This could get some excitement. Maybe a community planet challenge

[Guest]

I think it would be cool if KSP 2 team had a planet challenge at some point in the future and you design a planet and submit it. Winner get's their planet to be stock in game. I think it would be a cool idea but yet again most people don't know how to make planets. Thoughts?

[jastrone]

idk if its realistic or not but a gas "giant" made up of very heavy gases but being farely small making some parts bouyant when deep into the atmosphere

[K^2]

I hope they make adding star systems as mods relatively straight forward. Then most of the wilder ideas people have for planets can simply be fan-made mods.

Making planets is a lot of work, though, if you want to end up with something that looks nice in the end.

[Dr. Kerbal]

We could use Universe Sandbox to make the general look of a planet. Or Blender or you know.

[K^2]

14 minutes ago, Dr. Kerbal said:

Or Blender or you know.

For a shape, easy. For rough outline of continents, not too bad. But by the time you're starting to work on detailed coastline and mountains, you're going to be hating you ever started the project. You really need some user-guided procedural tools that can handle different models for weathering simulation. I know that tools for designing open world maps like this exist, but I don't know how much work it'd be to try and do something like this on a scale of a planet. Ideally, you want specialized procedural tools for this sort of work.

[Dr. Kerbal]

6 hours ago, K^2 said:

For a shape, easy. For rough outline of continents, not too bad. But by the time you're starting to work on detailed coastline and mountains, you're going to be hating you ever started the project. You really need some user-guided procedural tools that can handle different models for weathering simulation. I know that tools for designing open world maps like this exist, but I don't know how much work it'd be to try and do something like this on a scale of a planet. Ideally, you want specialized procedural tools for this sort of work.

I know. That why I said “the general look”. I’m just mean the general look of the planet not the mountains.

[jastrone]

8 hours ago, K^2 said:

For a shape, easy. For rough outline of continents, not too bad. But by the time you're starting to work on detailed coastline and mountains, you're going to be hating you ever started the project. 

you could use a displacement map

[pandaman]

If the idea is for entrants to create a 'concept', using maps and illustrations etc to help communicate the ideas, then I could see it being an interesting thing.

The actual 'creation' of the in game assets themselves would be much better done by the dev team IMO.  That would mean everyone has a chance for their idea to be considered, not just those with the tools and skillset.  And also ensure that it works in game and blends in visually with the existing planets etc.

[mcwaffles2003]

11 hours ago, K^2 said:

For a shape, easy. For rough outline of continents, not too bad. But by the time you're starting to work on detailed coastline and mountains, you're going to be hating you ever started the project. You really need some user-guided procedural tools that can handle different models for weathering simulation. I know that tools for designing open world maps like this exist, but I don't know how much work it'd be to try and do something like this on a scale of a planet. Ideally, you want specialized procedural tools for this sort of work.

you may wanna check this out:

Spoiler

 

 

[VoidCosmos]

Or create a name for a planet 

[K^2]

11 hours ago, jastrone said:

you could use a displacement map

And are you planning to create millions of square kilometers worth of displacement maps by hand? Even at 10m resolution, you're looking at something like 360,000 x 180,000 pixels for Kerbin.

8 hours ago, mcwaffles2003 said:

you may wanna check this out:

This looks good from the distance, but if you land on a surface of something like this, it's going to be the most bland and boring place.

Now, KSP planets aren't exactly exciting, except for a few hand-crafted features, and even these are pretty low resolution. But even getting just that requires far more work than shown in the video above. Sure, you can throw a noise map over the whole terrain, and you'll get bumps and "hills" that all look pretty much the same. To get a good looking planet, you have to generate actual features, be they mountains, rivers, canyons, craters, etc. in some combination. From the looks of things, a lot more care is going into KSP2 planets. To get to something that won't look out of place in the game, you'll need to put a lot of work by hand on top of having a generation system that will fill in the rest of the environment.

