About Store Forum Documentation Contact



Post Reply 
Questions about float accuracy and large worlds ...
Author Message
wysardry Offline
Member

Post: #16
RE: Questions about float accuracy and large worlds ...
(11-27-2011 03:17 PM)Demostenes2 Wrote:  40x40km is enough for any MMO ever.
40 by 40 km is 160 sq km. WoW has an area of around 207 sq km, without expansions. Lord of the Rings Online has an area of about 77,700 square kilometres.

160 sq km is barely large enough for an island, let alone a world. Ireland has an area of over 84,400 square kilometres and even the Isle of Wight is around 384 sq km.
11-27-2011 07:12 PM
Find all posts by this user Quote this message in a reply
paviii Offline
Member

Post: #17
RE: Questions about float accuracy and large worlds ...
Just a quick correction to the previous post, 40km x 40km is 1600 km^2. This is a pretty big space.

I setup a land mass with a previous engine that was about 35Km x 20Km. I set a run speed of 6 Km per hour which is somewhat on the quick side. You can do the math to figure out how many hours it took to run across.

Paviii
11-27-2011 07:26 PM
Visit this user's website Find all posts by this user Quote this message in a reply
rndbit Offline
Member

Post: #18
RE: Questions about float accuracy and large worlds ...
since discussion is about world size i am wondering how well esenthel could scale for a space game?

since in space distances are massive i wonder how we could go about it. well.. maybe tiny tiny models and tiny tiny speeds? like 0.001 being 1 kilometer, and that might not be small enough, since space is huge place..

i also wonder what about planet rendering. other engines do lots of weird trickery like render batching or however its called (when far away objects are combined into one larger object to achieve good performance without suffering any visual defects). api of esenthel is fairly high level, would it allow achieving something reliable in that field?
11-27-2011 09:20 PM
Find all posts by this user Quote this message in a reply
Tottel Offline
Member

Post: #19
RE: Questions about float accuracy and large worlds ...
40km x 40km is indeed 1600 kmĀ², as pavii pointed out.
Also, Esenthel clearly states he TESTED with 40x40, which was stable. He never said you can't go bigger than this.
11-27-2011 10:01 PM
Find all posts by this user Quote this message in a reply
impi Offline
Member

Post: #20
RE: Questions about float accuracy and large worlds ...
So let me understand this. What you are saying is its ok to call 40x40km "Unlimited"? This should be called 'large' terrain, not unlimited. Its completely misleading.

(11-27-2011 03:17 PM)Demostenes2 Wrote:  40x40km is enough for any MMO ever.
It never has been and never will be good enough for 'any' MMO, not even for 'any' game (eg flight sims).
Most certainly it will be good enough for 'many' games and MMO's ... but not 'any'.

If you are building a realistic game with change of environments etc (Icy north and humid south etc), you cannot have snow and 40km away tropical beaches that you can fly to in a few minutes then hit a wall ... thats just ridiculous. That means a dead standard 155mm artillery shell can be fired from the tropical southern rebel island and take out a military base in the icy north ... lets not think about missiles etc.

Not every game being built is dark-ages themed (with no powered vehicles etc).

(11-27-2011 03:17 PM)Demostenes2 Wrote:  Have you ever tried to populate lets say 30 square kilometrs with reasonable content?
I would say your generator needs improving. On large terrains designers should only have to 'tweak' selected areas of interest. Also on many modern engines (I'm not sure if Esenthel has something similar as I'm still learning it) you can insert a 'reference world' (like instances of smaller worlds) .. so you build a few nice terrain outcrops etc, and insert them all over changing orientation etc ... that makes for fast editing of generated content.

The whole idea of realistic modern combat games is to have large open spaces to cross and which are natural boundaries to seperate teams etc. And it doesn't have to be land-mass : 70% of earth is oceans, and in a game its great to navigate oceans.

