[0.90]NEAR: A Simpler Aerodynamics Model v1.3.1 12/16/14

[ferram4]

But that kind of compromise doesn't make any kind of logical sense, especially since it would make trying to figure out if something will work much more difficult. You wouldn't be able to simply ignore aerodynamics, like in stock KSP, nor could you base stuff on real life; it'll be some halfway that will leave everyone more confused in dealing with a feature that is certainly not an easy thing to deal with.

To me, arguing for drag depending on mass (in any way) is akin to arguing that the orientation of rocket engines shouldn't matter, that they should just work, regardless of where they're pointing. Neither one is realistic, neither one really makes sense, and both are only fun in the way that ignoring reality is fun, but then it becomes boring because the challenge is gone.

[rynak]

Not proposing a solution, just trying to identify the "problem":

I don't think any sane player expects aerodynamics to matter, while at the same time making loss of control impossible. However, in most "forgiving" plane simulators, it also is the case that:

1. The fly-by-wire controls tend to take this into account, reducing the risk of such a thing happening (SAS doesn't understand, right?)

2. There always is a way to recover in reasonable time (relative to the original intensity of control loss), as long as you got enough altitude/time to do so.

So, i don't think it makes sense to expect spins and stalls to not happen - what does matter i think, instead is recovery and SAS having the slightest clue about those effects (yes, i know the later might be hard or even impossible to fix, without outrightly replacing SAS with an own custom "AI" - not exactly lightweight, is it?).

[JedTech]

I agree, stock aerodynamics is kinda silly and not very logical. Well I guess a half silly / half realistic aerodynamic model is neither Near nor Far.

[ferram4]

@rynak: Well, the first thing to remember is that most flight simulators are already working with stable aircraft; KSP has no guarantee of that, and unless a plane is designed exactly right there's no way a control system can keep its instabilities under control. Then vehicles need to be designed to make recovery even possible; there are very specific things that need to be done to aircraft to make them want to come out of stall and allow them to come out of spins, and if a player doesn't add those (forward lift surface stalls first, tail has lots of control authority) you'll never be able to remove the issue.

Even if we had SAS that was capable of saving players from their mistakes it would only work if the mistakes were pure piloting and within a certain tolerance. The only way to maintain it within a certain tolerance would be to have always-on SAS, which would annoy good players and allow poor pilots to blame something external for their failings (even though it's off by default in FAR, that ended up being the reasoning used by some people and is one of my reasons for removing the control systems from NEAR).

The ultimate problem is that it's not just an issue of piloting. It's an issue of vehicle design. No amount of SAS can save a bad design short of reaching into the physics and changing them to save the player, which seems very... wrong. How is the player supposed to learn if the world isn't consistent and its laws will bend to save them? And that's before the condescending aspect of it, since it basically comes down to the developer saying, "We don't think you're smart enough to learn how to handle this."

Edited July 19, 2014 by ferram4
good != bad

[problemecium]

lol @ our overzealous censoring software butchering "a n a l y s i s" xD

[Kingsix]

I may have just found a bit of a bug with NEAR, but maybe its a bug with KSP 0.24. A particularly poorly planned mission of mine was headed down at about a 45deg angle to the water on an SRB. I figured Hadbel Kerman was a gonner. All of a sudden my rocket way flying under the water. I got rid of the SRB engaged the liquid engine and flew up and out of the water to safety. I must have been going about 300m/s when I hit and went down about 500m. NEAR is the only physics mod I have installed.

[Kingsix]

double sorry

Edited July 19, 2014 by Kingsix

[rynak]

No amount of SAS can save a good design short of reaching into the physics and changing them to save the player, which seems very... wrong.

Well, there is a wide spectrum in between "as realistic as possible" and "What? It's space... with drag and gravity. Done!".

The question is where does NEAR place itself? What i'll say next might offend you, but i'll explain why its pure logic afterwards: If you consider "bending physics" to be "very wrong" and out of NEAR's scope, then you are not targeting the playerbase you think you're targeting. You in fact are targeting a player who's skill, understanding and concern is almost on par with FAR!

