Sci-Fi Tech Starfall

JagerIV

Well-known member
Even more usefully, a scale of 424 km diameter reduces the surface area of the Earth from 510 million skm^2, to "only" 565,000 km^2. Thus, for roughly equivalent work the planet can have 1,000x the density of interesting stuff vs a real scale. 565,000 is still a huge map. As point of comparison Skyrim seem to be roughly 40 km^2. So, the earth would be 14,000 skyrim maps, as one world among many. This is not manageable as play areas.

However, its not quite that bad. First, we have land use: most land simply would have nothing interesting.

1920px-Global-land-use-graphic.png


First you have 70% of the surface area is ocean, which as a space game should be nothing but scenery for plane or space travel, and to break up land areas. So, that already reduces it to 163,000 km^2 of land, or 5,641 local maps if each was roughly 5 km wide (so, roughly an hour to walk, or 3 minutes driving.

Of those, roughly 30% of that land is wastes of ice or deserts that there's not much reason to go on. That's territory the player really shouldn't be traversing with any regularity, and if they do its quite acceptable that there's nothing there. A player might fly to the south pole, but if you want to walk there, your going through semi-procedural areas that have nothing in it. The challenge of those kinds of terrain is in fact the nothingness there.

That's down to 114,000 km^2 of actual territory that needs to have something in it. Or about 4,000 local maps. Most of these should be procedural with 1-2 minor landmarks. By which I mean something like a hill, or large ball of yarn. So, of the 2,000 odd forest/wilderness maps, it would have its set forest biome, a notable local cliff, the programed forest road and river to connect to the other maps, and the rest is nearly purely procedural. Or maybe semi: the smaller the "biome" seed, the less effectively procedural it is.
 

JagerIV

Well-known member
I think its important for the procedural setting to be as minimally procedural as possible. Set locations means exploring is finding real places, and thus a shared experience that can be discussed. Planet x having a randomly generated massive canyon is a thing you know was randomly generated by the computer in your game world. Cyan Canyon placed on planet Nectar, where it is always 100 km long, 10 km wide, and 30 km deep, is a location you can talk about and recommend to people. Its a shared experience, or potential shared experience.

Differences from procedural generation filling in the gaps provides more discussion, especially if there's some way to tie in how thing turn out. And many are unlikely to have that dramatic an effect: if procedural filling in changes the placement and number of trees in the vast canyon, that probably doesn't dramatically change the experience. Not when basic map information is so small so you can preserve a consistent basic shape: a topographic map doesn't have all that much data in it. So an areas hills and terrain follow a roughly correct pattern.

With such a design, players will know when they come across an empty area, that emptiness is to some great degree purposeful. Which gives it meaning, rather than being the computer failing to make anything interesting.
 

JagerIV

Well-known member
So, hopefully final thoughts on map scaling and the implied map, and coordinate data size for a 16 bit (65,535 objects) list.


64 bit Universe map: 1 meter scale is 2,000 light year wide map. 1 mm scale would be 2 light years. 1.7 Gigabytes of location data with 65,000 objects. Overall game scale, hold "local" maps of suns, planetary systems, asteroids, and space fleets. 2 LY to 30x scale would be roughly 30 LY/10 parsec diameter map. this would be roughly 400 suns, suggesting something like 1,000 planets, 3,000 moons, and 6,000 major asteroids, for roughly 10,000 major objects. Leaving 40,000 slots for man made deep space objects.

32 bit regional map: 1 meter scale is roughly a 4 million km map. 1-30 scale suggests the map can cover roughly 128 million km. close to 1 AU, so therefore most red dwarfs with much smaller suns can have the entire system within a local map. A large Gas giant like Jupiter would still have its entire hill sphere of roughly 100 million km within a single local map. 1 MB

