Concerned about KSP2 aerodynamics

[eekee]

I'm concerned because I heard that the developers of KSP2 are relying on the community's opinion of KSP1's aerodynamic model; saying it's predictable and intuitive. I was in open source software for a long time, and so I've seen many examples of people entirely forgetting the learning curve they once had to go through. They declare things to be "simple" and "intuitive" when they're really nothing of the kind, and I'm concerned that this is more of the same. Thinking lightly of KSP1's aerodynamics is nothing like the dizzying, surreal heights once reached by Git fanboys, but it still won't be good for newcomers to KSP2 who have never played KSP1. Let's compare KSP1 aerodynamics with real life.

Real Life

• If you're outside the vehicle, looking all around it, what you can see affects lift and drag. What you can't see doesn't. You need some education to understand how the profile of a wing affects its lift, and more to understand supersonic and hypersonic airflows, but these are all things which can be seen with the help of wind tunnels.

KSP1

• Parts with a designated wing area affect lift and drag even if they're hidden, unless the very center of the part (not shown in the editor!) is inside a cargo bay, unless again the part is a helicopter blade in which case it'll work inside a cargo bay too.
• Worse, parts which look hidden or at least shielded from airflow still create drag and are affected by reentry heating.
• Fitting parts side-by-side does not turn them into a lifting surface when it looks like they would be. Fitting hidden wing-parts to compensate adds mass without reason.
• Parts without a designated wing area have some effect on lift and drag, but it's inconsistent in unpleasantly surprising ways. The following examples may fairly be called bugs, but they are all the result of statically giving each part specific aerodynamic stats instead of evaluating the shape of the entire vehicle.
  • A 1.25m fairing made into a long cylinder with a point on the end will, when at an angle to the airflow, have several times more drag than an identically-sized cylinder of fuel tanks with a nose cone on the end. (There's some evaluation here, but only on the fairing part. It wasn't matched to the other parts.)
  • Mk2 parts have both wing and body drag, resulting in vastly greater drag than expected at some angles of attack. I wouldn't be surprised if this one detail is responsible for the majority of failed SSTO designs. It's infuriating to find, after very carefully fixing your craft's aerodynamics in the SPH, that it behaves completely differently in flight.
  • Mk3 parts = ??? I've seen very high drag from Mk3 cargo bays, but is the drag the same as a same-size Mk3 fuel tank, is it similar to 3.75m parts, and most importantly: why would a newbie ever think to check if cargo bays or all Mk3 parts are different? And it's not very easy to build a spaceplane big enough to compare big Mk3 and 3.75m parts.
• In place of wind tunnels, we have the aerodynamic overlay in flight. It's useful, but it's easy for a long drag triangle to be hidden inside a long fuselage.
• And why do the aerodynamic and cargo shielding systems work together so hard to prevent you making custom cargo bays?  I just watched a video with a nose cone attached to a hinge and a fairing closed on the nose cone. Perfectly reasonable, normal, intuitive engineering, especially when you consider the docking ports used as a catch to hold it closed. When it got out into space, the hinge was not allowed to open! What even is this? What line of logic tries to fight such basic simple practical engineering? It's as intuitive as a door in a house, but something about the design of KSP1's aerodynamics system made whole teams of developers think it was right to prevent it. And the alternative of using those tubular parts from Making History leaves everything inside the tube exposed to drag and heat because the node at the nose cone end isn't "properly covered". That's less intuitive than half of MS-DOS!

The only "intuitive" thing I can think of about KSP1's aerodynamics is that you can add up the wing area of all the parts which have such, but I am so tired of adding up all the different wing parts to even get an idea of my build's lift/mass. KSP is a computer game, it runs on a computer, why can it not compute the wing area and tell me what it is? And, here's the really interesting bit: if it's computing the wing area, why not instead make the more complex but realistic calculation of overall lift based on vehicle shape, as the FAR mod does?

