Scale of KSP

[Shadownailshot]

Perhaps you should re-read my post. Then do a wikipedia search on procedural generation. It doesn\'t have to have 'no character.' Only if the programmer is lazy. This is the reason Infinity: The Quest for Earth has been taking so long. The programmer kind of wants to make it so that it\'s not 'boring, without character.'

[Themohawkninja]

Perhaps you should re-read my post. Then do a wikipedia search on procedural generation. It doesn\'t have to have 'no character.' Only if the programmer is lazy. This is the reason Infinity: The Quest for Earth has been taking so long. The programmer kind of wants to make it so that it\'s not 'boring, without character.'

You beat me to it!

Now to go check out that game you mentioned...

[angryman2]

i think that the idea of ksp going into a galactic scale would be great even though it would take some time it would be fun, i think that as for distance and point precision error it could be solved by using stars and maybe even planets as 'hot spots' say you are headed to a star that\'s pretty far away from kerban like three light years. those two stars have both have point precision spheres that overlap. once you are in that overlapping spot the two stars and kerban (kerbans sphere is infinite so that it can give a third point of focus )then target your exact location by say hitting you with a signal or whatever computers do and the computer now knows exactly where you are in space and then there\'s no point precision error. as for loading up the game you could do it as a mixture of minecrafts chunk loading and kerbans sphere of loading quality first it loads the 'chunks' in low quality and as you get closer the quality gets better.

and sorry for the mouth full!

[Mr_Orion]

It\'s been proven to work with areas up to 30,000,000 m². That was with the game filled with hundreds of objects at any given time. It should work on an even grander scale, because in KSP there is much less occupying that same space.

Procedural generation has no upper limit whatsoever. Physics and number inaccuracies do, however; this is irrelevant though because of KSP\'s floating origin system.

[LukeTim]

Procedural Generation is alot better than you might think. You can set it to craft as many different types of planets/moons/stars/asteroids/comets as you want, with a TON of different details, and it will do that, randomly, but it can save the generated item so that when you come back to it later, it will be the same as you left it. It doesn\'t make carbon copies, unless you set it to make carbon copies. It\'s very versatile, and allows for much larger play areas with a lot less lag.

IIRC, Kerbin, the Mun and Minmus are all procedurally generated, just with a fixed seed all the time so they never differ.

[Themohawkninja]

IIRC, Kerbin, the Mun and Minmus are all procedurally generated, just with a fixed seed all the time so they never differ.

Than why did they procedurally generate them? Was it in preparation for actual randomness?

[Mr_Orion]

Than why did they procedurally generate them? Was it in preparation for actual randomness?

Nope, it\'s because it\'s takes far less memory to procedurally generate something than to actually store it; Kerbin is massive. You\'d need to store gigabytes of information to store the entire planets surface.

[vexx32]

It\'s also usually faster to just generate it on-the-fly than try to load it into memory from the hard disk.

[samstarman5]

I could not agree with you more! The future of the game in the next 6 months is so very bright. The potential is there for so much to explore and do just in the local solar system that we wont need anything else for a long long time.

I am of the same stance. I would be fine if KSP was released with just the Kerbol solar system to play around in. Interstellar travel and other solar systems can be added later through DLC or expansion if Squad wanted to take a break.

[Solusphere]

Personally, I think making the area outside of the main system explorable would actually detract from the game. Currently, the game is set in a near-modern-technology level, with a mostly realistic model of propulsion. To explore external solar systems in a timeframe the average player would be willing to sit through, you would either need to implement sci-fi FTL systems, or have an absolutely enormous time accelleration setting, both of which kind of feel like cheating. The game is called Kerbal Space Program, not Galactic Kerbal Empire. At most, I think there should only be one external solar system in range, kind of the equivalent of our Alpha Centauri, but closer. It could be the ultimate challenge goal for those who are willing to spend many hours of max-accellerated flight with an ion engine to get there.

[Bananasword]

I think the limit should just be the solar system. Add in a higher res static space background of stars just for giggles.

A must is of course a random but scarce mission where a huge asteroid enters the system on a crash course with kerbin and the KSC has to find a way to off set it.

you can\'t have other solar systems because then the programmers will have to widen the gravity scope so that the systems are moving in the arms of a galaxy with a black hole at the center.... all that math.... alll that work.

