Kerbal Space Program 2: Episode 5 - Interstellar Travel

[Ahres]

4 minutes ago, Bej Kerman said:

…the fact the developers would have told us years ago if large-scale procedural generation was involved. 

What’s your definition of large-scale procedural generation? You seem to be ditching the “everything is handcrafted” stance I just called you out on. 

[Bej Kerman]

7 minutes ago, Ahres said:

22 minutes ago, Bej Kerman said:

…the fact the developers would have told us years ago if large-scale procedural generation was involved. 

What’s your definition of large-scale procedural generation? You seem to be ditching the “everything is handcrafted” stance I just called you out on. 

I'm saying the developers won't be generating entire solar systems. Stop deliberately misinterpreting my words and using them as a "gotcha!" argument.

[Ahres]

@Bej Kerman I know it appears it was deliberate, I promise it wasn’t. I took it as meaning you were saying the devs were doing procedural generation at small scales, like minor surface features only seen at close distances.
 

But there is no misinterpretation of “everything is handcrafted.” And I just don’t recall this ever being stated. I’d still like a source if you have one. 

Edited August 21, 2022 by Ahres
Meant to quote Bej, but forgot.

[Bej Kerman]

3 minutes ago, Ahres said:

@Bej Kerman I know it appears it was deliberate, I promise it wasn’t. I took it as meaning you were saying the devs were doing procedural generation at small scales, like minor surface features only seen at close distances.
 

But there is no misinterpretation of “everything is handcrafted.” And I just don’t recall this ever being stated. I’d still like a source if you have one. 

I'll spell it out for you: I'm saying that the devs won't generate entire solar systems. That's what I mean by "everything is handcrafted" and "no large-scale proc gen". I figured this'd be very self-explanatory and that I wouldn't have to argue about the semantics. You can stop picking everything apart now. End of.

[Wizard Kerbal]

Ooh lemme get the popcorn

[Ahres]

7 hours ago, Bej Kerman said:

I'll spell it out for you: I'm saying that the devs won't generate entire solar systems. That's what I mean by "everything is handcrafted" and "no large-scale proc gen". I figured this'd be very self-explanatory and that I wouldn't have to argue about the semantics. You can stop picking everything apart now. End of.

Fair enough Bej. I’ll admit I’m especially nitpicky on you because… of reasons stated earlier in the thread. 
 

I think you’re right that there won’t be hundreds of planets to explore. But I also wouldn’t be surprised if there’s much more than you think there are. 
 

Sorry @Wizard Kerbal, show’s over 

[Pthigrivi]

Obviously, the truth is we just don't know. We've seen some things, but in Rumsfeldian fashion we cannot know what we have not been shown. Not being personally experienced in planetary game design I couldn't profess authority on the subject but Im going to guess that a fair degree of procedural generation goes into any space with a surface area measured 50+ km^2 let alone multiple orders of magnitude higher for a few or even several star systems worth of planets. PD/Incercept showed some interesting glimpses into the process in the Planetary Architecting dev vid,  and talked about how terrain and scatter might respond to hand-crafted and curated topology. That said, given the hints we've seen about the kind of technological scale and the earnest attempts to make the game accessible for new players I'd be pretty impressed if the the scale of robust, Adventure Mode gameplay decisions extended over the playscale of as many as 5 total systems and I'd be positively boggled if it extended to more than 10 at the time of launch. In a game where you're not just clicking on a UI element and being transported magically there, where you're being asked to navigate brachistochrone trajectories and exotic gravity wells with hard dV costs over interstellar distances and to scout and mine rare resources just to survive and expand I would hope the emphasis was on the quality of that experience which we all know could take thousands of hours over just 2-3 systems, vs the quantity of  dozens let alone hundreds of poorly considered systems that extraordinarily few players will ever encounter.

I mean just look at the Starfield reveal where viewers were all right there excited and on board until Todd Howard told us there would be a hundred+ systems and everyone realized just how empty and ubiquitous so much of that space was guaranteed to feel. Maybe Im just not up on whats possible but it sure seems like a huge mistake. 