[TROPtastic]

I agree that a lot of KSP1's aerodynamics aren't intuitive, especially for new players given the observations you've talked about. I think Intercept could go a long way to making the aerodynamic system easier to understand (perhaps that's what they want feedback for in Phase 1 of early access?) by getting rid of these inconsistencies and unintuitive behaviours. However, I'd like to address this line in particular as a KSP1 FAR noob:

11 minutes ago, eekee said:

And, here's the really interesting bit: if it's computing the wing area, why not instead make the more complex but realistic calculation of overall lift based on vehicle shape, as the FAR mod does?

I personally found the FAR mod to be a bit confusing to understand, in terms of the different readouts that are presented to the player in the VAB/spaceplane hangar. It's quite possible that most of these statistics were actually just "FYI", but if they weren't and they were indicative of all the stats you had to consider to make a flyable, controllable plane, it might well be too complex for the average player. However, Intercept could choose to implement a FAR-like aerodynamics model and only surface key indicators to most players (and do many calculations and optimizations behind the scenes), with detailed statistics and tuning offered to advanced players at the press of a button. That might be a good way of implementing a complex flight model without turning the game into XPlane or MSFS, for example.

[LHACK4142]

20 minutes ago, TROPtastic said:

I personally found the FAR mod to be a bit confusing to understand, in terms of the different readouts that are presented to the player in the VAB/spaceplane hangar. It's quite possible that most of these statistics were actually just "FYI", but if they weren't and they were indicative of all the stats you had to consider to make a flyable, controllable plane, it might well be too complex for the average player. However, Intercept could choose to implement a FAR-like aerodynamics model and only surface key indicators to most players (and do many calculations and optimizations behind the scenes), with detailed statistics and tuning offered to advanced players at the press of a button. That might be a good way of implementing a complex flight model without turning the game into XPlane or MSFS, for example.

They are mostly FYI. Scott Manley has a great video where he explains it all. Basically, the yellow line is stability, the blue line is lift, the red line is drag, and the green line is lift/drag ratio. If you go to the tab with the colored numbers in it, green numbers = good and red numbers = bad but might still work.

[Luriss]

3 hours ago, TROPtastic said:

I agree that a lot of KSP1's aerodynamics aren't intuitive, especially for new players given the observations you've talked about. I think Intercept could go a long way to making the aerodynamic system easier to understand (perhaps that's what they want feedback for in Phase 1 of early access?) by getting rid of these inconsistencies and unintuitive behaviours. However, I'd like to address this line in particular as a KSP1 FAR noob:

I personally found the FAR mod to be a bit confusing to understand, in terms of the different readouts that are presented to the player in the VAB/spaceplane hangar. It's quite possible that most of these statistics were actually just "FYI", but if they weren't and they were indicative of all the stats you had to consider to make a flyable, controllable plane, it might well be too complex for the average player. However, Intercept could choose to implement a FAR-like aerodynamics model and only surface key indicators to most players (and do many calculations and optimizations behind the scenes), with detailed statistics and tuning offered to advanced players at the press of a button. That might be a good way of implementing a complex flight model without turning the game into XPlane or MSFS, for example.

FAR isn't that bad, its like KSP 1 aero model; as long as you know the basics you can make a plane fly.

More importantly, the difference between FAR and stock is night and day. The KSP 1 aero model severely limits creativity for the reasons eekee mentioned and FAR fixes all of it.

[Bej Kerman]

I honestly can't see any conceivable reason Intercept wouldn't want to go as realistic as possible here.

[The Aziz]

I made a car and a dumpling fly in stock aero. How does that limit creativity?

[t_v]

I can see both sides of the argument here. On the one hand, while I played FAR, there was a much smaller range of shapes that worked in flight. Planes had to be shaped like planes and rockets like rockets to fly, there wasn’t really an opportunity to go for unconventional shapes because they would just be unstable (like in real life I guess, but figuring out the way to make something stable was infuriating trial end error for me). On the other hand, in stock, there is a much smaller range of what you can do inside your ships. If you want to make a big boat, everything you put inside it throws off the drag, and figuring out what is causing the problem and how to solve it is again, infuriating trial and error. 
 

