Space Sim (how big is to big?)

#1 How many is too many options is hard to say. The more you have, the longer it takes to complete the product. Think of it in terms of cost - how long will each feature take to implement? Along the same lines, how much would you have payed your programmer(s) working on the feature during that time? Is the feature worth that price?

As far as polygon counts, having 2000x1200 polygon models means you need to render around ~2.5 million polygons every frame, or have roughly a fill-rate of 144MTri/S. That's simply not going to happen on the average consumer computer - you'd be fortunate to push 10MTri/S on a GeForce1. 2000x300 is just 600K polygons per frame, meaning you need roughly a 36MTri/S fill rate. These are all worst-case numbers of course, you're actual numbers would likely be somewhat less.

Another note: balancing 10 playable races is going to be nigh impossible. IMO you should tone it down to at most 5, and even less if possible. Also, I think that 2000 ships in one area would require quite a lot of A.I. and physics processing time - not to mention all the special effects: weapons fire, engine trails, coronas, lens flares, etc. That's a steep requirement that's going to be tough to reach, if at all.

#2 Sound to me like your game is being designed by a marketing department...Not game designers...I meen you are describeing your game with a "feature list" of all things.

Sorry I don't meen to be harsh...but haveing X number of races doesn't make a bit of difference to me if they all play the same, have the same units (even IF they use different textures/models), and require the same stratigy to defeat each other...It would make it easy to balance things out if the races and units were given the "game token" approce..Example: a scout ship would fly, fire the same weapons, have identical armor rateing accross each of the races, even though each race uses it's own 3D model of this craft...well if this is the type of thing you want to do...why even have different races...why not just color code them differently (and in this case why stop at 10 races?).

I only say this because your "feature list" indicates that the races will all have the same number of units...which indicates, to me, that the game would develop along those lines...Instead, I think, it would suit the game better if you only concentraited on a few races and trried to develop a "rock-paper-sissors" approce...

*one race could have mastered light technology...they can use it for weapons, shields, cloakeing devices, etc...developing further they could be a more "specialised" race with many units specificly designed for a single perpose....

*another race could have mastered gravity technology...they can create gravity pools, mini-black holes, etc..thier weapons are gravity baised...they may only have a few different units..but these units can be used in a large variety of ways...

*Yet another race has highly developed nano-tech technology...to the effect that thier units are constructed from millions of these tiny robots...these units can thus reshape themselves, combine with each other to form larger units, split appart, etc...they may only have 3 "Base" units...but like Lego toys...they can build a nearly infinate array of different specialised units

There...only 3 races...but vastly different...different number of units (the "light" race could have say 15..the "gravity" race 5...and the "nano" race only 3(with the ability to combine them, createing more))...Honestly something like this would be VERY difficult to balance gameplay wise...but the results would be worth it

#3 About the Poly count. Even if the ships have 1200 polygons, you only want to show that many when the player is very close to it - when the ship is so far away that it's only 3 pixels long, it doesn't make sense to show that many polygons. Use different levels of details (L.O.D's) for the models, the further away they are, the lower the detail on the model. That would reduce the poly count by a huge amount.

#4 Well, you're still presented with a few problems to overcome.

You probably can't use a dynamic LOD scheme, as you'll kill any T&L performance. Plus for 2000 objects, you'll be losing a great many CPU cycles to the while the LOD algorithm thinks - there are a few tricks around this, such as group/area schemes, prorated calculations, etc.

If you go with static LOD, you've got to deal with what... three times the storage requirements? Five times? With so many ships, you're also going to be at risk of swapping the stuff from the HD or pagefile, unless almost all of those 2000 ships are the same.

Certainly not impossible stuff, but... you have to realise that there are reasons you don't see stuff quite that huge; nobody can afford to alienate the common gamers - the ones with 1 or 2 year old systems (unless your that small company making the Next Doom).

(Now playing: Squarepusher - Music is Rotted One Note)

#5 Yea its a features list to a point. In the end its what we would like to do. What doesn't work or is to time consuming we will leave out or change. only a few of the ships are over 800 polys and they are going to be limited.

The races will all play differant. The weapons differant and the styles ships differant. we are still working on the game mechanics we hope to use the V12 engine.

#6 OK i think i can help a bit here mate as far as polys are concerned the average poly count in a fighter ship in the likes of Starlancer is about 600 polys some are more and some are less a LOD is also used in this i have built ships that will fly in that engine with 2000 plus polys but it groans on my poor PC, most 3D player models in most games currently run at or around this count, but bear in mind that if your project takes 3 years then you could probably increase it to account for advances in hardware and graphics cards i myself love space sims but wish they could be less linear than what most are.

#7 I have been thinking about a space sim and one of the biggest features that I look for in such a game is the ability to design your own ships in detail.

At the sorts of ranges that you will typically be engaging (if your taking anything like a realistic approach) most ships will much be less than 2 or 3 pixels. My approach would be to represent each ship with a box and label like a HUD. The ship that you are currently tracking will be blown up on an MFD.

I agree that 2000 ships is too many for AI and in addition I don't think giving orders to large numbers of ships in single player or LAN games is fun.