Furthermore, I suspect a world of 40x40km etc is going to perform like a dog when its a MMO. If its a multiplayer with 10 or 20 players that might be ok, but a MMO with hundreds / thousands of logged in players - all on one server ... I don't think so. Any MMO needs to be divided up to attempt to distribute players over more servers - so in Esenthels case multiple worlds.

On a positive note, these smaller worlds can still work though : if worlds can be 'joined' (viewed from neighbor world and be navigable - ie. a player can cross over into the next world - ie. transfer onto another server) .. then you can have say a 30x20km world (island) next to say a seamass (or a few) of 40x40 and next to that another world (perhaps a few small islands) of say 20x20km etc etc.
11-27-2011 10:44 PM
Visit this user's website Find all posts by this user Quote this message in a reply
Demostenes2 Offline
Member

Post: #21
RE: Questions about float accuracy and large worlds ...
(11-27-2011 10:44 PM)impi Wrote:  I would say your generator needs improving. On large terrains designers should only have to 'tweak' selected areas of interest.

I consider my generator quite OK:

http://forum.mymmo.cz/demo/screen4.jpg

And this was what we exactly did, we did terrain by generator and designeres were making points of interest. I know what does it mean to populate world and even making nice terrain 30sq km with lots of places of interest takes years. If I am able to do lets say 300 nice points of interest in 3 years, it is far better to fit them into 50sq km map and have "full map", where players have nice places everywhere, then 10k sq km terrain with same number of points of interest, but nobody would never find it and terrain would look empty.
I know what does it mean to make map for MMO and I totally agree with size of WOW and other MMOs. Going above this is technicaly almost no difference, but there is no point in having huge and empty world. Players want to do something in the world, not waste hours by trips and desperately searching for any dungeon. Yes, you can even atomatically distribute spawns, etc...but such games are trash, for good game is handwork esential.

Totaly different story is flight simulator, car racing, tank simulator, whatever...this is really case, where you can put it to generator and dont care about places of interest.
(This post was last modified: 11-28-2011 12:17 AM by Demostenes2.)
11-28-2011 12:08 AM
Find all posts by this user Quote this message in a reply
impi Offline
Member

Post: #22
RE: Questions about float accuracy and large worlds ...
Nice thick forest scene. Perhaps get the generator to rotate the trees randomly when placed - most are orientated the same.

(11-28-2011 12:08 AM)Demostenes2 Wrote:  Totaly different story is flight simulator, car racing, tank simulator, whatever...this is really case, where you can put it to generator and dont care about places of interest.
And in all these cases 40x40km is no good ... as we wrote before you threw your toys out of your cot.

What you design in your game is your business. What others design in theirs is theirs. They have no business telling you how to design yours, and you have no business telling them how to design theirs. Just because you chose a dense build with many "point-of-interest" within walking distance doesn't mean it will suit every game.
11-28-2011 12:59 AM
Visit this user's website Find all posts by this user Quote this message in a reply
Zervox Offline
Member

Post: #23
RE: Questions about float accuracy and large worlds ...
(11-27-2011 10:44 PM)impi Wrote:  So let me understand this. What you are saying is its ok to call 40x40km "Unlimited"? This should be called 'large' terrain, not unlimited. Its completely misleading.

(11-27-2011 03:17 PM)Demostenes2 Wrote:  ...
It never has been and never will be good enough for 'any' MMO, not even for 'any' game (eg flight sims).
Most certainly it will be good enough for 'many' games and MMO's ... but not 'any'.

If you are building a realistic game with change of environments etc (Icy north and humid south etc), you cannot have snow and 40km away tropical beaches that you can fly to in a few minutes then hit a wall ... thats just ridiculous. That means a dead standard 155mm artillery shell can be fired from the tropical southern rebel island and take out a military base in the icy north ... lets not think about missiles etc.

Not every game being built is dark-ages themed (with no powered vehicles etc).

