Is it possible to have handcrafted AND proceedurally generated planets

[mcwaffles2003]

Just wondering, to those who develop code, if there is a hard limitation that makes it so these can't co-exist. Also, is this kind of thing possible through modding if it isn't natively placed in? If it is possible would there likely be a performance hit caused by this and what are the likely limitations to implementing this?

I realize these might be difficult to answer without the game here with us but I figured I would see what others think is possible on this front

Edited January 23, 2020 by mcwaffles2003

[micha]

Yes it is possible. Always depending on how the system is built, of course.

But there are many games out there which do this successfully. Just to mention 2, Elite Dangerous and No Man's Sky.  The former has entire hand-crafted systems (eg, Sol) as well as hand-crafted locations on procedurally-generated planets (bases etc), but most of the galaxy is procedural.  The latter is completely generated, but allows players to deform the terrain and add hand-crafted bases anywhere.

Generally speaking you either suppress the procedural algorithm for the areas which you want to hand-craft, or you override them after the procgen is finished.

[Lisias]

To tell you the true, I think it's already made!

It's an external application, it doesn't runs inside KSP - what's usually a good idea, since you don't want the galaxy to change once you start a game, and KSP can't have a per savegame GameDatabase (the thingy where everything on the game exists, from parts to planets)

 

[razark]

Is it possible? Yes.

game_cost += artist_salary;

[mcwaffles2003]

10 minutes ago, Lisias said:

To tell you the true, I think it's already made!

It's an external application, it doesn't runs inside KSP - what's usually a good idea, since you don't want the galaxy to change once you start a game, and KSP can't have a per savegame GameDatabase (the thingy where everything on the game exists, from parts to planets)

 

Does this actually work? O_O

 

[micha]

What I'd personally have loved to see is a procedurally-generated Kerbol system.  The dev-argument was always that they want people to play in the same environment so they can exchange stories and knowledge etc.  But as long as everybody uses the same starting seed, that can still be true.

Then when people get bored of the default, they can change the seed and try a different system.  As people try out different seeds, some will be silly or even unplayable and others will be awesome. Bingo, people share popular system seeds to play in.  This would also allow for true exploration since each system is unknown to start off with.  Add in a better discovery mechanism (eg, planetary surfaces are blurred until mapped using a suitable instrument) and it opens up a whole new gameplay.
 

[Guest]

Honestly I don't know if procedurally generated systems ar worth the effort, this isn't Elite Dangerous in which you can hop on your Diamondback Explorer (if you use anything else you're not a true explorer but a spoiled tourist) and visit 50 systems in an hour.

Jool missions are something big to plan and execute and bigger and more powerful engines will not change that, just reduce the amount of time you spend in time warp (and they will require better and bigger orbital facilities).

In KSP2 Jool will just be the first step, like Mun is in KSP1, interplanetary missions will be something even bigger and more complex, especially if there's some sort of life support.

I think that if you can jump from system to system so fast and easy that you can get bored with the 2 or 3 (4, 5 or 6 the number doesn't change the point ) initial systems there something really wrong with the game difficulty.

Also I passed 2 months designing and testing an electric plane to return from Eve by launching at 20000 meters, it's the most fun I've ever had in 2000 hours of KSP. I want more design puzzles like that, handcrafted puzzles to solve and, with time, master enough to build a colony.

I'd also prefer to have environmental concerns for existing planets (pressure and heat on Eve, radiation on Laythe, but even rugged terrain or problematic rocks around interesting landing zones) before having procedurally generated systems that relies on fewer variables.

Edited January 23, 2020 by Guest

[mcwaffles2003]

4 minutes ago, micha said:

What I'd personally have loved to see is a procedurally-generated Kerbol system.  The dev-argument was always that they want people to play in the same environment so they can exchange stories and knowledge etc.  But as long as everybody uses the same starting seed, that can still be true.

Then when people get bored of the default, they can change the seed and try a different system.  As people try out different seeds, some will be silly or even unplayable and others will be awesome. Bingo, people share popular system seeds to play in.  This would also allow for true exploration since each system is unknown to start off with.  Add in a better discovery mechanism (eg, planetary surfaces are blurred until mapped using a suitable instrument) and it opens up a whole new gameplay.
 

This is why I like the idea of a handcrafted stellar neighborhood with progen galaxy. That way everyone can learn to play in the same playground, that is beautifully tailored and nuanced; and if anyone makes it far enough they can finally leave that playground and wander in the random infinite. I doubt most will make it that far but there's just something nice to always knowing there's more out there, you just have to be willing to go out and look. Brings a feeling of being less constrained.

With up coming multiplayer I could see people being scattered about in the same galaxy exploring together but at a distance.