No thanks. just a nice solar system with planets being similar to our own. asteriod belt is probably asking to much lel.

[RayneCloud]

I think the limit should just be the solar system. Add in a higher res static space background of stars just for giggles.

A must is of course a random but scarce mission where a huge asteroid enters the system on a crash course with kerbin and the KSC has to find a way to off set it.

you can\'t have other solar systems because then the programmers will have to widen the gravity scope so that the systems are moving in the arms of a galaxy with a black hole at the center.... all that math.... alll that work.

No thanks. just a nice solar system with planets being similar to our own. asteriod belt is probably asking to much lel.

Not true, unity supports seamless scene switching and scene streaming. Adding another solar system would simply require building a new flight scene in that area and as you approach it, unity unloads the old scene and streams in the new one. You wouldn\'t notice.

But for now, this solar system that we have is about 1/11th scale in regards to *most* planets, save for Kerbin\'s main moon, which is bigger in proportion then Earth\'s moon is to Earth. Just wait though, they\'ll be plenty of explorable space.

[Bananasword]

Fair enough you have the technology. But I still don\'t see why adding another solar system in would be useful. like I said as a solar systems are part of a gravitational pull of super black holes at the center of their galaxies you\'d have to simulate that for the kerbin ss and the foriegn ss. okay you have the technology perhaps but you run into the problem where that looks ugly.

You\'ll have two ' real' solar systems against a 2d space backdrop. Over eager fans will demand for more solar systems to break up the weird scene of just two lonely 'real' solar systems chillin on a galaxies jpg arm. Where as if you keep the game on the one solar system it\'ll aesthetically look complete.

besides the single solar system can be full of tonnes of interesting stuff, if you spend more time on the details rather than pooping out random solar systems.

Otherwise its your game to make and I totally will respect and fly to whatever solar systems you decide to add. I guess fingers crossed they are the last part of the development cycle.

[vexx32]

Considering that the planets are generated from seed numbers on the fly, it\'s plausible to generate an entire galaxy that way, with some predefined rules and some pre-made textures that can later be expanded upon, perhaps overlayed with colors depending on the seed for certain planets, etc.

You can make an entire galaxy the same way Kerbin, the Mun and Minmus were made -- predefined paramaters and seeds. The detail level would be almost identical to what it is now, and can certainly be improved upon later.

Or at least, that\'s one way it can be done. I do tend to agree with the viewpoint that one system is enough, primarily because at this point in time it would be pointless and rather unusual for players to venture beyond it. In future... well, we\'ll see. I just had a thought -- what if the devs setup a thing where players could sit down and put a few code snippets up in pseudocode that is easy-to-understand and the team can have an automated script convert that into seeds and textures and other terrain features for planets throughout a galaxy? A single galaxy, created by thousands of hands... of course the pseudocode would need to be carefully monitored so that nothing malicious can possibly be created.

[Skyro]

You could make all of the planets outside the solar system of Kerbol procedural, like Space Engine or something like that, but I bet that would take a while to do, and would take one of the further out updates, like 0.30\'s. I wonder how that would affect how big the scale of KSP is. Of course we would settle out the bugs of interplanetary speeds and such, but I bet we could think to procedural generation of inter-solar planets.

[rkman]

You can make an entire galaxy the same way Kerbin, the Mun and Minmus were made -- predefined paramaters and seeds.

I\'m pretty sure Kerbin, Mun and Minmus are (at least for the most part) not generated procedurally.

The dev who made the new terrain engine also made the Mun\'s texture (which was greatly improved a few releases ago).

Probably 'scatter' (rocks, trees) and perhaps some texture noise is placed procedurally.

like I said as a solar systems are part of a gravitational pull of super black holes at the center of their galaxies

Or they could be 'on rails' just as the planets that are in KSP now.

In fact they could just be stationary because movement relative to the galaxy is imperceptible anyway.

[vexx32]

Well, let\'s just clear this up as fast as possible. I\'m tired.