32 bit local map: 1 mm scale gives a 4,000 km map. This would be planetary maps where surfaces may be important. A 1/30 scale Earth would only be roughly 400 km wide, so you could have an entire earth plus a fair bit of orbital space. This would also only be 1 MB of data, but at this scale tracking more than 65,000 objects could be necessary. Earth at this scale would still be 500,000 km^2, so one object per km^2 would overwhelm the system, let alone one object per meter, which would be 500 billion objects. You could upgrade to a 32 bit list, and thus could track up to 4 billion objects. Not enough for per meter tracking, but you could have 1 km grid coordinates tracked as a biome, say a forest or city, with procedural filling in gaps. 4 billion objects tracked in a 32 bit coordinate system would be roughly 56 Gigabytes. If a dense world only had 1 million objects, say 1 token per km^2 to set general biome for 500,000 on an earth sized world, plus 500,000 objects of specific note, a planet would only have roughly 20 MB of location data, with space for up to a few billion objects than can be tracked if necesary, say a battle where you need to track a lot of bullets and such, the procedural temporarily has to track a bunch of local detail for a Forrest or city.

16 bit Object map: 1 cm scale means you can track up to a 655 m wide object. This would be used to track a local "object", such as a ship or large building. This sets a maximum scale for ships and other objects treated as a single object. This should be a pretty light limit: current longest ship is 458 meters, and the largest stadium is roughly 300 m round. Considering how big a level a full cruise ship or stadium would be, this should not be a major limitation on map design when a hand crafted area is needed. A full object with 65,000 parts would only be roughly 500 kilobytes of data. Individual objects can probably get away with 1-10 MB of data per model on average, especially using a shared bank of textures and such. So, if the universe could have 65,000 object list with an average data of 1 MB, the object list then might be 65 GB.

So, in summary, map data needs may be achievable in roughly so:

250 KB of location data for roughly 10,000 independent objects at the top map scale. Round up to 1 MB

There then might be 1,000 dense regional maps, say one per star plus larger gas giants. 1 MB per, so 1 GB data overall.

You then have a 10,000 local maps. Some are going to be much more empty, some much more full. If each has 20 MB of data, that would be 200 GB. A lot would be much smaller than earth though, so most can probably get away with 1 MB of data on average, which would be 10 GB. Assuming 1,000 dense local maps with 1 million parts at 20 MB per planet for 20 GB, 3k with some detail say 5 MB per local map for 15 GB, then the rest for 1 MB maps gets local map data for roughly 10,000 local maps, may full planets and moons, to 40 GB.

Finally, objects. If you had 1,000 high detail objects, say 10 MB per, those would be 10 GB. 4,000 mid detail of 5 MB gives another 20 GB. And then say 10,000 small low detail objects, coins, plates, paths, etcetera at 1 MB each for a final 10 GB, so total object list from a list of roughly 15,000 objects could be done in 40 GB.

Thus, 400 suns, 10,000 explorable planets, with 2 billion points of interest, pulling from a list of 15,000 objects might be fit within under 100 GB of data. Which isn't a small amount, but compared to the scale, impressive, and also quite doable on modern systems.
 

JagerIV

Well-known member
Now, we have to come to a question of how people could interact with this in an engaging way. Big maps are a staple of flight sims, because the high speeds require a large map to interact with, but on the other side of that it drives very low content per area. If one wants one to fly to 1-2 locations on a planet, then there's really not much reason to not have something like fast travel, or even smaller planets, like what No Man's Sky does.

I think the goal of something like this would be "The Galaxy at a human scale". To make use the the scale, the travel has to be a large part of the game. A core of that, is to Oregon trail it, and lean on logistics. This can drive through a variety of experiences, and make one experience the distance.

Lets start with a sandbox assumption for exploration and colonization. The process of colonization within a solar system would be:

1) Start on colonized world.
2) Enter Orbit
3) interplanetary travel
4) Enter New Orbit
5) Land on New world

Now, one way to shake things up is to change at which stage a player starts. This can give a particular run its own feeling.