One of the biggest issues I had playing with FAR is that compared to the stock model, a lot of the effects were difficult to account for and limited flying. Sure, not being able to turn as fast at high speeds is realistic, but it also makes amazing flying stunts impossible, or less impressive at lower speeds. I think that the stock aero model is intuitive and predictable not necessarily in the craft design but in flight. Once you’ve got something that can get into the air, it will if you just pull up. Once you are in the air, if you start spinning out with a vertical stabilizer, it is because you are holding a key. If you turn, you will turn the same way at all speeds, and then maybe have your wings ripped off as a result. If you are at a high angle of attack, there aren’t really any unforeseen aerodynamic considerations, other than that your wings may be dragging you back to prograde. 
 

I think that having a model that accounts for the shape of the craft would be better for those who want to be creative on their interiors, but we don’t need to include the more punishing or unintuitive aspects of FAR for those who want to be creative on their exteriors. 

[Bej Kerman]

8 minutes ago, t_v said:

Sure, not being able to turn as fast at high speeds is realistic, but it also makes amazing flying stunts impossible, or less impressive at lower speeds.

Well it'd be nice to be able to teleport anywhere in the Universe instantly but that's not realistic. Ultimately realism matters more than being able to make a grand piano fly.

[KerikBalm]

On 10/29/2022 at 7:02 AM, eekee said:

• • A 1.25m fairing made into a long cylinder with a point on the end will, when at an angle to the airflow, have several times more drag than an identically-sized cylinder of fuel tanks with a nose cone on the end. (There's some evaluation here, but only on the fairing part. It wasn't matched to the other parts.)

I wasn't really aware of this, can you supply some pics showing the drag readouts.

How much difference are we talking? 

I have taken to using very long cylindrical fairings at structural elements for lower part counts.

While I normally try to keep my AoA close to 0 and rely on incidence, obviously there is some deviation from this

[Dinlink]

Stock aerodynamics is so broken, that you can stabilize a rocket only by adding boosters, because, even if the center of lift indicates that your rocket is unstable, the fake drag of the boosters automatically stabilize any rocket...

If it's about intuition, I think this and other derivative behaviours get very not-intuitive,. Like you drop the boosters and suddenly the whole ship is unstable... Or if you add any part to the command pod, it's automatically unstable on re-entry... And so on...

In my opinion, simplified Lift/drag per part makes the game not intuitive and frustrating when trying to play with the aerodynamics (No SAS flywheel control).

[Nazalassa]

You can even use heat shields as wings...

[ColdJ]

I am hoping that being able to make things lifting bodies in configuration stays. Otherwise sci fi ships like from star wars would become impossible. Things that look good in a movie don't often do well with real aerodynamics coming in to play.

[t_v]

23 hours ago, Bej Kerman said:

Well it'd be nice to be able to teleport anywhere in the Universe instantly but that's not realistic. Ultimately realism matters more than being able to make a grand piano fly.

What I’m saying is that in this situation, realism cuts into the gameplay and makes it less compelling. Instead of going for more realism in the aero model like FAR does, KSP 2 could make the model more lax while keeping the benefits of a voxel based model. And gameplay matters more than realism. 

[Bej Kerman]

5 minutes ago, t_v said:

23 hours ago, Bej Kerman said:

Well it'd be nice to be able to teleport anywhere in the Universe instantly but that's not realistic. Ultimately realism matters more than being able to make a grand piano fly.

What I’m saying is that in this situation, realism cuts into the gameplay and makes it less compelling. Instead of going for more realism in the aero model like FAR does, KSP 2 could make the model more lax while keeping the benefits of a voxel based model. And gameplay matters more than realism. 

That'd hold water for something like Elite Dangerous where realism can take a backseat to gameplay (or what little gameplay there is in that specific example), but KSP 2 is supposed to be teaching players how rockets and aircraft work in the real world, not precisely the opposite for the sake of a few laughs. If anything, not being able to make literal made-out-of-clay bricks fly is the best outcome here.

[t_v]

22 minutes ago, Bej Kerman said:

That'd hold water for something like Elite Dangerous where realism can take a backseat to gameplay (or what little gameplay there is in that specific example), but KSP 2 is supposed to be teaching players how rockets and aircraft work in the real world, not precisely the opposite for the sake of a few laughs. If anything, not being able to make literal made-out-of-clay bricks fly is the best outcome here.

