Ascent - The Space Game

Buy the PC version now!
  Click here to play The Space Game now!
Try it first with our Web Player demo
(with Unity3D Web Player)
Home Videos Press What's New? What's Next? Design and Vision Story About Us Help and Community History
Return to Design and Vision

Modelling the Milky Way


Our game universe now consists of over 200,000,000,000 star systems, most of which have planets. This environment is (to the limits of available scientific knowledge) an accurate model of the Milky Way as of available exoplanet and galactic observation data 10/12/2013. Based on the number of systems combined with the size of each system (stellar systems can be more than 100 AU across and these distances are simulated in-game), we believe it to be the largest virtual environment ever constructed for any game, and the most accurate depiction of the Milky Way in any game to date.

Stellar Density

image courtesy NASA / JPL-Caltech / R.Hurt (SSC-Caltech)
Image courtesy NASA / JPL-Caltech / R.Hurt (SSC-Caltech)

In order to take a photograph that depicts our galaxy in the same layout as the image above, one would need to send a space craft around 100,000 light years north and perhaps another 15,000-20,000 light years in a straight line from our Solar System towards the galactic centre. You would then need to travel forward in time over a hundred thousand years so the image your camera takes is of our galaxy as it stands today.

This is quite beyond current human technology. The image above was constructed by NASA based on what humans can observe and theorize, as best we can, the Milky Way to look like. It has the shape and density of our galaxy as best we understand it, with the prominent features - the spiral arms and the central bar/bulge - clearly marked.

As such, FKS chose this image to be our guide to both the Density (stars per sector) and the thickness of our galaxy. Our model of the Milky way is 100,000 light years wide, 100,000 light years long, and averages around 1,000 light years thick. However, in its thickest parts it is up to 4,000 light years thick. We have excluded off-plane globular clusters and the ultra sparse spherical cloud of stars, concentrating our model on the spiral/bar disc. These can be added later if a gameplay reason for them is found.

Image is a derivative work of density data courtesy NASA / JPL-Caltech / R.Hurt (SSC-Caltech)

This image is a single slice, or "layer" of 10x10x10 light year sectors. Each pixel is not an individual star, but is coloured more brightly or darkly based on the number of stars in the sector that each pixel represents. In order to fit the image on your screen, just over one fifth of all sectors in this one layer (of 400 or so total layers) are visible. This image was originally never intended for publication. It is the result of a system test - our generation code was run in order to confirm that it was indeed following the specified stellar densities.

Using this data and our persistent random algorithm, we have re-created the Milky Way in a way that can be generated precisely the same every time, obviating the need to store every sector, star and planet in a database. Database entries are only stored for systems once they are visited by players - recording the discoverer and the time/date of discovery. Later we'll also store terraforming actions/results and naturally who built what.

Stars and their properties

This image is a screen shot of the in-game sector map. The player's current sector is displayed in this map, along with all adjacent sectors. The names of systems are the discoverer's username and a number which increments with each new system that player discovers. At the time of writing a few hundred systems had been discovered. It is not possible in a human lifetime to visit every system in the game.

As can be seen in the image, stars have different sizes (some are obviously either giants or very close to the 'camera') and different spectral classes. The sizes in the map are not proportional, and the displayed spectral classes (colours) are based on the traditional, rather than the observed visual colouring of these stars. This is to ease player familiarity.

The most massive star known in the game is a Wolf-Rayet star of just under 150 solar masses (150x the mass of our sun). This star is relatively young, having spent a very brief time as an 'O' class 'bright blue' star, before becoming an ultra-bright Wolf-Rayet star, emitting a great deal of ultra violent and x-ray radiation. This star puts out more than 1.5 million times the radiation of our sun.

The least massive star is an M class 'Red Dwarf' star of just .08 Solar Masses, which puts out around 0.001 of the radiation of our sun.