(11-27-2011 03:17 PM)Demostenes2 Wrote:  ...
I would say your generator needs improving. On large terrains designers should only have to 'tweak' selected areas of interest. Also on many modern engines (I'm not sure if Esenthel has something similar as I'm still learning it) you can insert a 'reference world' (like instances of smaller worlds) .. so you build a few nice terrain outcrops etc, and insert them all over changing orientation etc ... that makes for fast editing of generated content.

The whole idea of realistic modern combat games is to have large open spaces to cross and which are natural boundaries to seperate teams etc. And it doesn't have to be land-mass : 70% of earth is oceans, and in a game its great to navigate oceans.

Furthermore, I suspect a world of 40x40km etc is going to perform like a dog when its a MMO. If its a multiplayer with 10 or 20 players that might be ok, but a MMO with hundreds / thousands of logged in players - all on one server ... I don't think so. Any MMO needs to be divided up to attempt to distribute players over more servers - so in Esenthels case multiple worlds.

On a positive note, these smaller worlds can still work though : if worlds can be 'joined' (viewed from neighbor world and be navigable - ie. a player can cross over into the next world - ie. transfer onto another server) .. then you can have say a 30x20km world (island) next to say a seamass (or a few) of 40x40 and next to that another world (perhaps a few small islands) of say 20x20km etc etc.

I think you are reading way too much into this...
first of all if WoW can do two continents and entertain 12 million people on 207 square kilometer, then 1600 square kilometer on EE is going to be able to do the same.

Sure not all games etc will be limited to "x" space but 1600 square kilometer is pretty darn huge, if you want unlimited, you know it is possible to write procedural algorithms. Esenthel engine can handle unlimited worlds, but it is limited like all other engines and games to the hardware, in this case...storage, which again comes to mind most space games use procedural algorithms in sorts to prevent space requirements go out of extreme proportions.

If you want clustering(eg zones/servers handling x amount of areas) on a MMO perhaps you should start looking into how these are done,I am pretty sure it is possible in EE you know, just needs planning and design(Experimentation as well).

About changing environments in large world scenario, well as was proposed, scaling down size of objects and speed they travel can make the world seem even larger. which is in fact done in alot of games.
(This post was last modified: 11-28-2011 01:24 AM by Zervox.)
11-28-2011 01:23 AM
Find all posts by this user Quote this message in a reply
Demostenes2 Offline
Member

Post: #24
RE: Questions about float accuracy and large worlds ...
(11-28-2011 12:59 AM)impi Wrote:  Nice thick forest scene. Perhaps get the generator to rotate the trees randomly when placed - most are orientated the same.

Especially for performance reasons LOD is billboard. This scene has 28k trees on 2000*2000m terrain. If put some random rotation on tree, number of billboards will drastically increase.
11-28-2011 01:58 AM
Find all posts by this user Quote this message in a reply
impi Offline
Member

Post: #25
RE: Questions about float accuracy and large worlds ...
(11-28-2011 01:23 AM)Zervox Wrote:  first of all if WoW can do two continents and entertain 12 million people on 207 square kilometer, then 1600 square kilometer on EE is going to be able to do the same.
Impossible.

WoW has thousands of servers to host that number of players. Esenthel MMO has one server for the world, so you're not going to run thousands/millions of players on that.

(11-28-2011 01:23 AM)Zervox Wrote:  If you want clustering(eg zones/servers handling x amount of areas) on a MMO perhaps you should start looking into how these are done,I am pretty sure it is possible in EE you know, just needs planning and design(Experimentation as well).
Now you're on the money. In order to do any MMO with Esenthel you would need to be doing this (else it'll be a MO). Theres two main approaches here :
1) Throw multiple servers at a world to share the load. This means you would not be using Esenthel MMO as you'd be redoing most yourself.
2) Logically breakup the 'world' into smaller worlds - each on its own server (like current Esenthel design). This also allows easy (future) growth/expansion (new quests etc) of the game. Which is why I added a feature request to have visibility across worlds - then each region would have its own server to handle the battles and number of logins. That would also overcome the 40x40km limitation ... just add servers (and therefore landmass).
11-28-2011 02:11 AM
Visit this user's website Find all posts by this user Quote this message in a reply
wysardry Offline
Member

Post: #26
RE: Questions about float accuracy and large worlds ...
Why some designers want very large areas or how they would fill it with content is beside the point.