What I’m trying to say is that there is a balance between unreasonably lenient and unhelpfully realistic. There’s a point where the player understands that the aircraft can stall if they go beyond a certain angle of attack, and making the players’ unconventionally designed wing shapes stall out at every opportunity isn’t helping them learn how real airplanes fly, is is preventing them from really diagnosing the problem since it is happening under too many situations. Airplanes that perform as well as the ones in real life are surprisingly hard to design, and players will frequently be left with planes that don’t lift at low speeds, have a very narrow speed range, stall very easily, cannot turn even though the elevons are the same, the wing shape is just different, etc. The point is that the player will be flying really bad planes until they go through a hefty iterative design process for each plane they build that will leave them looking unfortunately similar to each other. Toning back the effects, increasing the tolerances, adding leniency will still teach players how to build and fly planes realistically but opens the door for new players to easily start that learning curve, beginning with getting off the ground, and also for more unique (still not completely unrealistic) craft to be built and flown. 

[Guest]

39 minutes ago, Bej Kerman said:

That'd hold water for something like Elite Dangerous where realism can take a backseat to gameplay

ED, like NMS and a bunch of other sci-fi space games follow their own version of what realism is for KSP, which is a specific style of sci-fi.

NMS prioritizes looking like a vintage sci-fi music album cover over everything, realism and gameplay included. ED prioritizes keeping the old Elite from the 80s as canon as possible over any kind of consideration for gameplay or style in this decade (that's why almost every ship looks like a door wedge or an arrow head).

 

You can find a similar trap in KSP when realism is considered or the educational purpose of the game is considered.

I don't use FAR, but I want a more realistic approach.

FAR is a bad target for a game because is pure realism without any consideration for gameplay. KSP2 should aim for a better model than what is stock now, but still something way more forgiving than FAR while also keeping in mind that they're erasing a decade of intuition in building planes in KSP while changing model and that requires a ton of in-game tutorials and explanation to get people up to speed as fast as possible on the differences.

I don't use FAR because it goes against everything I learned on how plane work in KSP and I don't want to start digging in 5+ years old forum discussions or YT videos to learn how to make planes in it. It's not worth my time.

[Bej Kerman]

51 minutes ago, t_v said:

The point is that the player will be flying really bad planes until they go through a hefty iterative design process for each plane they build that will leave them looking unfortunately similar to each other.

35 minutes ago, Master39 said:

I don't use FAR because it goes against everything I learned on how plane work in KSP and I don't want to start digging in 5+ years old forum discussions or YT videos to learn how to make planes in it. It's not worth my time.

If KSP has messed up your idea of how aerodynamics actually works to the point of giving up on FAR so you don't have to abandon being able to make bricks fly, then it's done a terrible, terrible, terrible job, plain and simple.

[Dinlink]

59 minutes ago, Master39 said:

ED, like NMS and a bunch of other sci-fi space games follow their own version of what realism is for KSP, which is a specific style of sci-fi.

NMS prioritizes looking like a vintage sci-fi music album cover over everything, realism and gameplay included. ED prioritizes keeping the old Elite from the 80s as canon as possible over any kind of consideration for gameplay or style in this decade (that's why almost every ship looks like a door wedge or an arrow head).

 

You can find a similar trap in KSP when realism is considered or the educational purpose of the game is considered.

I don't use FAR, but I want a more realistic approach.

FAR is a bad target for a game because is pure realism without any consideration for gameplay. KSP2 should aim for a better model than what is stock now, but still something way more forgiving than FAR while also keeping in mind that they're erasing a decade of intuition in building planes in KSP while changing model and that requires a ton of in-game tutorials and explanation to get people up to speed as fast as possible on the differences.

I don't use FAR because it goes against everything I learned on how plane work in KSP and I don't want to start digging in 5+ years old forum discussions or YT videos to learn how to make planes in it. It's not worth my time.