Stars from the main sequence in every spectral class are represented, as are most types of giants known, documented, or theorised to date. Stellar remnants in the model include white dwarfs, black holes and neutron stars. Care has been taken to ensure that all stars generated by the model would be considered plausible by scientists.

While single stars are visible on the map, the model includes a calculated proportion of binary, trinary, quaternary and so on systems. These systems also follow the latest available information and thinking on their formation. For example binary stars tend to be either very close in mass and spectrum (e.g. two Class G stars of around the same mass orbiting each other) or a single star with a much smaller one such as a red or white dwarf orbiting it. (technically also orbiting each other but that isn't how it would look)

Planets and their formation

This image is of a generated gas giant within the model. Pictured behind it is a generated star field and nebula (these are strictly visual effects, although they are also carefully plausible where possible). Within the model, gas giants can be any combination of colours, although colours are statistically moved towards the black-white-grey spectrum. The model allows any colour more for gameplay and aesthetic reasons than scientific.

Little is known for certain about the formation of planets, and each new exoplanet discovered seems to challenge the existing theories once more. As such, considerable risk has been taken in the model by basing it on the current theoretical planetary formation models and available exoplanet data; quite simply in a year, two years, or ten, the model could be thoroughly disproven. As it is a computer game, and players will be living on these planets, constructing cities, mining bases, farming bases, manufacturing, military and research facilities, as well as constructing facilities in orbit, the planetary generation model can not be significantly altered in the future to reflect new scientific knowledge - as this might, for example, drop a city into an ocean, or drop it to -150 Kelvin, or flood the air with ammonia etc.

The planetary formation model closely follows the present thinking; simply put, a small portion of material is left over from the formation of a star. This material remains roughly disc shaped in space, orbiting the star, but is pushed outward by stellar winds. Less dense material tends to be pushed further than denser material before accreting into planetary bodies. Beyond a certain point, the level of radiation receieved from the star falls to below the point where water can freeze to a solid. This is known as the snow line, or frost line. Typically during stellar system formation, the first planet, a gas giant, forms just beyond this line, as a large mass of ice can accrete relatively quickly, drawing in more and more mass around it. In our solar system, this planet was (and is) Jupiter.

Clearing the neighborhood

Based on the planetary mass and its orbit around its parent object or objects, a Hill Sphere and a Sphere of Influence is calculated. A simplified model incorporating these results in cleared areas around each object's orbit. Planetary masses forming within a certain number times the Hill Sphere of an existing object are destroyed, eaten, or ejected over galactic time.

Within a gravitation sphere of influence (a subset of a hill sphere), and increasing in probability further into a hill sphere, moons can be formed or captured. A moon cannot be too close to a planet (within a simplified Roche limit) or they will be broken up by tidal forces (differences in gravitational forces from one side of an object to another due to the sides' varying ranges from the gravitational mass of the body they orbit). These may later become asteroid belts. At present a simplified moon generation model is used, and this can and probably will be updated prior to allowing settlements on or around moons.

Planets continue to form from the available remaining material. Beyond the frost line, gas giants are formed, and within it (closer to the star) rocky planets are more common. Rocky planetoids outside the frost line tend to become the moons of the gas giants, and smaller planetoids within the frost line either become the moons of rocky planets, or collide and are destroyed or ejected.

Gas giants can be many times the mass, and around twice the diameter of Jupiter, while for practical reasons rocky planets are currently restricted to earth's diameter, although they can be more massive. The diameter restriction is in place because the game engine literally simulates a full earth-sized object with varying terrain, which is profoundly enormous to any game engine and any player, and objects of this size require a number of "tricks" to operate relatively stably inside a modern computer game engine. Modern video cards are still almost completely incapable of rendering scenes that are based on 64bit floating point vector data, so all of the accuracy limitations of 32 bit floating point vector data apply.