I've linked a bunch of videos with some workflow examples that are either fully procedural or artist-driven procedural in another post.

On 12/23/2020 at 3:31 PM, K^2 said:

That is a bit of an understatement. Just to get the very basics of what the planets and moons look like, you need a ton more parameters. Here's a good exploration to get you started.

But this just gets you the basics. To make something that looks good up close, you need to do many times more work. Here's a sampling of talks I'd start with if I was planning something like this. Note that some of these go over artist-driven tools, but can be extended to procedural generation.

  Reveal hidden contents

you may wanna check this out:

Some of these techniques I've worked with, and they require a fairly significant team to pull them off. Now, it looks like we're already getting some H:ZD style vegetation placements, so some of the work has to be done regardless, but if I was asked to plan procedural generation of planets for a game like KSP2 on the current schedule, I would budget for 3 engineers, at least one senior and two others at least solid mids, a lead, at least one additional tech artist with some experience in terrain tools, and either a design-minded artist or an art-minded designer with experience in terrain as well. Team of that size is also going to be basically a full-time job for a producer. So we're talking about addition of at least 7 fairly senior developers for two years.  A senior technical dev in  Seattle area might average, ballpark, $170k per dev per year (benefits and all), so the total cost is going to be in the $2M - $2.5M range, depending on how senior the team is.

This is a lot of money for something fairly experimental, and something that will certainly not have the same quality as the bespoke planets and star systems and with questionable benefit to the players.

I would very much like to see KSP2 team keep this open for modders, though. As I've noted, some procedural generation is almost guaranteed to be present, so having an API to control what and how gets generated on modded planets would be fantastic. And that will open up the path for either modders to create custom star systems and planets or even experiment with procedural generation. Though, I do not have high expectations, unless a talented mod team makes it their passion project and spends a few years on it.

[jastrone]

15 hours ago, K^2 said:

And are you planning to create millions of square kilometers worth of displacement maps by hand? Even at 10m resolution, you're looking at something like 360,000 x 180,000 pixels for Kerbin.

not like that. rather fractal noise like where you ex have one map for sea and land one for mountains and then one small that repeats but gives more detail and one mor smaller to make some variation

[K^2]

38 minutes ago, jastrone said:

not like that. rather fractal noise like where you ex have one map for sea and land one for mountains and then one small that repeats but gives more detail and one mor smaller to make some variation

That will never look good by itself. Real terrain has a lot of spatial correlation that you inherently don't with noise, so any terrain based on just noise always looks both bland and very fake. You don't want just random variations in height. You want mountain chains, riverbeds, canyons, maybe craters. And you don't want these to have the same surfaces either. Whether pristine on an airless world or eroded over time on a world with atmosphere, a rim wall of an impact crater can't look the same as the rest of your terrain.

Yeah, you can generate all these features from noise, but at each level of magnification, you are looking at a completely separate generator, and for every type of world it's going to be different. You are looking at dozens of processes that all influence each other, and you need to make sure they all work together. There are tools for doing this kind of work for games and cinema. And I suggest you take a look at how involved they get just to get a single scene of a single environment. And while the process isn't going to be fundamentally different for doing this at planetary scale, I'm not aware of any tools that work on a globe. Intercept will probably need to either roll their own or adapt one of the existing tools. In either case, I doubt we'll have access to such tools, which means it will be an enormous amount of work for any artists who'll want to build their own worlds, either because of huge amount of work that will have to be done by hand or because of additional tools that will need to be made.

Here's an example of a tool used to create terrain for games. Take a note of how complex the graph gets.

 

[pandaman]

I strongly suspect that some people are greatly underestimating the amount of work needed to actually model a planet well enough to make it interesting enough and suitable for use in game.

Yes, like any CAD or graphics software, pencil and paper or paint and canvas, with a little practice you can do a lot of the big stuff quite well and quite quickly, but it's the little details that it needs to give it character and 'polish' that can really take the time, and that just does take time even for highly skilled experienced artists.