What direction do you want KSP to take.

[rkman]

Bad design would be to generate planets RANDOMLY...But worse the community really wouldn't be able to relate with each others..

That problem is not inherent to procedural content, it is inherent only to using random seed numbers to generate the procedural content.

There is no reason not to have procedural content with fixed seed numbers (so that all players get the same universe), other than the fear that a mod that can randomize the seed numbers would become more popular than stock ksp.

[Dunbaratu]

That problem is not inherent to procedural content, it is inherent only to using random seed numbers to generate the procedural content.

There is no reason not to have procedural content with fixed seed numbers (so that all players get the same universe), other than the fear that a mod that can randomize the seed numbers would become more popular than stock ksp.

The idea that randomly generated content must necessarily be boring is also a misunderstanding of how random content can be generated. It can be generated within limited parameters and with good algorithms operating from the random seeds to guarantee interesting landforms and interesting stuff is made. Take Minecraft as a good example. Not only can the entire terrain be re-genned from just one seed number, but it doesn't make flat uninteresting terrain.

[Dunbaratu]

rather procedural being a subset of random.

A tiny example of pseudocode to generate some content procedurally but not randomly:

for x = 1 to 10.

print x.

endfor.

A tiny example of changing it to add some random starting point to it:

x = milliseconds since epoch.

startX = rand(x)

for x = startX to startX + 10

print x.

endfor.

Stop pretending just so you can have an argument to derail the suggestion. That's two tiny examples of algorithms that generate admittedly boring content, both procedural, one fully deterministic, and one only partly so because the algorithm began from a more random starting point.

And they are making asteroids procedural last I heard.....

Using an algorithm that doesn't give the same deterministic content every time because it's seeded. By a value as close to randomly generated as is actually possible on a computer without using inputs external to the computer like waiting for how long it takes for a particle to be ejected from a block of radioactive material.

Edited March 31, 2014 by Steven Mading

[Dunbaratu]

You may have seen me write this before, but since we have paid for early access for a game in beta, I don't think there should be a group of "testers" who test the updates and get them early: as we buy the game before it is complete, I think every player should be testing at the same time.

We seem to have to wait for updates to be complete, a disadvantage usually found in released games...

And yet when we get the updates, they usually have some sort of bugs, a disadvantage usually found in early access games.

Please excuse my rant, but I object to having to wait longer when I am essentially paying for a game which is not complete.

I'm not sure I'd agree, having been a developer of stuff myself. If you release everything to too large an audience immediately, before it's tested, then you get the same exact errors being reported again and again and again, clogging up the issue tracking list. The big problems that everyone runs into get reported by everyone. Instead you first release to a small population, discover the big problems that are so common everyone hits them, fix those problems, and then for round two you release to a larger population to find the slightly less common bugs, then for round three a much larger population, and so on.

Even open source projects stored on public sites like github, that theoretically have every little change released to everyone in the public at once, still tend to flag the code as "developers only for now" and then promote it to "beta for testers" and only after a while do they promote it to "ready for the public now". You can access the newest code at any time but are warned in a big way that a lot of it doesn't work yet and unless you're trying to become part of the development process by making bug reports and participating in the QA of it, you should still stay away. (That if you're just trying to use it as an external end-user, it's not ready for that yet.)

[TheTennesseeFireman]

I think a lot of the worries here are unfounded. Realism seems to be a sore point for some, but I feel like the developers have struck a good balance with that already. For example, I don't think life support would add much to the game since time is so difficult to judge, but if it were ever implemented, I'm sure it would be done in a way that's balanced and not obtrusive, like small pods being attached on longer flights to make time another resource to be balanced against weight. Unless the devs break several of their own design rules or actively avoid making interesting content, I'm not too worried about them straying down the wrong path.

At its core, KSP is a game about space exploration. The moment you stop visiting new places and building new things, you lose all replay value. The ARM/0.24 updates are a good indicator of what I'm hoping for going forward: new places to visit, parts that make new structures possible, and new mechanics that add depth to the existing gameplay without undermining it.

As for procedural generation of planets/systems, I doubt it can be done well. Maybe you can tweak the size/color/atmosphere/orbit numbers, but how do you make them real places with interesting terrain, textures and gameplay? I'd take a limited number of well-thought out places with interesting mechanics and scenery over a bunch of different-colored clones of planets we already have. Perhaps there's an algorithm that can prove me wrong, but I have doubts.