From what I\'ve read from the last few months from the devs, here is how it works (basically):

The Mun, Kerbin and Minmus have a very low-res heightmap of sorts that\'s wrapped around the sphere that is the basis of the planet (we\'ll use Kerbin as the example). This allows the devs to add very basic land forms and features, but only very vague ones. Then, the devs fix a seed number that never changes and is stored somewhere in the game files. This number is used to generate terrain 'noise' and make it look a TON more detailed -- small hills, lakes, whatever -- which is applied over the top of the very low-res heightmap to collectively produce a convincingly detailed terrain map. Textures are then applied to this, and in the case of Kerbin, the ocean is added (which is essentially a sphere, but is actually a hedron with many faces, which can actually be configured to be low or high-res. The texture itself is always applied around a much more perfect sphere, so if the ocean\'s hedron isn\'t high-res enough, you start to get those issues where the ship 'hits' the water, but seems to be floating above it instead. Textures don\'t match up and such). Then, terrain scatter is added, which may or may not be affected by the seed (I have yet to see confirmation either way).

If you want proof of this, go look back through Harv, Mu, and Nova\'s posts over the last few weeks and a bit further back for Mu and Harv. This is how I understood it to work, but I sure as hell ain\'t a dev and probably misread something somewhere, but this is the general gist of it.

[Shadownailshot]

I\'m pretty sure Kerbin, Mun and Minmus are (at least for the most part) not generated procedurally.

The dev who made the new terrain engine also made the Mun\'s texture (which was greatly improved a few releases ago).

Probably 'scatter' (rocks, trees) and perhaps some texture noise is placed procedurally.

Or they could be 'on rails' just as the planets that are in KSP now.

In fact they could just be stationary because movement relative to the galaxy is imperceptible anyway.

Procedural Generation does not mean random. It can do things seemingly at random. But it does not mean random. Pick apart the words. It\'s the english language. Procedural- means it goes through a procedure- Generation- it generates things from a smaller item. Nothing about 'random' in that description.

[Alchemist]

It doesn\'t have to be a large spiral galaxy. Imagine dwarf galaxy like Magellanic clouds.

We can have a small cluster of stars outside of a large galaxy (so you can see it in the background but are unable to reach).

So we can have couple dozens of stars of which:

3-5 systems with large numbers of planets and moons to explore (including Kerbol system)

Up to a dozen systems with some interesting features (cool to visit and look from close distance but nothing much to explore), like:

• 
  [li]multiple-star systems (there is a known case of 6-star system)[/li]
  [li]maybe some unusual case of close trinary system with uncommon trajectories? [/li]
  [li]a lone gas giant in ultra low orbit around its star[/li]
  [li]a planetary nebula - remnants of a red giant with a white dwarf[/li]
  [li]a dwarf star stealing gas from an old red giant (that\'s how to get white dwarf around white giant like Sirius)[/li]
  [li]a back side of the previous, when the formed white giant becomes unstable and the white dwarf (remnants of the original giant) starts stealing the gas back, which (because of the temperature of the white dwarf) every several years (when there gets enough hydrogen) starts fusion resulting in a several days long flash of a nova. P.S. it\'s very bright. if it happened in Sirius system, we here on the Earth would see it as bright as our Sun![/li]
  [li]Maybe also the classic case of black hole + red giant = a disk of a superheated gas that emits strong X-ray radiation[/li]
  [li]etc...[/li]
  

Just some lone stars of different sizes

The only problem is the interstellar gravity in a cluster - it\'s not a Keplerian point-source field.

It can be approximated by a point-source field with intensity g®=M*m®/r², where m® is mass of the stars that are closer to the center (can be approximated with some distribution formula).

Well, that depends on what we\'ll get. If periods of star orbits will be at least several millenniums, it\'s not worth simulating, let\'s make the stars static and interstellar SOI with zero gravity. If we get orbital periods of at least some stars less than couple centuries, it\'s worth to make them moving around the cluster at least in circular orbits (and maybe couple stars in non-circular orbits

[Endeavour]

How would we reach these other solar systems? If the other stars are moving also, that would make launching to them a nightmare.

[Cthulufaic]

How would we reach these other solar systems? If the other stars are moving also, that would make launching to them a nightmare.

Well, with the new 100,000x timewarp coming in 0.16 then I think that it will be less of a hassle waiting for the right moment. But maybe the stars won\'t be moving relative to each other.

[Shadownailshot]

Where was it said that they\'re bringing a 100,000x timewarp?