Don't know if any of that applies to your game though.

#8 I have seen great space crafts with less than 300 polygons.
This can be hard at times, yes. With around 600 polygons you can have a very good looking model.

Most of the detail should lie in the textures. Have you seem modeled cars in many video games? They are almost boxes at times! But the textures are photorealistic, which is the key.

V12 has a progressive mesh algorithm for distanly viewed models. You could make a 2D map system for ships that are too far away to see, and view culling ranges, maybe ones you can change depending on computer performance.

Designing your own ships in detail is a very interesting idea. Maybe some sort of lego block type modeling program? Place plates on your ship, guns, that sort of thing.


>Sorry I don't meen to be harsh...but haveing X number of races doesn't make a bit of difference to me if they all play the same, have the same units (even IF they use different textures/models), and require the same stratigy to defeat each other...

The rock paper siccors idea is good.
Maybe create a 6 or 9 way rock paper siccors?
Everything for the races should be different. Take a look at starcraft.

>*Yet another race has highly developed nano-tech technology...to the effect that thier units are constructed from millions of these tiny robots...these units can thus reshape themselves, combine with each other to form larger units, split appart, etc...they may only have 3 "Base" units...but like Lego toys...they can build a nearly infinate array of different specialised units

I like this idea, how about have different weapon/armor choices for units to make it more dynamic?

>If you go with static LOD, you've got to deal with what... three times the storage requirements? Five times? With so many ships, you're also going to be at risk of swapping the stuff from the HD or pagefile, unless almost all of those 2000 ships are the same.

A lego like system could help here. There can be many ship parts that are re-used in each model, and each ship has a different configuration, so you can end up with millions of combinations, and infinite general combinations of the ship frame.

I don't see why some models would be so many polygons unless they are really big, like a space station or something. If you have good modelers, they can probably do wonders with 600-800 polygons for larger ships(cruiser types), 300-600 for smaller ones(fighter crafts), of course, depending on your design. Some space sims have smooth designs to save polygons, they generally look as good as others, and the others have more rough shapes, more polygons....

Depending on the engine and other things, polygons that are exactly paralell to another(sorry, that is the best way I can describe it) can be combined with that one, creating a larger one, and the area that isn't suppose to be there can be transparent. It does wonders with cylinder objects where the tops and bottoms can be one polygon.

I hope I helped!

#9 That is fine and dandy about the polygon counts and the spacecraft.so tell me how are going to add hundreds of spacecraft plus infantry mechs and ground vehicles?
i mean you will probably need death animations for infantry different explosions for each craft and vehicle.
but never the less best of luck to you guys.

#10 2 years later and starting from scratch. I learned some valuable things. Most importantly I learned you can set your sites for the stars when you don't even have a ship yet. We started this program with modelers and texture artist and no programmers. We set our sights for a 2003 release. With what at the time seemed like high graphic reqs. For 6 months we worked hard and came up with no programmers and the team fell apart. I have assembled some of my old team and im looking for programmers. This time ill wait till I find them to rush in to the project.

My question is the same though how big is too big .

We want a space sim on a level of size and intricacy that has not been done yet. I know there was a game similar in idea to ours but none were near the scope. Some say we should focus on single player and make that massive and use what time we have left for multi. Some think we should focus on multi and do single player as a way to set the story for multi.

Right now we are focusing on multi from a design point of view. Originally we wanted huge space battles off set with fps battles in the corridors of the ships and mech battles on the ground. Looking back I think we were out of our minds lol. This time around we are going after the huge space battles. If we have time we may go back and see about the other stuff.

Huge space battle may be a daunting task in its own right. Getting 1000's or even 100's to work correctly may be near impossible. Let alone getting that many players on to a sever with out extreme sever lag or cost.

The thing we want to focus on is huge space battles even when the player base is limited to 16-128 players. From an idea perception we have come up with allowing the players to use "rank points" that they earn in the game to acquire wing men or npc help. Here's and example:

Player 1 spawns in and starts with 5 rank points. He spends the five to buy him self a wing man. The 2 of them attack an enemy out post each scoring 10 kills. The player would get 10 rank points. ( player kills + npc kills / npcs). using those 10 when he returns to base he can rise in rank , giving him access to new ships, or he can get more wing men.

Player 2 spawns in and starts with 5 rank points. he saves his points and goes on defense. He kills 20 fighters and took out a destroyer. he gets 50 rank points (20 for the fighters 30 for the destroyer). Using his rank points he rises to captain and takes a carrier(40 points for the example) and 10 fighters.

the 10 fighter are npc by default. However if a player chooses to they can spawn as one of the fighters that he still has on his deck. This would allow for new players to get closer to battles.

The point in this that when you have 32 players with fighters or cap ships you could theoretically have a battle of 300 or 3000 ships. I know very little programming but it seems feasible to program several AI attitudes. so if you have 30 different AItudes then you could have a far more interesting battle in stead of all the AI acting the same. If we have time we may not stop at 30 but that is just a guestimate.