[Kanukki]

I don't think that the debate is really between a 100% common fixed solar system and and 100% random/procedurally generated system, different for all players. The ARM update shows the way towards a balance between keeping the current 'known' solar system as is for all players, and adding some element of the unknown and exploration with one or two extra procedurally-generated bodies outside the known system. This would allow players to learn all the core elements of the game from the community by having how-to guides, common reference materials on orbital mechanics, transfer windows, etc, but also give an opportunity in the late-game for unique player experiences with greater elements of risk and independent problem-solving, since no one would have gone to your particular generated 'new planet' before and uploaded simple instructions on exactly what you need to do to get there.

[rkman]

As for procedural generation of planets/systems, I doubt it can be done well. Maybe you can tweak the size/color/atmosphere/orbit numbers, but how do you make them real places with interesting terrain, textures and gameplay? I'd take a limited number of well-thought out places with interesting mechanics and scenery over a bunch of different-colored clones of planets we already have. Perhaps there's an algorithm that can prove me wrong, but I have doubts.

It has been done, see Elite, and more recently Spaceengine.

That is not to say it is easy to make procedural content that looks nice, but it can be done.

[cantab]

( SQUAD has said they don't want to do procedural content.

I wonder if Squad themselves didn't conflate procedural with random here. After all, they already put procedural non-random craters on the Mun.

Oh, and there are plenty of ways to get true random numbers into a computer. ERNIE uses thermal noise, random.org uses atmospheric noise, radioactive decay based RNGs have been done, and though you can argue whether it's "truly" random user input such as mouse movement is commonly used.

[Kegereneku]

That problem is not inherent to procedural content, it is inherent only to using random seed numbers to generate the procedural content.

There is no reason not to have procedural content with fixed seed numbers (so that all players get the same universe), other than the fear that a mod that can randomize the seed numbers would become more popular than stock ksp.

Unlimited number of Fixed Seed doesn't solve the problem of unrelatable game experience in any way.

Except maybe if it become a running jokes that nobody except a few hardcore fan will play on more than the "standard seed" because everybody else are satisfied with it.

I would prefer to have more thing to do on a few token planet, that having nothing to do but land on an infinite number of planet.

Remember : KSP =/= Minecraft.

Random seed in Minecraft was a necessity because the whole interest lies in being creative with your surrounding.

A non-randomly made unique solar system in KSP is a necessity because the interest lies in creating increasingly efficient design for given parameter, the risk of finding randomly boring seed defeat this.

Furthermore, the completed game is expected to have a career mode with budget limitation and a long playtime. Unlike Minecraft you cannot just "move on" once you get bored somewhere.

So in term of game-design you either go with an infinite number of barely-interesting planet, or you make sure the one solar system is choke full of everything you want to do.

I would love to see other solar systems implemented (but only when the Kerbol system is complete.) I want NO FTL, but other more realistic systems (perhaps this is something that Squad can consult with NASA on?) perhaps they could have Cryo-Sleep?

People asking for that usually don't realize how boring non-FTL flight would be, even in a video game. Especially if you keep withing the classic all-contained spaceship design (as opposed to, say, solar-sail impossible in game for other reasons)

What would you even do with such possibility ? Build even bigger rocket made of 99,99% FUEL, timewarp for eternity to get slingshot opportunities, then rage-quit because on arrival you see that you forgot a piece on the tiny probe you managed to launch.

Better go with a carefully balanced fictional FTL. It would at least be interesting to play.

It has been done, see Elite, and more recently Spaceengine.

That is not to say it is easy to make procedural content that looks nice, but it can be done.

Both game rely on the feature for other reason than KSP. Elite is Pioneers IN SPACE, SpaceEngine is centered around a stellar scale, and in both you don't design your rocket up to the very control system.

[rkman]

Unlimited number of Fixed Seed doesn't solve the problem of unrelatable game experience in any way.

I definitely did not say there should be an "unlimited" number of seed numbers.

Moreover, you misunderstand what i mean: I did not mean there should be multiple seeds numbers to choose from which each would generate a different universe, instead i assume multiple seed numbers are required to generate various aspects of one and the same universe (e.g. a seed for star distribution, another seed for planet distribution).

That's why i said "so that all players get the same universe".