[Lisias]

9 minutes ago, micha said:

Then when people get bored of the default, they can change the seed and try a different system.  As people try out different seeds, some will be silly or even unplayable and others will be awesome. Bingo, people share popular system seeds to play in.  This would also allow for true exploration since each system is unknown to start off with.  Add in a better discovery mechanism (eg, planetary surfaces are blurred until mapped using a suitable instrument) and it opens up a whole new gameplay.
 

I remember an Add'On to automate it, Planetary-Diversity!

[mcwaffles2003]

3 minutes ago, Master39 said:

Jool missions are something big to plan and execute and bigger and more powerful engines will not change that, just reduce the amount of time you spend in time warp (and they will require better and bigger orbital facilities).

But more powerful engines will reduce the amount of trips you need to get a payload out there. Im not looking for base game FTL travel but if you can have a game with 2 or 3 star systems, why not make a galaxy?

 

7 minutes ago, Master39 said:

I think that if you can jump from system to system so fast and easy that you can get bored with the 2 or 3 (4, 5 or 6 the number doesn't change the point ) initial systems there something really wrong with the game difficulty

Im not looking for the game to make it easy to get to these places but maybe if there were a rare resource mechanism brought to the game where only a few bodies had one of these specific resources it would make a player incentivized to explore the galaxy in search of them, creating elaborate networks in the wake. Hopefully some of the monotony could become automated as the game progresses, like resource transfer between well established bases

 

11 minutes ago, Master39 said:

I'd also prefer to have environmental concerns for existing planets (pressure and heat on Eve, radiation on Laythe, but even rugged terrain or problematic rocks around interesting landing zones) before having procedurally generated systems that relies on fewer variables.

I think that would be a cool mechanic as well and agree these kind of steps are higher priority, why make a bunch of crap systems instead of refining the mechanics of the few we already have. But also, why not make many of those mechanics global once they're developed? Hand crafted planets will always be exceptional but random planets here and there could always throw you through a loop as well even though most will likely come out more plain.

 

There's just something, to me, about having the ability to explore a world of exotic stars, wandering black holes, and stellar/planetary nebula. Even if they dont get visited... just knowing you can

[Guest]

7 minutes ago, mcwaffles2003 said:

But more powerful engines will reduce the amount of trips you need to get a payload out there. Im not looking for base game FTL travel but if you can have a game with 2 or 3 star systems, why not make a galaxy?

 

Im not looking for the game to make it easy to get to these places but maybe if there were a rare resource mechanism brought to the game where only a few bodies had one of these specific resources it would make a player incentivized to explore the galaxy in search of them, creating elaborate networks in the wake. Hopefully some of the monotony could become automated as the game progresses, like resource transfer between well established bases

 

I think that would be a cool mechanic as well and agree these kind of steps are higher priority, why make a bunch of crap systems instead of refining the mechanics of the few we already have. But also, why not make many of those mechanics global once they're developed? Hand crafted planets will always be exceptional but random planets here and there could always throw you through a loop as well even though most will likely come out more plain.

 

There's just something, to me, about having the ability to explore a world of exotic stars, wandering black holes, and stellar/planetary nebula. Even if they dont get visited... just knowing you can

My point is that a procedural system is something big to develop and implement into a game like KSP, games like No Man's Sky and Elite: Dangerous are built around their procedurally generated worlds.

I dont think it's worth the effort, remember that most people never left Kerbin's SOI.

[mcwaffles2003]

2 minutes ago, Master39 said:

I dont think it's worth the effort, remember that most people never left Kerbin's SOI.

I hear that but I think that largely comes down to 2 things:

• The KSP 1 difficulty ramp is insane and has no tutorials
• Performance quickly degrades when you start building 200+ part ships for interplanetary travel

KSP 2 is addressing these issues on multiple fronts.

• They've promised large improvements in performance/part
• They're adding tutorials to help people get up that difficulty ramp
• They're improving the graphical scenery (creates incentive to go to new places)
• They're integrating extraplanetary launch pads (getting out of Kerbins atmosphere and SOI wont be a prerequisite to every single launch anymore)

[Guest]

3 minutes ago, mcwaffles2003 said:

I hear that but I think that largely comes down to 2 things:

• The KSP 1 difficulty ramp is insane and has no tutorials
• Performance quickly degrades when you start building 200+ part ships for interplanetary travel

KSP 2 is addressing these issues on multiple fronts.

• They've promised large improvements in performance/part
• They're adding tutorials to help people get up that difficulty ramp
• They're improving the graphical scenery (creates incentive to go to new places)
• They're integrating extraplanetary launch pads (getting out of Kerbins atmosphere and SOI wont be a prerequisite to every single launch anymore)

Yes, and those are all welcomed additions that makes interplanetary missions more intresting, but going interstellar is a whole new layer on top of that and a way bigger jump to do.