The some of the other things we want to work on is the size of the game world and evolution of tech. With being pretty much a team based game in space we don't want REALLY long travel times. Yet we still want to give players a chance to prepare. so we figure a grid system will work to set up the galaxy. on small grid you could limit the time to cross the map like 15 mins (long time but ill explain). On a larger grid maybe 30 mins. also with it being space a 3x3 grid would really be a 3x3x3 grid. If you spawn the systems at random on the grid it would provide a use for scout ships and give the players a little more time to prep for war.

Now I will explain why I seemed to have contradicted my self. We really don't see doing this game with out a resource system. so much like in a real time strategy, resources must be located and harvested. using the resource the team can advance the level of technology or production of ships. That could be done with a vote majority rules. when a team has raised their engine tech high enough they could fold space allowing for 5 or 10 secs of travel across the map instead of 15 or 30 mins. Collecting resources or shortening travel time is as easy as flying a construction ship to a point in space and setting up an outpost or collection facility.

The goal in the game is to drive your enemy from known space. several game types in thought. catchy names are just how we refer to them

Space arena -Arena style 1on1 combat

Typical DM - pick a ship and kill. increasing the size of your ship as your rank increases. small maps with no tech building

David vs. Goliath team DM - one team can only pick fighters while the other team can only pick cap ships and it switches after one team wins. small maps with no tech building

Typical CTF[b] - a flag is set up in space and you have to grab it and get back to base. small maps with no tech building

[b]Capture the planet - the server sets the number of possible planets and 2 teams try to take all the planets with orbital shelling and such. Resources required to develop. Fairly large maps with tech building.

Intergalactic War - dedicated servers could set up a multi race/faction war. Giving the players 3-10 different sides to join. Similar to CTP orbital shelling and resource play big part in out come. Huge map with tech building. (Just a thought it's what we wanted to begin with but is a high goal)


Those are some of the more interesting idea that we have. I would like to get some input on what you guys think. :) remember we will be using the torque engine and we are in need of team members preferably from north east Ohio but anywhere will work. I look forward to your response

#11 Ok take this from a Developer thats going to be releaseing Two different products later this year.

First of all like everyone mainly says. Start small and then work your way up. If you have more time and resources just add onto what you have.

My what was once small company started with a team of 10 people. We have now grown to a team of 50. We decided to split the team 25 / 25 to complete both projects in a timely manner.

One of our products that is in development and about to enter testing is a MMORPG. (Yes Another MMORPG. One that is very different then others out there) Anyways the game started very simple. 1 race and a nice world of some 25 to 30 different zones on it. 15 or so different monsters and a handfull of items and weapons to start with. Once we had the above completed or near completed we began to start expanding on it. Adding another race. Then another. Now we have 5 different races and a world that spans 100 different zones and about 5 times the weapons and items we planned at start.

So from experience you need to start out small and dont set your hopes and dreams so large to start with. One small step then work your way up.

Theres a saying in the industry that goes like this. Your first 10 games will suck.

#12 Yes! I've just made my number 8 game on the PC... Two more and I'll make a game thats cool! :)

Now, seriously... The rank point thing sounds very good indeed... Although the multiplayer doesn't attract me too much (I prefer the single player experience)...

But don't underestimate the requirements of the AI... You can program only one "attitude" as you described, but making 100's of units use that AI is going to suck processing power like a sponge... And consider also the weight of the special effects... A space game should have LOTS of special effects!

#13 Doh, I've only made 3, and I had high hopes for my current project. )C:

#14 this is what we hope to accomplish in the end. To me its like running a 5 km race. If you know before you start that its a 5 km race you can pace your self so that reguardless of of who wins you will at least finish the race. If you do that same 15 km race and you don't know that its 15 km then you may burn out long before you get to the end.

we are going to work in Phases/steps.
Phase 1- (kind of an over view)
step 1 is to find some programmers (2or3) that can do what we need.
step 2 is to get the foundations of the game laid out such as the physics.
step 3 is to get a model in game.
step 4 is to apply some control over the model.

When we get this far i'll worry about the more advanced things. to me thats a small mildstone for celebration :)

Game design: secrets of the sages is a great book i have read it thru several times. lol it was when i read that book that i realized iots alot more then "i wanna make a game" BOOM "im a millionair" i have no delusions this will be the greatest sim of all time i just really have to give it a try it. If i don't try to give it 120% and i compromise i'll look back and say we should have done this or that.

AI its been my understanding that you can program several levels of AI into any game. thats what i ment in terms of the AI. As for processer reqs we aren't aiming for any real release time (way to early for that) but we are aiming for the high end comps near 2005. i know we can't alienate the low end consumer but in truth if we aim for that but don't finish till 2006 we will be right on target :)

#15 "...size matters not."

Why do you immediately assume that having a space sim with thousands of units in action at once would be fun to play?

Before you spend another two years thinking about how cool it would be to have a galaxy-wide battle of millions of fleets from hundreds of thousands of worlds with billions of ships all fighting at the same time, perhaps you should spend a few minutes thinking about what the player gets to do.

What does the player get to do?

Why are those activities fun?

Would anyone want to do that for any length of time?

If you can answer those questions to your satisfaction, I think the answer to your "how big is too big" question will be a lot more meaningful.

--milo
www.starshatter.com