Elite is Pioneers IN SPACE, SpaceEngine is centered around a stellar scale, and in both you don't design your rocket up to the very control system.

SpaceEngine allows visits to the surface of planets/moons and those are not uninteresting, contrary to you objection to procedural content. The fact that SpaceEngine does not (yet) include rocket building is irrelevant to that.

[Red Iron Crown]

A non-randomly made unique solar system in KSP is a necessity because the interest lies in creating increasingly efficient design for given parameter, the risk of finding randomly boring seed defeat this.

That's *one* type of interest in KSP, but it's not the only one. I'm sure many players would like exploring bodies no one has visited as much or more than refining their craft further and further. For that matter, have a poke around the Spacecraft Exchange, you'll see that many players don't care about efficiency at all.

So in term of game-design you either go with an infinite number of barely-interesting planet, or you make sure the one solar system is choke full of everything you want to do.

I'm not sure that's a valid dichotomy. There's no reason random, procedural content can't be interesting; certainly the procedural engine has to be more sophisticated, but it's not impossible.

[DaRocketCat]

As realistic as possible as long as it doesn't detriment the gameplay too much.

And keep the UI simplistic, intuitive, and easy to understand, as it is now.

For instance, I think re-entry heat should definitely be added, despite the fact that it will incinerate unprotected ships.

Edited April 1, 2014 by DaRocketCat

[edallison655321]

I love the update, best one yet. I would like to see planets with tilted axes like Uranus and Venus and moons in a retrograde orbit.

[wasmic]

I would like the game to continue the direction it is on now, generally. Probably with simple reentry mechanics, but I'd love fully functional FAR-like aerodynamics, which should have been there from the beginning IMO. I'm not keen on life support, though.

[Kegereneku]

I definitely did not say there should be an "unlimited" number of seed numbers.

Moreover, you misunderstand what i mean: I did not mean there should be multiple seeds numbers to choose from which each would generate a different universe, instead i assume multiple seed numbers are required to generate various aspects of one and the same universe (e.g. a seed for star distribution, another seed for planet distribution).

That's why i said "so that all players get the same universe".

[.....]

SpaceEngine allows visits to the surface of planets/moons and those are not uninteresting, contrary to you objection to procedural content. The fact that SpaceEngine does not (yet) include rocket building is irrelevant to that.

Changing the parameter is no different from generating unlimited amount of custom/random planet, and the problem isn't in the number (although it's an aggravating factor) it's in the fact that players would be playing different system. They can't be in the "same" Universe/Galaxy/System.

Unlike KSP which is/was Firstly about designing and navigating SpaceEngine exist for the purpose of sight-seeing and playing with orbital mechanics, it do not compute anything the same way and is not suitable for a KSP-like gameplay (there was a thread about that with technical explanation).

That's *one* type of interest in KSP, but it's not the only one. I'm sure many players would like exploring bodies no one has visited as much or more than refining their craft further and further. For that matter, have a poke around the Spacecraft Exchange, you'll see that many players don't care about efficiency at all.

[............]

I'm not sure that's a valid dichotomy. There's no reason random, procedural content can't be interesting; certainly the procedural engine has to be more sophisticated, but it's not impossible.

I don't want to sound like I know better than everybody but procedural generation can only do so much without human input, as of now there's not much to do on planet, that's probably why people wish they could generate themselves hollow asteroid, Land of Water and Magma, low gravity moon with insane pressure, gaz giant with one solid landing zone...etc. But random generation don't guaranty interesting planet. Statistically it should be averagely boring.

Then, there's also only so much you can do with input from player who don't know how procedural generation work and what is best (or how long it take to generate a planet).

About my mention of efficiency, what I mean is that there is players who like to build complete infrastructure (satellite and all) and would be frustrated to know they devoted hours of work for a solar system that isn't the best the devs could make. We cannot consider the KSP BETA's save-breaking to be a normal.

All the problem with unrelatable game experience still apply too.

[rkman]

Changing the parameter is no different from generating unlimited amount of custom/random planet, and the problem isn't in the number (although it's an aggravating factor) it's in the fact that players would be playing different system. They can't be in the "same" Universe/Galaxy/System.

It is you idea to "change parameters", not mine. It is entirely possible to have procedurally generated content that is identical for all players. That is what i meant and that is what i said, and others in this discussion have explained the same.

