Procedurally Generated Celestial Surfaces and Biomes

[Gumby]

I wanted to suggest this to create a semi-unique experience each time you embark on a new Journey (begin a new universe).

Features:

Kerbin: "biomes" alike in other procedural terrain engines, could be an area that cultivates an expected result, such as mountains where mountains should be and desert plains wherever the game feels comfortable placing them. It wouldn't be heavy during the game as it would enter a pre-generation state, generating terrain for all celestial bodies +- Kerbol

Craters: im a big fan of holes in the ground, but when their bowl shaped and great for driving around in, oh boy! My point is, is that celestial bodies that have craters would be alot more interesting if those same craters weren't there in a second career playthrough.

Mapping: this feature would now have a legitimate purpose, as it would allow pilots to plan their next celestial vacation on the currently mapped body as well as perhaps science for such accomplishments.

Mapped bodies should display number of unique biomes

Science: oh yeah science... this will be a lot more non linear collection of important data and will give a more exploratory feeling to the well known system we love and perhaps more incentive to enjoy a terrain generated jewel. Also, to refer back to mapping, the science menu should catalog known biomes, visited biomes, and perhaps "discovered" biomes (feel free to use a less confusing word by all means); these biomes are not discovered through mapping but through close up analysis, kind of like sub-biomes where the terrain is slightly different, but not unexpectedly different from the biome in general and can be used to key in a extra 5 to 10% of science points.

I hope you like the suggestion! feel free to add comments!

[aeTIos]

This is on the "Do not suggest" list. Squad wants the game to be the same for everyone who plays the game, so you can share screenshots and say "I've been there too!", amongst other reasons.

[LethalDose]

Craters: im a big fan of holes in the ground, but when their bowl shaped and great for driving around in, oh boy! My point is, is that celestial bodies that have craters would be alot more interesting if those same craters weren't there in a second career playthrough.

I think most of the craters on the Mun are already procedurally generated. The large named craters and some larger highland/midland craters are set, but the rest are randomly placed each play-through.

[vexx32]

There's a very big difference between "random" and "procedural". The craters on the Mun are not random, they are procedurally generated with a fixed seed, so as to be in the same place on every play-through.

While this particular thing is not on the WNTS, it does bear mentioning that HarvesteR has in the past stated that having a random universe is not on the cards for the stock game, as it detracts from the community aspect of the game if everyone has different planets all the time.

[draeath]

Hmm, I would think going procedural but using a particular seed (unless overridden in settings?) would be a great compromise. Anyone who doesn't open up the INI and set a seed gets the same universe. Those who want to deviate, can, and they can share the seed for others to use if they wish.

I'd even go so far as to omit it from the menus - make them dive in the INI like we do for the conics patch setting. (of course, it would need to be written into the persistence file when a game is created - I'd imagine the hilarity that could result otherwise!)

[Kasuha]

As long as KSP is already generating all of the terrain procedurally ... and there already are different terrain classes as for instance it is way harder to land a plane on terrain with rocky texture than on terrain with grass texture ... it could assign biomes to each location automatically as well.

Especially since the current bitmap-driven implementation is horribly ïnaccurate, at least on Kerbin.

[r4pt0r]

There's a very big difference between "random" and "procedural". The craters on the Mun are not random, they are procedurally generated with a fixed seed, so as to be in the same place on every play-through.

I think most of the craters on the Mun are already procedurally generated. The large named craters and some larger highland/midland craters are set, but the rest are randomly placed each play-through.

so it is always the same?

[Dunbaratu]

The current systems is procedurally generated but the algorithms operate from fixed hardcoded starting data, guaranteeing an identical deterministic result every time.

I'd like them to allow mods to fiddle with those starting points to open up the chance of those mods adding variation to the universe. And that variation could include some randomizers.

(I don't think it's as simple as one single seed number like in Minecraft, though - it's a large collection of starting data, but at the moment that data is not exposed to the modder API so nobody can change it even in a mod.)

[vexx32]

so it is always the same?

Pretty much. It moves around a little bit on the Low terrain detail, but is generally speaking in almost exactly the same place.

[DoctorCruz]

I wanted to suggest this to create a semi-unique experience each time you embark on a new Journey (begin a new universe).

Features:

Kerbin: "biomes" alike in other procedural terrain engines, could be an area that cultivates an expected result, such as mountains where mountains should be and desert plains wherever the game feels comfortable placing them. It wouldn't be heavy during the game as it would enter a pre-generation state, generating terrain for all celestial bodies +- Kerbol

Craters: im a big fan of holes in the ground, but when their bowl shaped and great for driving around in, oh boy! My point is, is that celestial bodies that have craters would be alot more interesting if those same craters weren't there in a second career playthrough.

Mapping: this feature would now have a legitimate purpose, as it would allow pilots to plan their next celestial vacation on the currently mapped body as well as perhaps science for such accomplishments.

Mapped bodies should display number of unique biomes

Science: oh yeah science... this will be a lot more non linear collection of important data and will give a more exploratory feeling to the well known system we love and perhaps more incentive to enjoy a terrain generated jewel. Also, to refer back to mapping, the science menu should catalog known biomes, visited biomes, and perhaps "discovered" biomes (feel free to use a less confusing word by all means); these biomes are not discovered through mapping but through close up analysis, kind of like sub-biomes where the terrain is slightly different, but not unexpectedly different from the biome in general and can be used to key in a extra 5 to 10% of science points.

I hope you like the suggestion! feel free to add comments!