I must disagree, I have found FAR mod more fun and useful than stock aerodynamics... FAR is not unforgiving, the opposite, it helps you to design planes that fly by themselves, it helps you to know your plane before you even launch it... It's true that the User Interface is not perfect, and that's what I would expect from KSP2, to improve the user interface expérience while keeping a realistic aerodynamics model.

You don't need a ton of tutorials to build a good plane, in fact with FAR is just looking one graph, the lift graph and one number, the speed for sustained flight. FAR even helps you to position and size the tail of your plane to have an improved and smooth stable expérience (with the stability graphs)... Which in FAR user interface is not that obvious, but with a nice interface building planes would be as easy or easier than building rockets....

What's sets KSP apart from all the other games is its capability of making realistic simulations approachable... If we remove the realism of KSP we end up with another run of the mill "space game"...

 

Edited October 30, 2022 by Dinlink

[Guest]

1 hour ago, Dinlink said:

with FAR is just looking one graph, the lift graph and one number, the speed for sustained flight.

And to understand how that interface works and which number means what you have to dig through 6 years of posts.

 

1 hour ago, Dinlink said:

Which in FAR user interface is not that obvious, but with a nice interface building planes would be as easy or easier than building rockets....

Which is what I'm saying.

FAR is a terrible example if you want to propose a better aerodynamic model.

 

The same is valid for every other feature added by mods that, while still working and being maintained, have the required documentation and tutorials lost in time and in the spam of the forum.

 

1 hour ago, Bej Kerman said:

If KSP has messed up your idea of how aerodynamics actually works to the point of giving up on FAR so you don't have to abandon being able to make bricks fly, then it's done a terrible, terrible, terrible job, plain and simple.

I don't play KSP to study how aerodynamics work.

That's not KSP job or responsibility.

I play KSP because it's fun, and after 9 years I don't see the point of change how everything work for the sake of realism.

I gave up on FAR as soon as I saw that there wasn't the needed documentation to bridge the gap with the stock model.

The added gameplay challenge should be 100% in game. If the added difficulty is only in out of the game study and digging in the forum than it's not worth it.

[Bej Kerman]

16 minutes ago, Master39 said:

2 hours ago, Bej Kerman said:

If KSP has messed up your idea of how aerodynamics actually works to the point of giving up on FAR so you don't have to abandon being able to make bricks fly, then it's done a terrible, terrible, terrible job, plain and simple.

I don't play KSP to study how aerodynamics work.

That's not KSP job or responsibility.

If it isn't then it shouldn't have had a SPH in the first place. If KSP 2 is going to support aircraft like KSP 1 did then it will need to do a competent job of not letting you do silly and ridiculous things that should not be possible.

18 minutes ago, Master39 said:

I gave up on FAR as soon as I saw that there wasn't the needed documentation to bridge the gap with the stock model.

Yes, FAR is that far ahead of stock. There's no documentation to "bridge the gap" because the stock model is terrible and, as I stated earlier, is so unrealistic and easy to grasp that it's put a wave of newbies and veterans off of using something like FAR. The stock model needs to be left behind to rot in the stone ages.

20 minutes ago, Master39 said:

The added gameplay challenge should be 100% in game. If the added difficulty is only in out of the game study and digging in the forum than it's not worth it.

I'm sorry that you consider not being able to fly a literal grand piano difficulty. But, as I said, the stock model is archaic, and is almost actually insulting to anyone with a shard of curiosity about aviation given the time the devs have had to improve it. It's like if KSP 1 had infinite fuel on by default for the sake of "fun", because players would rather build silly things than learn how spaceflight works properly and find any fun in overcoming challenges. It'd be insulting to anyone with even a tiny grasp on how rockets actually work.

[Guest]

41 minutes ago, Bej Kerman said:

If it isn't then it shouldn't have had a SPH in the first place. If KSP 2 is going to support aircraft like KSP 1 did then it will need to do a competent job of not letting you do silly and ridiculous things that should not be possible.

Hard disagree, as far as flying in games goes KSP terrible model is way above average, it's terrible, but exactly as for orbital mechanics it's still better than most games (if you exclude dedicated flight sims, obviously).