[KASASpace]

Elite was procedural, yet it always had Lave (the starting planet), the same galaxy, and always had the Thargoids. Every time you started a new save.

In minecraft, when you put in the same seed every time, you get the same world.

So, with Procedural generation, you CAN make the same world for everyone.

It would make the game smaller (less memory).

Plus I think the Asteroids in ARM are procedurally generated.

[KASASpace]

A tiny example of pseudocode to generate some content procedurally but not randomly:

for x = 1 to 10.

print x.

endfor.

A tiny example of changing it to add some random starting point to it:

x = milliseconds since epoch.

startX = rand(x)

for x = startX to startX + 10

print x.

endfor.

Stop pretending just so you can have an argument to derail the suggestion. That's two tiny examples of algorithms that generate admittedly boring content, both procedural, one fully deterministic, and one only partly so because the algorithm began from a more random starting point.

Using an algorithm that doesn't give the same deterministic content every time because it's seeded. By a value as close to randomly generated as is actually possible on a computer without using inputs external to the computer like waiting for how long it takes for a particle to be ejected from a block of radioactive material.

Oh, you want some code?

Fortran(old, but it's mathematically based):

C23456789012345678901234567890123456789012345678901234567890123456789012
      PROGRAM ProceduralTest

      Implicit None

      Real xSeed, ySeed
      INTEGER IX, JX, KX, MX, IY, JY, KY, MY, ICNT
      INTEGER JX2, KX2, MX2, IY2, JY2, KY2, MY2 
      LOGICAL found

C-----------------------------------------------------------------------
C -- This program will procedurally generate a set of numbers
C-----------------------------------------------------------------------

C -- These are the seeds, or the base numbers
      xSeed = 1342
      ySeed = 3429

C -- Calculate the points from the x seed
      IX = NINT(xSeed)
      JX = IX/100
      KX = IX - JX*100
      MX = JX + KX
C -- Calculate the points from the y speed
      IY = NINT(ySeed)
      JY = IY/100
      KY = IY - JY*100
      MY = JY + KY

C -- Output the two separate numbers and the seed
      WRITE(6,100) xSeed, JX, KX
100   Format(/1x, 'The seed was........ ', F6.1,
     $       /1x, 'The first number is. ', I5,
     $       /1x, 'The second number is ', I5,
     $       /1x, '---------------------------------------------------')

C -- Output the two separate numbers and the seed
      WRITE(6,200) ySeed, JY, KY
200   Format(/1x, 'The seed was........ ', F6.1,
     $       /1x, 'The first number is. ', I5,
     $       /1x, 'The second number is ', I5,
     $       /1x, '---------------------------------------------------')

C -- Here is the "found" variable
      found = .false.
      MX2 = MX
      MY2 = MY
      ICNT = 1 ! Delete later
      Write(6,*) ICNT-1, MX2, MX2, MY2, MY2 

C -- This is the "Do" loop
      DO while ( .not. found )
         MX2 = MX2 + KX
         JX2 = MX2/100
         KX2 = MX2 - JX2*100
         MY2 = MY2 + KY
         JY2 = MY2/100
         KY2 = MY2 - JY2*100
         Write(6,*) ICNT, MX2, KX2, MY2, KY2
         If ( KX2 .eq. MX ) found = .true.
         If ( KY2 .eq. MY ) found = .true.
         ICNT = ICNT + 1
      END DO

      STOP
      END

That's fortran. That last "do" loop was to actually procedurally generate the numbers. It also checks when the numbers start to repeat.

Oh, yeah, I also made one with user input, and once I actually completely filled the entire screen. Overload.

Edited April 5, 2014 by KASASpace

[Vanamonde]

Since the subject of the thread is the overall direction the game might take, let's please not get into arguments over specifics of coding things which aren't even in the game yet, okay guys? Some of you need to take the angry tone down a notch, and if you do want to discuss details of coding, please take it to the off-topic lounge, since it's not really about the game anymore at that point. Thanks.

[P.Funk]

Every hardcore sim I can think of has a menu that lets you turn it from something so realistic that you need to read at least one 500 page PDF file and study it for weeks down to something thats as arcadey as a gameboy game. Options are always better than no options. Options are the holy grail of gameplay in my opinion. Its the difference between making a good game for most people and a great game for everyone who plays it.