I would like to see a mapping system when biomes are implemented on other planets

[LethalDose]

Pretty much. It moves around a little bit on the Low terrain detail, but is generally speaking in almost exactly the same place.

Wait, srsly? The procedural crater generator is always handed the same seed?

Also, can we please give this tired "procedural isn't random" business a rest? It's completely condescending to anyone you may be talking to unless you're going to deign to actually explain the difference to who you must see as the ignorant masses.

For clarification: "Procedural" is a program term for creating something (an image, terrain, number, character string, etc) based on a set of commands, typically based on one or more values. Given the same input, a procedural process will always produce the same output. This gets confused with "random" because very very frequently, objects that are procedurally generated by computers pass the generation process a random value (often referred to as a "seed"). The results of similar seeds are often wildly different, meaning it is difficult or impossible to actually predict the outcome of

Another reason that it's so insanely condescending to parrot "procedural isn't random" is that even what is widely accepted as random in computing is almost never actually random, but is really pseudorandom. Pseduorandom generators are actually procedural because the same seed (e.g. system clock time) will always produce the same result. The only way around this is to provide the pseudorandom generator's seed, in turn, from some physical source that actually has entropy (nuclear decay, noise, etc).

[vexx32]

I said procedural is not the same as random. Which is true. I only said it because in your earlier post, you seemed to confuse the two, from how I read it.

I am perfectly well aware that pseudorandom generators are procedural -- the distinction is merely the input values assigned. If the input values differ in some way, the output will end up being random to some degree, determined by the code used to generate the output. If the input seed is always the same, it is merely a useful way of generating a large amount of output without having to manually code or model each and every piece of it.

And yes, the crater generator must, by necessity, be handed the same seed all the time. If it is not, when you load up a save game with a craft landed on the Mun inside a crater, you might find that it is suddenly and inexplicably underground, and the crater it once occupied is now either nonexistent, or a long way from where it was previously.

[LethalDose]

I said procedural is not the same as random. Which is true. I only said it because in your earlier post, you seemed to confuse the two, from how I read it.

I am perfectly well aware that pseudorandom generators are procedural -- the distinction is merely the input values assigned. If the input values differ in some way, the output will end up being random to some degree, determined by the code used to generate the output. If the input seed is always the same, it is merely a useful way of generating a large amount of output without having to manually code or model each and every piece of it.

And yes, the crater generator must, by necessity, be handed the same seed all the time. If it is not, when you load up a save game with a craft landed on the Mun inside a crater, you might find that it is suddenly and inexplicably underground, and the crater it once occupied is now either nonexistent, or a long way from where it was previously.

First off, the explanations above RE: procedural, random, and pseudorandom wasn't to explain the differences to you, but to anyone who doesn't know what the differences are and to point out why the statement is condescending. Despite all the times I've seen "Procedural isn't random" stated, I've never seen anyone actually explain the difference. And, as I stated, merely pointing out "Procedural isn't random" is condescending because:

• People who understand the difference, well, obviously know the difference and don't need to be told. If they know the differnce, they're probably making the substitution by convention instead of making a mistake.
  
• People who don't understand the differences are left in the dark about why they're being corrected. You're just telling them they're wrong without an explanation.
  

Now, all that being said, I'm actually very surprised that the crater generator always uses the same seed. Obviously, within saved games, it would be absurd to change the seed for the reasons you pointed out above (landers on the surface). But that doesn't mean that the seed isn't changed between saves. I created a new game and paid close attention to the new crater positions and they do appear to be identical.

Again, I'm really surprised the devs would go through the process of creating the crater generation algorithms and not feed it a random seed for each new save. But they did, and when I said they were randomly generated above, I was mistaken.

[vexx32]

Welp, there's twice I've misread your posts. Sorry about that.

At a guess, I'd say that feeding the same seed into the generation algorithm is another of their efforts to make sure that there is a "community" aspect to KSP -- someone finds an especially interesting crater on the Mun, posts coordinates, and everyone else can go see the same crater and so forth.

[Kegereneku]

This is a suggestion that come really often "Should SQUAD make it possible for players to start in a different solar system (if possible) ?".

Now some obtuse proponent pseudorandom-procedural-generation might jump on the feasibility ("It's possible I swear, look at my unrelated examples !") without considering the actual question.

Myself I say : "Not worth it"

For starter it Would "break the community", how would new players even ask about how to land on a planet/moon if nothing they can describe will make sense to other ?

If you got bored navigating a solar system with custom-made diversity, you'll get bored even faster with automated generation of not-so-random relief.

This game isn't like Minecraft, you don't shape your environment until it suit you, you face it and take it for what it's worth.

Lastly it might not be possible at all or without huge sacrifice. (pro-tips : procedural generation is used best to fill the void, not create content)

[Dunbaratu]

And yes, the crater generator must, by necessity, be handed the same seed all the time. If it is not, when you load up a save game with a craft landed on the Mun inside a crater, you might find that it is suddenly and inexplicably underground,

You are pretending the only two choices are:

1: seed is fixed forever for all games.

2: seed changes every time you load the game.

And you know that's false. You can generate a seed when first starting a game and then keep that seed saved with that game. To get a new seed you start a new campaign.

[Gumby]

All of you recommendations and clarifications are much appreciated and the responses given are remarkably enlightening, so thank you. But i still cant wrap my head around the understanding of how it might detract from the community's experience with the game. If any of you are willing to elaborate in more depth it would be great.