Commercial games with maps covering well over 1600 sq km have been made. Daggerfall, the largest at 161,600 sq km, was released 15 years ago so modern hardware can handle them if the software is written in a way that avoids precision issues.

The feature list for Esenthel says:-

Quote:Streaming of Unlimited Sized Worlds:

Esenthel Engine supports streaming world data, which means that during the game only some parts of the world are loaded into the memory, while other parts are awaiting on the hard drive. This allows you to immerse into virtually unlimited sized worlds, being limited only by the size of your hard drive.

This seems to say that it can handle any area you can fit on a hard drive, which apparently is not the case. It may well be possible to modify the code to avoid the floating point accuracy limits, but the need to do that is not mentioned.

The features list is one of the first places that potential users check when comparing engines, so it is important that they are accurate. Unfortunately, in my experience, they rarely are.
11-28-2011 02:16 AM
Find all posts by this user Quote this message in a reply
Demostenes2 Offline
Member

Post: #27
RE: Questions about float accuracy and large worlds ...
(11-28-2011 01:23 AM)Zervox Wrote:  If you want clustering(eg zones/servers handling x amount of areas) on a MMO perhaps you should start looking into how these are done,I am pretty sure it is possible in EE you know, just needs planning and design(Experimentation as well).

This has nothing to do with theoretical limit of world. Once you have proper streaming, there is no difference (except hdd space) if you have 20 terrain tiles, or 2000 (from the point of client). Real problem is network and especially arbitration. So problem is on the server side, not client. It can be solved, but this is totally different story and actually there is no MMO on the world, which implements necessary technology. So this is reason, why they have cap like 10-20k players on one instance of world, more is impossible to handle by conventional techniques.

Btw, i remember Dagerfall and i NEVER traveled through that generated huge terrain. It had no sense, it was better to travel from city to city or dungeon using fast travel. Yes, it was possible to go for hours through terrain, but you would die of boredom, there was NOTHING.
(This post was last modified: 11-28-2011 02:28 AM by Demostenes2.)
11-28-2011 02:23 AM
Find all posts by this user Quote this message in a reply
impi Offline
Member

Post: #28
RE: Questions about float accuracy and large worlds ...
(11-28-2011 01:58 AM)Demostenes2 Wrote:  Especially for performance reasons LOD is billboard. This scene has 28k trees on 2000*2000m terrain. If put some random rotation on tree, number of billboards will drastically increase.
Thats a lot of trees ... good occlusion tools will help.

The best solution for treescapes like that (to randomize them) is a directional impostering system - so the lowest LOD will be a billboard which draws the tree depending on the angle its orientated.
11-28-2011 02:28 AM
Visit this user's website Find all posts by this user Quote this message in a reply
Demostenes2 Offline
Member

Post: #29
RE: Questions about float accuracy and large worlds ...
(11-28-2011 02:28 AM)impi Wrote:  Thats a lot of trees ... good occlusion tools will help.

AFAIK there is no oclusion tool on EE? Only chance is to implement something of your own? And trees are transparent anyway, so no point using them as occludor. Only chance is to use terrain, or mesh rocks as occludors.
(This post was last modified: 11-28-2011 02:32 AM by Demostenes2.)
11-28-2011 02:31 AM
Find all posts by this user Quote this message in a reply
impi Offline
Member

Post: #30
RE: Questions about float accuracy and large worlds ...
Yes no occlusion ... but its on roadmap. You would probably need it to get good response with multi-players on that terrain. Terrain would be best occludor in those scenes - lots of hills and valleys there.
11-28-2011 04:19 AM
Visit this user's website Find all posts by this user Quote this message in a reply
Post Reply