 

42 minutes ago, Bej Kerman said:

Yes, FAR is that far ahead of stock. There's no documentation to "bridge the gap" because the stock model is terrible a

If you make a mod that changes stock and fail to provide the documentation on those changes than I'm sorry but you did less than half of the work.

I don't care how accurate FAR is or how terrible the stock model is, if it's not documented then it may as well not exist.

 

45 minutes ago, Bej Kerman said:

I'm sorry that you consider not being able to fly a literal grand piano difficulty. But, as I said, the stock model is archaic, and is almost actually insulting to anyone with a shard of curiosity about aviation given the time the devs have had to improve it. It's like if KSP 1 had infinite fuel on by default for the sake of "fun", because players would rather build silly things than learn how spaceflight works properly and find any fun in overcoming challenges. It'd be insulting to anyone with even a tiny grasp on how rockets actually work.

You missed the several times I've explained that I'm in favour of a better model.

I'm just pointing out that if you want a more complex aero model you should first and foremost have better in game explanations and tutorials.

 

[Bej Kerman]

1 minute ago, Master39 said:

52 minutes ago, Bej Kerman said:

If it isn't then it shouldn't have had a SPH in the first place. If KSP 2 is going to support aircraft like KSP 1 did then it will need to do a competent job of not letting you do silly and ridiculous things that should not be possible.

Hard disagree, as far as flying in games goes KSP terrible model is way above average, it's terrible, but exactly as for orbital mechanics it's still better than most games (if you exclude dedicated flight sims, obviously).

Terrible isn't good enough.

1 minute ago, Master39 said:

53 minutes ago, Bej Kerman said:

Yes, FAR is that far ahead of stock. There's no documentation to "bridge the gap" because the stock model is terrible a

If you make a mod that changes stock and fail to provide the documentation on those changes than I'm sorry but you did less than half of the work.

I don't care how accurate FAR is or how terrible the stock model is, if it's not documented then it may as well not exist.

Toyota isn't showing you how to upgrade your stone cart into a car. It's not happening.

3 minutes ago, Master39 said:

54 minutes ago, Bej Kerman said:

I'm sorry that you consider not being able to fly a literal grand piano difficulty. But, as I said, the stock model is archaic, and is almost actually insulting to anyone with a shard of curiosity about aviation given the time the devs have had to improve it. It's like if KSP 1 had infinite fuel on by default for the sake of "fun", because players would rather build silly things than learn how spaceflight works properly and find any fun in overcoming challenges. It'd be insulting to anyone with even a tiny grasp on how rockets actually work.

You missed the several times I've explained that I'm in favour of a better model.

I'm just pointing out that if you want a more complex aero model you should first and foremost have better in game explanations and tutorials.

So, guess I should forget all the bits you said you hate FAR because it doesn't let you make unrealistic abominations that shouldn't fly?

[The Aziz]

A lot of big names in this community wouldn't be there if it wasn't for those unrealistic abominations they managed to fly and crash. Going weird was always a part of KSP experience and that shouldn't change.

Also, toyota gives you a user manual with a car.

[Guest]

48 minutes ago, Bej Kerman said:

So, guess I should forget all the bits you said you hate FAR because it doesn't let you make unrealistic abominations that shouldn't fly?

If you want to have a debate at least have the decency to read who you're replying to and their posts.

[Bej Kerman]

1 hour ago, Master39 said:

2 hours ago, Bej Kerman said:

So, guess I should forget all the bits you said you hate FAR because it doesn't let you make unrealistic abominations that shouldn't fly?

If you want to have a debate at least have the decency to read who you're replying to and their posts.

I've gotten the gist of it - KSP 1 aero works "well enough" and isn't worth changing.

That mindset has never led to anything improving ever.

KSP 2 needs something at least close to FAR. With the amount of devs working on KSP 2 and how much of an upgrade everything else is, and how good FAR was with just a few modders having contributed to it, there's no excuse to leave the aero model behind in the dust. Just because it's harder doesn't mean it's worse. Avoiding challenge is never a good reason to shoot down improvements from KSP 1.