 64 player mini-MMO. :)

[[GU]Elmur_fud]

Hellgate London's implementation is intriguing. The cities are true mmo but the actual gameplay is party based. So you aren't sharing kills with other people, maps can be smaller, and the actual client overhead is less.

Archlord has a streamlining idea that is well overdue in mmo's and that is eliminating alot of the usual in town lag by making the user store centralized in one location. Everyone just sets up there items from anywhere to sell and they get listed in the "auction house" for any1 to browse. Then you go about gameplay. No afk vendors crouding around NPC's in that game. I would have implemented it slightly differently but wouldn't we all.

The cities could be further broke up districts not loading 1 till you enter the previous section. Say like:


͏͏͏͏͏͏͏͏͏□͏͏͏□͏͏͏□͏͏͏□͏͏͏□͏͏͏□͏͏͏□
͏͏͏□͏͏͏□͏͏͏□͏͏͏□͏͏͏□͏͏͏□͏͏͏□
□□■■■□□
□□■▪■□□
□□■■■□□
□□□□□□□
□□□□□□□

Where;
□ = city block
■ = loaded area
▪ = you

That way the area surrounding you is loaded b4 you get there. This could be done potentially with zone volumes I would think, but obviously would require custom code.

[GeoffW]

Well as was said earlier in the thread a little bit at a time i think.
first off i think you start of with 2 maps and 2 servers that join together - something basic
when thats ok start looking at the mysql for keeping stats
then start looking at player count and progressively up the detail just to see where it can go what its limitation are etc

[KingLollo]

I was thinking about creating a MORPG.

Something like starting small.

Adding Melee Fights and Spells for online deathmatch. ( like good old Rune)
Then adding spells.
Then start over with party based dungeon maps, where you have to kill some trash mobs and bosses and get some loot with about 5-10 players.

Then adding some inventory stuff.
I guess it should be possible to save the inventory of ur characters on a masterserver.

I played some WoW a while ago and even if people like those MMORPG games, I think the only relevant part of those games is the dungeon raiding.

I could possible code ( would lern it.... ya i know C, C++ or Java, so UScript isnt that hard ) and create maps.

[innociv]

i'm pretty sure you'd need the engine sourc eto make an mmo..

Alot of stuff is hard-coded into the game I'm sure like it was with ut2004.

[noz]

I'm working as a level designer on a professional MMO using unreal 2.5 tech. Let me tell you, for a company it's ALOT of work, time and effort, for a mod team it's impossible.

Start small, make a few coop maps with some mobs in them.

[warlord57]

I'm working as a level designer on a professional MMO using unreal 2.5 tech. Let me tell you, for a company it's ALOT of work, time and effort, for a mod team it's impossible.

Start small, make a few coop maps with some mobs in them.

I agree with you, start small then slowly add to it and change it and it will grow. I plan to create my own RPG mod for UT3 and I am just going to start by getting the base working such as experience and smaller stuff. Once that is all working then I will move to simple skills for the mod (probably just weapon skills). I eventually will have a basic rpg mutator then I will slowly add content to it turning the mutator into a mod.

[NobodyImportant]

mysql would be extremely easy, ...even easier is just go download mysterial's source code UTRPG and see how he did it there.Mysterial didn't use MySQL. UT2004RPG saves the character data to the host's UT2004RPG.ini directly in an index of array variables which includes your name and PlayerID for authentication.

This data did not follow you across servers, so if you used multiple world servers with this approach you'd also need a way to transmit this data back and forth between them and the character server automatically while the game is still running.

I don't know if UE3 will be different, but I know 2k4 used Running.ini when the game was launched and didn't read external .ini changes. I don't remember if Running.ini pre-loaded mutator configs or if they loaded on map launch, but either way the mutator wrote the data locally at the end of the match, and without a way to transmit that data to a master server it was impossible to create consistant data across servers, at least not without a lot more code. If you're going to do that, you might as well use an external database anyway.

[[GU]Elmur_fud]

Mysterial didn't use MySQL. UT2004RPG saves the character data to the host's UT2004RPG.ini directly in an index of array variables which includes your name and PlayerID for authentication.


ooo I forgot that. Can't fathom why, goodness knows I have plenty of experience with the ut2004rpg.ini file. I have edited it on servers I ran many times.

mysql still isn't hard.

[kisk]

what we're doing with this mmo mod is starting small... why the hell would we be thinking 1000+ players/per/server in even the 10th alpha test or 3rd open beta? Yes, netcode is going to be challenging.