Truck Start

For example, a player could already start on Mars. This skips all the space travel, and so instead the player simply needs to purchase a car, plane, or even bus ticket and then hoof it out into the wilderness. This simplifies the logistical element, and gives the player a big cushion: since there's already a settlement reachable by truck, you can easily move between the homestead and civilization to buy and sell things. Maybe if established enough you can have things directly delivered to you! You thus only have to interact with planetary travel and base/colony building, with self sufficiency not required, so you just have to concern yourself with making money. Which, because you have other people to sell to, the amount of things you can profitably sell by truck is pretty high.

This would probably be the initial player start, where one can start building capital to afford other starts.

Drop Start

A drop start would start with an orbital drop. You can have an expendable pod which drops you down with you and whatever starting supplies you can stuff in your drop ship. You are then on your own: assumedly your dropping because your going somewhere without civilization, and because you were in a drop pod, you can't get to orbit on your own. Your either completely on your own and just have to survive and build civilization on your own, or you have some schedule of trips where traders can come by and either drop additional supplies, and maybe even purchase things. Though building up enough infrastructure to support any exports may be a challenge in and of itself, unless you can sell things that don't require physical export.

As a player develops wealth, the scale of the drop pod affordable and the initial amount of equipment can both go up, allowing the player to organically skip earlier levels and develop a bigger, more impressive colonies easier and faster. For example, the most basic available drop pod might be something on the scale of the Soyuz return capsule, with enough space for maybe 3 people, plus maybe a couple hundred kg of supplies. This starts the player with virtually nothing.

iu


Meanwhile, with more money one might be able to afford, maybe up to a 400 ton drop pod, so you can start with probably several vehicles and prefabricated structures. Skip the early grind of starting from absolutely nothing. This keeps the focus very much on surface colony operations, but dramatically raises the stakes and makes means one must understand all the systems to a much higher degree, and not simply make enough money to buy any other deficits. A lot of stuff can't be bought, and with much higher shipping costs, what can be sold profitably is a much smaller selection.

Shuttle Start

Like the Drop Start, but you have a shuttle. So, you start with the ability to leave the planet and enter orbit, but with somewhat realistic shuttles, this is not a trivial thing to be able to do. The most basic realistic shuttle for an Earth level gravity I'm aware of is the DH-1

rocketCompany.jpg


And for an independent colony, represents a large investment to fly: The thing has an empty weight of 25 tons, requires 75 tons of methalox and 40 tons of hydrolox, requiring 3 different fuel liquids, and will deliver about 2 tons to orbit. So, the colony needs to manufacture about 120 tons of fuel per 2 tons delivered to orbit. Now, being able to ship in some precious 1 ton of goods could be the difference between survival and death for a colony, and there are things that 1 ton would be quite valuable to sell. 1 ton of gold would be about $50 million dollars. 1 ton of silver, a much less valuable material, would still be worth a cool half a million. And if your colony could produce a ton of something worth as much as silver every week or month, that's a nice cash flow to buy other necessities to be carried back or delivered by drop pod. Assuming there's someone to sell to in orbit....

A more substantial shuttle would be something like the Venture Star. Takes roughly 1,500 tons of fuel, and is 40 meters long, but its all hydrolox, so as long as you have power and a water source the fuel is fairly easy to manufacture, and you can do something like 50 tons of cargo, so about 2x as efficient in fuel consumption, and single stage. Its also substantial enough to make distant trips.

iu


If you had one launch on Earth, you can refuel it for a relatively speedy trip to Mars with a good 50 tons of cargo to start a new colony, and then if you can fully refuel it on Mars, theoretically it can deliver roughly 400 tons of goods back to Earth, which means the amount of things that can be profitably sold goes up and the freedom to buy things expands again. This would also be the point where a colony building focused player has to really start interacting with the space elements.

Carrier Start

The scale where you can own the ship that is bringing the colony drop pods and the person the colonists are trying to sell goods to. You really have to start focusing on the space elements: the orbital plane of the ship determines partially what viable surface to orbit locations there are. You really have to balance shuttle use vs infrastructure construction: is it worthwhile to consume a lot of fuel to do orbital transfers, or is it better to drop a bunch more colonies to support road connections to a centralized space port where you can run more efficient surface to orbit means, such as laser launch?
 

Users who are viewing this thread

Top