Even if all those "never left Kerbin SOI" people will finally begin doing interplanetary missions most of them will never probably leave Kerbol's SOI 

[micha]

My point wasn't _multiple systems_ within the one game/save for people to travel to.  My point was going back to an early dev discussion in KSP1 where the devs were against procedural generation because they wanted people to have the same experience.  Procedurally-generated does not equal "random".

Hand-crafting an experience often leads to a better game (eg, by having the Mun in an equatorial orbit combined with the launchpad being on the equator makes reaching it a lot simpler as a first goal after reaching orbit). But it's possible to have BOTH a procedural system and hand-crafting. 

So best of all worlds (imho) : Start everybody off with seed #0 which is heavily hand-crafted, but allow them to choose a random seed instead for a random experience.

[mcwaffles2003]

19 minutes ago, Master39 said:

Yes, and those are all welcomed additions that makes interplanetary missions more intresting, but going interstellar is a whole new layer on top of that and a way bigger jump to do.

Even if all those "never left Kerbin SOI" people will finally begin doing interplanetary missions most of them will never probably leave Kerbol's SOI 

Well at that point what do we do? hold it back for the people who wont play the game much or expand it for the people that do?

I agreed earlier that planetary mechanics should be well revamped, but I dont see why that would be constrained to an entirely planet by planet basis as opposed to a global feature limiting the devs ability to allow procedural generation. Also I'm not necessarily asking for this to happen at launch, it'd be nearly pointless to do so since everyone's gonna take a while just to go interstellar in the first place. It sure would make for a great DLC once the games had a few updates, become stable, and most of the star systems have been well excavated though.

[Guest]

2 minutes ago, mcwaffles2003 said:

It sure would make for a great DLC once the games had a few updates, become stable, and most of the star systems have been well excavated though.

If this is your view than I agree with you, my point was that what is late game now in KSP will be mid-game at best in KSP2 without changing the difficulty that much, the scenario in which the user base needs more systems than the ones Star.Theory can hand-craft is a far-future one.

@micha we were talking about the original proposition, I get your idea, it's not that I don't like it (ok, I also don't personally like it, but that doesn't matter), it's just that they already confirmed that the Kerbolar system will have minimal changes, so if anything official will be procedurally generated it will be in other systems.

But I hope there will be an alternative mode in which you can start as a launch capable colony on a planet of your choice and build your space program from there, it should add a lot to the replayability factor whitout needing to implement a procedural system.

I'll be surprised if such a thing isn't already planned.

[mcwaffles2003]

10 minutes ago, Master39 said:

But I hope there will be an alternative mode in which you can start as a launch capable colony on a planet of your choice and build your space program from there, it should add a lot to the replayability factor whitout needing to implement a procedural system.

I'll be surprised if such a thing isn't already planned.

I cant cite the specific place/interview, maybe it was the one with shadowzone, but I think this was brought up and nate said that wouldn't be the case. I hope I'm wrong though.

Also, to my earlier DLC comment, I hope that it would simply be included in a later update so literally everyone has it, so it could see more use in what multiplayer may become, but if it's DLC I'll probably buy it.

  

10 minutes ago, Master39 said:

If this is your view than I agree with you, my point was that what is late game now in KSP will be mid-game at best in KSP2 without changing the difficulty that much, the scenario in which the user base needs more systems than the ones Star.Theory can hand-craft is a far-future one.

I wonder if we will be able to have mobile launch platforms integrated with interstellar ships... that would be dope

 

Edited January 23, 2020 by mcwaffles2003

[Guest]

3 minutes ago, mcwaffles2003 said:

I cant cite the specific place/interview, maybe it was the one with shadowzone, but I think this was brought up and nate said that wouldn't be the case. I hope I'm wrong though.

It's like "unmanned before manned" for KSP1 I think is such an obvious thing that if is not vanilla it will be a mod in the first months of the game, probably more than one.

4 minutes ago, mcwaffles2003 said:

Also, to my earlier DLC comment, I hope that it would simply be included in a later update so literally everyone has it

Basically I think the same, except I don't think I would buy it as a DLC, It will take hundred of hours to build the ideas I already have and do the missions I want to do, I think I'll have years to play in the starting systems before thinking about wanting more of them.

[mcwaffles2003]

1 minute ago, Master39 said:

It's like "unmanned before manned" for KSP1 I think is such an obvious thing that if is not vanilla it will be a mod in the first months of the game, probably more than one.

I really hope they remap the tech tree (and completely throw out career for something completely different but that's another topic) as unmanned before manned/PBC make a lot more sense progression wise.