It'll be fun to play with though -- we've been doing a lot of research on what type of netcode mods we'd have to do and we're pretty confident.

Graphically, think about it -- this is a brand spankin-new engine... early final betas will PROBABLY look like ut99, but we're looking at a long-term approach. By the time everybody's systems have upgraded (and the fact that we'll have to learn the engine and the workarounds) -- **** it might take 2 years -- the engine will still be around testing everybody's rigs. By that time, graphically, textures/models/polys will be updated to really push the engine and players machines.

We don't want to choke everybody just yet :D

-kisk

[upL8N8]

I think you would really want to research how games like WOW allow continuously flowing maps and determine whether that's possible in UT3. I still remember the everquest system where as you moved from zone to zone, you had to go through a portal since I assume multiple servers were running each zone.

If it was possible to create a massive map, which consists of multiple sectors (or smaller maps), whereas each sector only loads depending on your relative position, I think you'd have something. This of course would all be client side.

The other thing you have to look at is the data that is being sent to and from the server. FPS games are extremely dependent on character and projectile position. For instance, if there are 3 players in the game and player A shoots a rocket at player B, but as the rocket gets halfway there, player C jumps in front of the rocket, the rocket needs to hit player C. So the coordinates of the rocket, player B and player C need to be compared constantly to determine their positions. Not to mention the coordinates of the projectile to any walls or objects on the screen.

Whereas in an RPG, when you cast a spell, the spell is guaranteed to hit its target (not counting dice rolls :-P). The path of the projectile doesn't matter because it's always going to follow the enemy no matter where he goes. In fact, the projectile animation doesn't need to be tracked by the server at all. The client computers need only know that the projectile was sent from the position of player A and will end its route at the position of player B. The start point is static and the end point is dynamic. As long as the client computers know where player B is, both player A and B's computers can draw the route of the projectile. This is much less effort on the part of the server. The server simply told the client computers what happened and where player B is, and the client computers take it from there.

For FPS games, the server needs to be in constant contact between the players and projectiles on the screen. Hence why you can only have so many players on the server at any given time. The server simply can't keep track of 24 - 48 players (depending on the map/server)

You'll also notice that WoW's servers will lag out if there are too many players in one area. This is because when one character performs an action, the other 100 players standing around him need to be notified of what he just did so they can update their client computers. If this was an FPS engine with people shooting their rockets at each other, the server would die a grizzly horrible death :(

Building an MMO off of an FPS would be an awesome idea if you could keep with the idea of having dynamically tracked projectiles, instead of the WoW type projectiles as I mentioned above. However, to do so, you need to consider how many players are on the screen, whether you can have multiple servers running the various sectors of the global map, and what happens when too many users access the same sector of the map.

Then there's the whole persistence thing, but I think using a SQL server would be easy enough for that. You just need to make sure it's efficient and fast. There are already mods out there that are doing persistant data tracking though, so no problem there.

This would be fun to play around with.

[campbullard]

Stargate Worlds is a MMORPG which will be using UE3.
http://www.play-girlz.com/wp-content/uploads/2006/11/stargateworlds.jpg

[Nawrot]

I would say start it like NWN, aim for 1-16 players small private servers. Make everything step by step.

First some RPG "engine" that means quests, items, npc's, monster, persistent world support. In same time release asset packs and maps to public to advertise and assimilate more mappers, coders, artists. Then try to get momentum for serious things.

[skipgamer]

OK, to the op cos i cbf reading through 6 pages of posts.

Anyone that knows anything about game development knows "I want to make a mmo" Is THE biggest mistake of any beginning game developer. Its too ambitious (especially for a mod) and you will end up destroying your own project with feature creep (trying to add too many features and pushing ur release too far back) and burnout (you wont get the results fast enough and you will end up giving up)

At the end of the day, as everyone else has said its much better to start with something small. Hell, for my first mod (3 people ninja mod) we just started by adding one weapon, then a character, and it grew from there.

First off write down your idea, then I suggest start implementing things roughly, then just build up from there. You WILL NOT be able to just "build an mmo" from scratch with the ut3 engine. If its even possible at all in a mod form. You've got to take into account saving characters on a server and all that, securing the netcode so its not possible to hack the server and just make yourself lvl XX. And to be a real mmo, you'd need to change the netcode significantly i reckon.

Stop dreaming, get a real programmer/scripter and tell them to start playing with ut (im not going to say its impossible, because they said you couldnt rag doll, then come back to control in ut2k4 and someone figured it out) But it will be very very code intensive if it is possible at all...

short answer... no, dont try, unless you have a Very, Very talented group of people to work on it with.