However - the model is generating these planets at their actual radius and providing an earth-radius capped data set to the client. It is possible in the future to extend the radius of capped planets, although again this may not be possible where players have settled as it will distort (stretch) terrain features

Planetary Composition and Climate

Here an iron rich moon orbits a turbulent gas giant. The moon is actually quite small, as can be seen by its clearly non-spherical shape. There is a very large (proportionally) valley near the centre of the image. Rocky planets are given a detailed terrain by the model and the game engine itself. Much higher detail renderings of this terrain are a high probability for future updates of the game engine.

Planets within the model have a generated composition. Care was taken to ensure again that plausible planets were generated. Rocky planets have a mineral composition - the varying proportions of silicates and compounds of iron, carbon, aluminium and titanium, as well as a 'potential fluids' composition - the varying proportions of water, ammonia, methane, carbon dioxide etc. Based on the total received stellar radiation (which can be received from more than one star in, say, a binary system, and is calculated using the level emitted and the distance from the star or barycentric point of multi-stars) a surface temperature is calculated. This surface temperature determines the chemical state of these potential fluids - solid, liquid, or gas. Gaseous fluids can have further greenhouse effects, resulting in raising surface temperatures.

In this way, planets or even moons with methane or ammonia oceans are not only possible, they are a statistical inevitability.

Planets also have a generated level of geological activity and/or magnetosphere (magnetic field). These affect the atmosphere greatly. No magnetosphere usually means little to no atmosphere.

This model can and will be refined further in the near future, as it forms the basis of settlement and terraforming play. One of the game's design goals is to include complex terraforming activities for planetary bodies. For example, where a planet has liquid water in its oceans, and dissolved carbon dioxide, the addition of algae could release free oxygen into the system, which could then oxidise compounds such as methane and ammonia, moving that planet's atmospheric composition more towards one breathable by humans, and inevitably also dramatically altering its climate.

As the game already includes a complex scientific research system, a very detailed terraforming science branch is envisaged. This is flagged at present as one of the more complex elements of gameplay for advanced players and/or science nerds. Other players who would rather just blow up space pirates will likely do so from bases built by terraforming industrialist players.

Fun facts from the model

Take or leave these observations with a grain of salt; this is a computer game model after all not a scientific model. With that in mind, we noticed:

  1. Stars with no orbiting planets and no other stars in their stellar system are quite rare. It's really hard to form a star with nothing left over. This was accepted as it seems to agree with exoplanet data
  2. Systems with more stars actually tended to have less planets, as the many stars disrupt planetary orbits and eat or eject planets over galactic time, unless the stars were so far apart people might argue that they're not really part of the same system.

3. Earth like planets are hilariously rare. There are just SO many variables that it's just really unlikely we'll find a habitable planet within hundreds of light years. We'll refine this one more later as we refine our climate and atmospheric/oceanic models, but even as it stands...

Note the requirements of LIFE are much broader/more permissive. Human life outside a space suit is the narrow definition here.

Closing Remarks

There is no conclusion to draw here. This is a beginning for us, not the end. However, I would like to say constructing this model has been one of the most incredible experiences of my life. I am completely in awe of the scientists who made it possible (and no doubt if any read this they'll realise I've wrecked it or something and giggle) with their work. Yet at the same time, I find myself struck dumb at how little we know for sure about formation of even our own sun or our own solar system. As full of conjecture and guesswork as this crazy computer game model is, it's probably a reasonable metaphor for the current state of our knowledge.

I ran into a lot of situations where one thing I read completely contradicted another. The definition of 'planet' is still being argued to some extent, let alone how they're formed. As for the formation of a stellar system like our solar system, there are still huge gaps in the real models, let alone this one. I get to wallpaper over those gaps by slapping in a number... I don't envy the scientists who aren't allowed to do that...

By that I mean simply this - we have a LOT of discovery yet to do in our own solar system, let alone the other 200-400 billion systems we think are in our own galaxy, let alone the other 100-200 billion galaxies...

James Hicks, December 10, 2013.