Too few developers appreciate the value of offering options to their players. Consoles particularly have gutted this I believe and it has leeched into some of the PC gaming mindset. Nevertheless this is a sandbox game at heart. Sandboxes are all about options and thats inherent to gameplay. Why shouldn't we have all the options we want on the meta level as well?

Give us hardcore difficulty with reentry, aerodynamics, life support, remote communications, the works. Then let everyone switch those off and play a stripped down version if they so choose. It doesn't harm the gameplay. Its easy to strip a difficulty feature from a game and see no change than to try and shoehorn one into a game that doesn't quite fit it. In the end playing KSP with all those realism switch turned off will be akin to playing KSP by deactivating your mods.

I have no idea why some people feel upset with options. Sometimes it feels like someone else playing the same game at a harder level than them emasculates them or something. I know for a fact that making all these "hardcore" options off by default will go over better than if they are on. I just know thats how the psychology will break down.

[karolus10]

Stuff like deadly reentry, arcade/ (more) realistic aerodynamics, wind, weather, radio communications and delays or crew need to breathe and eat are perfect examples of features that ability to turn it ON and OFF decide about game difficulty level.

Adding more complex set of game mechanics, flight instruments and editor/flight-planning tools is a good and desirable thing (It's adding more depth to late game) until you aren't forced to use them all to achieve anything in game - You can still slap few parts together and "go for it", but you will discover and use more and more of the game capabilities and tools as You progress (i.e. get some skill and knowledge) and look for another challenge.

Also I believe that after game is pretty much finished it will receive more developed tutorial and training scenarios, making entry even more painless and giving more tools for the community to create complete missions and scenarios.

[Kegereneku]

A last post from me, some people misunderstood me.

It is you idea to "change parameters", not mine. It is entirely possible to have procedurally generated content that is identical for all players. That is what i meant and that is what i said, and others in this discussion have explained the same.

Elite was procedural, yet it always had Lave (the starting planet), the same galaxy, and always had the Thargoids. Every time you started a new save.

In minecraft, when you put in the same seed every time, you get the same world.

So, with Procedural generation, you CAN make the same world for everyone.

It would make the game smaller (less memory).

Plus I think the Asteroids in ARM are procedurally generated.

Before, people were talking of Procedural Generation based on "seed", which implied different universe based on "parameter" and as argued before, pose numerous problem, from planet with less attention to detail (low memory = lesser texture), requiring powerful computer just to generate a planet and also the whole "break the community" thingy with unrelatable game experience.

NOW, if you want a UNIQUE solar system Procedurally Generated By your own computer, just know that procedural generation of small asteroid is EASY, big planet not so much. Plus that's not the key to galaxy sized universe (supposing it's even fun to navigate between stars without FTL gameplay)

Also : The game Elite was repetitive as hell and didn't have much of a physics engines, aerodynamic, or anything too complicated. You might believe everything can scale up proportionally in term of Virtual World feature but no, some things follow the cube law when computer follow mostly square augmentation.

[aeTIos]

I know how many people are hyping for procedural planets. May I point out that while Minecraft's procedural world is excellent, this wouldn't work in KSP? Remember how Minecraft looks very familar every time you load up a new world. This'll also happen with procedural generation. There will be a kerbin-like planet, that looks the same, breathes the same feeling, is only more boring because you know Kerbin inside out. Same goes for Eve, Moho, Duna.... I believe it's more valuable if the devs add a whole bunch of planets one time with interesting features and/or add more interesting features to existing planets. Like Novasilisko once suggested: geysers on Eeloo, volcanoes on Laythe and Kerbin. May I add a few? I'd like a mount olympus like mountain on Duna, hexagonal-ish clouds on Jool just like Saturn has. PROCEDURAL weather. Because procedural things can be used to spiff non-procedural things up. They can be a part of existing systems, just like the asteroids: They are all the same, but different.

Just my $0.02

[Kamsko]

Every one keep saying features should be added but with an on/off switch.... why? Mods are by their very nature exactly this.

[cantab]

We can already turn off fuel usage, and I believe part destruction on crashes. Things like life support would be reasonable extension.

On the other hand, I feel that in a game like KSP the basic physics should not be player-selectable (aside from mods). If the stock aerodynamics get improved, I don't think there should be a button that says "use old aerodynamics". (There could, though, fairly be an option for NO aerodynamics, like we have an option for no gravity.)