Edited August 22, 2022 by Pthigrivi

[Wizard Kerbal]

3 hours ago, Ahres said:

Sorry @Wizard Kerbal, show’s over 

That was the goal. I wanted to show you two it looked like we were on Reddit, not the KSP forums.

[Fullmetal Analyst]

did they actually promised sub-millimeter level of precision or did i misheard?

[t_v]

6 hours ago, Fullmetal Analyst said:

did they actually promised sub-millimeter level of precision or did i misheard?

I think the smallest scale of precision was described using the world “millimeter.” But I can’t remember what piece of media it was from, and I’m not sure it was this one. They were talking about spatial subdivisions in a dev diary or something. 

Edited August 31, 2022 by t_v

[Pthigrivi]

17 hours ago, Fullmetal Analyst said:

did they actually promised sub-millimeter level of precision or did i misheard?

Yeah right at the beginning of this vid. Absolutely bonkers. 

[The Aziz]

My guess it's because of used data type. It's not like we, as players, are gonna need this kind of precision, but it's there for stability reasons. Like, if there's a calculation error of sorts, it's going to solve itself within that 0.001m range, which means it won't affect the craft in any visible or significant way.

[t_v]

44 minutes ago, The Aziz said:

My guess it's because of used data type. It's not like we, as players, are gonna need this kind of precision, but it's there for stability reasons. Like, if there's a calculation error of sorts, it's going to solve itself within that 0.001m range, which means it won't affect the craft in any visible or significant way.

The instability of the data is proportional to the size of the numbers, and while different data types will have different precisions, an instability of 1mm every single time the game updates can cause some serious perturbations. But the reasoning and implementation of this is discussed in a dev diary, I thought so!

In this diary, there are these statements:

On 6/12/2020 at 10:39 AM, Intercept Games said:

if we state that our lowest spacial resolution is 1mm (which, honestly, is not nearly small enough for what we do), the maximum signed distance we can represent is 9.2 Trillion kilometers, which is just shy of a Light Year.

On 6/12/2020 at 10:39 AM, Intercept Games said:

we’ve solved this by implementing a Spacial Scene Graph at Interstellar Scales, which allows us to arbitrarily “break off” sections of space and simulate them with a high degree of precision while still fully understanding their physical and positional relationship to the stars and planets around them,

So, what does this mean? first, it means that 1mm precision is actually meaningful for the game. If you think about a kerbal moving at 1 m/s (not sure of their actual speed), 1 mm precision allows them to stop at 1000 points along that mater, which is helpful to not have stutters of 1 cm at a time (which would be noticeable both space-wise and time-wise as the kerbal would stutter like Mario between frames of movement). you might also want to position parts with a 1mm level of precision to gain even greater control. And parts colliding should only dip millimeters into each other, not centimeters. There are tons of reasons why 1mm can be useful as a highest precision, but let's talk about the other end:

the maximum signed distance being just under a light-year means the maximum playable area is a box of just under 2 light-years on a side. After that, even if there is nothing out there, leaving that box would require implementing a new coordinate system and managing the transition between the two. Not to mention that you still have to deal with the crazy bugs that happen with large floating-point numbers. Instead, why not accept the fact that multiple coordinate systems will be involved and make it so that the physics works near flawlessly by keeping numbers small and not using floating-point?

so, here's how the spatial scene graph works, at least from what I understand of it: you start with your imprecise coordinate system that only measures in increments of 100,000 m. but you don't want your ship doing 100,000 m jumps, you want it moving as smoothly as possible through space. So you take the area where your ship is and you simulate that to a precision of 1 m. This area either stays stationary or moves with the ship (we're really looking at how the origin moves) and its position and velocity is kept track of.

So your ship is moving in increments of 1m in a bubble of space that is at a set position (but that position is in an increment of 100,000m). Once your ship reaches the edge of that bubble, it moves 100,000 m and your ship keeps going. But 1 m precision also isn't good because all the parts would be clipped to that 1 m scale, and any kerbals on EVA or any moving parts would jump one meter with respect to the ship. 