4 minutes ago, Master39 said:

Basically I think the same, except I don't think I would buy it as a DLC, It will take hundred of hours to build the ideas I already have and do the missions I want to do, I think I'll have years to play in the starting systems before thinking about wanting more of them.

I think the extraplanetary launch pads are going to accelerate the exploration of this game by some insane amounts since you'll only have to do interstellar travel once per system essentially, though the time needed for building colonies might slow it right back down depending how its done.

[Guest]

52 minutes ago, mcwaffles2003 said:

I think the extraplanetary launch pads are going to accelerate the exploration of this game by some insane amounts since you'll only have to do interstellar travel once

I surely hope that, at least for Kerbin, making colonies is something a little more complex and organic than bringing in a single ship with a "colony in a box" kind of equipment and then just waiting until it develops launching capabilities.

We're going back to:

2 hours ago, Master39 said:

I think that if you can jump from system to system so fast and easy that you can get bored with the 2 or 3 (4, 5 or 6 the number doesn't change the point ) initial systems there something really wrong with the game difficulty.

 

The point is that we have only a general idea of the features of this game but we have no idea on how complex, entertaining and layered getting to those features is.

I hope that, except for sandbox, the game makes you build a lot of infrastructure to build an maintain Colonies, extraplanetary VABs and that orbital Spaceports are hard to build and maintain.

There's a lot of gameplay potential in these features I hope they don't oversimplify it into the Supermarket Loyalty Card system that science mode is.

[Guest]

My experience with procedural generation is that it usually results in dull, same-y locations. If you have good assets, you can get something that looks nice, but for true uniqueness, you need either an artist, or a lot of luck. Not to mention it's nice when the planets make sense physically, with procedural ones it's rarely the case unless some heavy constraints are placed on the generator.

[DStaal]

You can also do something like XCom2's semi-procedural generation: generation using large assets that have to fit together in specific ways.  If the scale is large, this could be individually crafted planets/solar systems in a procedural universe, or it could be 'continents' on a planet.

[mcwaffles2003]

10 hours ago, DStaal said:

You can also do something like XCom2's semi-procedural generation: generation using large assets that have to fit together in specific ways.  If the scale is large, this could be individually crafted planets/solar systems in a procedural universe, or it could be 'continents' on a planet.

This is along the lines of what I was thinking. Create a library of a lot planetary surface features and develop an algorithm that intelligently selects good combinations of features.

As for the base layer for the foundation of the planet a library of images could be selected from and a system as shown in the video could be automated:

Spoiler

 

Another method would be to develop a basic deep learning AI and have it spit out planets which could be voted on, thereby developing the AI's heuristics through community voting and once sufficiently advanced used to implement it to make randomized planets for everyone. The initial training for this AI could be based out of already made community based planets to give the AI a "feel" for what a planet should look like.

If a youtuber on the internet in college can develop an AI capable of making its own dance moves, jazz music, and rap lyrics... I see no reason an AI couldnt be developed by experience coders to make surface texture maps for planets:

Spoiler

 

 

[Superfluous J]

If No Man's Sky's infinitely variable planets are any indication of the state of the art of procedural planet generation then no, I would prefer dozens of hand crafted worlds to billions of the same thing over and over with a different palate and land masses in different spots.

In fact I wouldn't want them sullying the actual game content either, so wouldn't want them in addition to good planets either.

Make it a DLC for sure, so I can not get it and therefore not have to see them.

[mcwaffles2003]

4 hours ago, 5thHorseman said:

I would prefer dozens of hand crafted worlds to billions of the same thing over and over with a different palate and land masses in different spots.

Isn't that essentially all that differs between planets though? Gravity profile, atmosphere profile, color, and continent shape. Being real here the stock planets arent all that amazing, sure we have things like the mohole that add funny character to them but overall the greatest part about them, to me, is getting to them. As far as exploring them the lack of scenery in general is pretty intense which could be fixed up a bit with some random scatter of rocks/trees. But overall so long as the planets arent exactly alike with just a shifted hue I think somewhat random generation wouldnt be that bad. Simple rules like if it has water then craters should be minimal , If its tidally locked craters should be more prevalent on the side facing outward, and other simple rules that determine the topography of each planet in real life.

I feel if these exist and there are artist planets sprinkled in it would be similar to any painting or bouquet where you have the generic background that makes the truly unique and nuanced worlds really shine. We can go in depth about all the mechanics of planets formation and what not, but over a real galaxy with trillions of planets, the vast majority of them will most likely be very similar to one another among a small set of categories (gas giants, rocky planets, super earths, etc) which is perfectly suited for a random noise generator.

Just make sure the artist made planets are predominately in nearby star systems.