So, you now take a smaller, fully precise bubble of space just around the ship. this bubble jumps 1m at a time inside the bigger bubble that jumps 100,000m at a time, while on the very inside, you can do your micro maneuvers to your heart's content. you have full precision while storing relatively small values in data, and you avoid any of the kraken-inducing bugs that floating point errors produce. It's a win-win-win! I wonder if it is now being implemented everywhere in the game instead of just interstellar space, since the devs are 'slaying the kraken'

Sorry for the long explanation, I just wanted to make sure that there were little to no misunderstandings for this particular thing. 

 

[Sky Kerman]

Where is the Kerboler system located in the galaxy? That green galaxy kind of looks like a magnetic cloud!

[magnemoe]

On 8/21/2022 at 5:59 PM, Bej Kerman said:

I'll spell it out for you: I'm saying that the devs won't generate entire solar systems. That's what I mean by "everything is handcrafted" and "no large-scale proc gen". I figured this'd be very self-explanatory and that I wouldn't have to argue about the semantics. You can stop picking everything apart now. End of.

Agree, I assume its very much like KSP 1 Mun, maps are used to map out craters biomes and resources then the small details are procedural generated, perhaps also with maps for how much rocks an area have.

[magnemoe]

On 9/1/2022 at 12:47 AM, The Aziz said:

My guess it's because of used data type. It's not like we, as players, are gonna need this kind of precision, but it's there for stability reasons. Like, if there's a calculation error of sorts, it's going to solve itself within that 0.001m range, which means it won't affect the craft in any visible or significant way.

Hopefully avoiding issues like this 

On load the two tanks moved around 70 cm to the side. 
Luckily it did not explode but it was hard to get into orbit coming in from Moho. Did not dare to aerobrake it so put it in an an high orbit around Kerbin. 

[The Aziz]

Exactly. I remember my asteroid catcher, being eaten in half by the asteroid. Most of the Kraken attacks probably happen when there's a positioning error on load and all the parts try to turn into Astral spikes resulting in the craft being yoten into nothingness at FTL speeds.

[Bej Kerman]

On 9/6/2022 at 1:32 PM, magnemoe said:

Hopefully avoiding issues like this 

On load the two tanks moved around 70 cm to the side. 
Luckily it did not explode but it was hard to get into orbit coming in from Moho. Did not dare to aerobrake it so put it in an an high orbit around Kerbin. 

Playing this mess for another year = a good alternative to KSP 2 EA, they said.

It'll be fun, they said.

[Bej Kerman]

I had a far worse issue - on two different crafts

On both of them, I used a cubic octagonal strut to place stuff inside - on one, this was to put a MkI passenger module on a node already occupied by another bunch of stuff, and on another, the same thing but for an engine plate. Timewarping them too much caused the strut to wobble until it wobbled so much it shook the craft apart.

Not an alternative to KSP 2 EA to me. Even small crafts can barely be flown in KSP 1 because it's so buggy.

Edited January 13, 2023 by Bej Kerman

[magnemoe]

On 1/13/2023 at 5:22 PM, Bej Kerman said:

I had a far worse issue - on two different crafts

On both of them, I used a cubic octagonal strut to place stuff inside - on one, this was to put a MkI passenger module on a node already occupied by another bunch of stuff, and on another, the same thing but for an engine plate. Timewarping them too much caused the strut to wobble until it wobbled so much it shook the craft apart.

Not an alternative to KSP 2 EA to me. Even small crafts can barely be flown in KSP 1 because it's so buggy.

The weirdest thing here is that everything except the 2.5 meter docking port, the beams and the wings are surface mounted on the two tanks who is surface mounted on an beam going down to the docking port. 
It was to keep the tanks inside the heat shield. 
But as its surface mounted it should have moved with the tanks. It was an pretty bizarre bug.