It's not just that those other flight simulators - especially earlier ones during the 90's - star well designed and stable planes. It also is that they indeed do bend physics to let the player eventually escape any loss of control, and pay for it with time and/or altitude. After all, injecting a slight downward bias so that eventually, the plane will end up facing down and become stable again - this IS bending physics! Any kind of "bias" no matter how limited to specific situations and no matter how subtle, is messing with physics! If you don't want to have anything like this, then you're still very much on hardcore realism terrain, simply because just the most basic physical laws in a fluid, have very heavyhanded consequences - as you explained so well in your previous post, there is no escape from the mentioned effects, unless one understands almost as much, as a skilled designer and player using FAR, save perhaps mach effects and similiar.

So, pun intended, NEAR and FAR are quite close, as is. And that's fine if this is what you want it to be! But understand that in such a case, you're not targeting a "simple and lightweight aerodynamic model" - FAR from it Rather, it is more like FAR but without the hardest stuff.

Edited July 18, 2014 by rynak
typos

[Geoff-AU]

This mod sounds promising. One of the things I found really frustrating with going too far down the path of "realism mods" (FAR, DREC, RSS/RO, etc) was that it was impossible to balance the game such that it was still fun. It got harder, sure.. but DREC seemed to alternate between too easy and too hard, and the engine packs never worked nicely (the engines were the wrong size, wrong thrust, had nodes in the wrong place, wouldn't build shrouds or something else). FAR on stock kerbin made it too easy to achieve orbit, but on RSS it was bloody impossible to achieve orbit (not FAR's fault, the rockets weren't big enough.. but just saying too many mods unbalances the game).

For 0.24 I went back to STOCK, I wiped all my mods and it's awesome fun. I'm interested in trying NEAR as a standalone mod (and maybe DREC again, if I can get the parameters right) but any more than that ruins the balance of the game or makes it do really annoying things that prevent you from flying missions effectively.

[ferram4]

@Kingsix: You'll need to post a copy of your output_log.txt from KSP_Data and a full list of reproduction steps. So far, your bug report does not have enough information to determine what, if any, behavior is due to NEAR.

@rynak: No, I know exactly who I'm targeting, and I'd argue it's no more hardcore realism than the rest of the game.

This is a game where gravity is the primary thing that causes you to fail. Gravity doesn't magically become weaker when the player needs it to be to survive a bad situation.

This is a game where off-center thrust will cause things to go spinning out of control. The CoT doesn't magically relocate itself if a player screws things up.

This is a game where things can flex all over the place and cause vehicles to lose control and possibly break as a result. Parts don't magically act like they are somewhere they're not to save the player.

Why should aerodynamics be different? Or should the game be made easier in all these other fields as well? Should gravity at Kerbin's surface become weaker if you're falling uncontrollably towards it and you lack the thrust to arrest the fall? Should sideways-thrusting engines simply not matter in stability if they're making things too bad? The game already smashes bad designs and piloting in the face, but the good thing is that it's consistent. You never have to rely on the physics bending to save you.

But let's say I were to add something to force planes to aim forward. Okay, define forward. Easy for a simple plane. Easy for a rocket. What about for a crazy contraption? Once you've built the expectation that if a player intends a vehicle to be stable in what they think is forward, you have to apply it everywhere. What happens if a player suddenly changes the reference part for controlling, so that their plane is now "facing backwards?" Does the plane suddenly become unstable? Do you simply ignore that and cause one direction to be privileged above the others in terms of stability? If you don't, can't that be exploited to cause really weird things? KSP is not like other flight simulators in that the player has too much control over what is happening and any attempt to subtly save players from themselves will either cause blatant violations of common sense (followed by raging), provide easy means to do things that cannot be done in real life (Scott Manley's Tony Probe coming out of the old RCS code that made it impossible to build something that wasn't easy to control with RCS), or simply cause players to decide that the game makes no sense because, well, it's acting contrary to how they see reality.

So, how do you solve those problems?

Edited July 19, 2014 by ferram4

[Black-Talon]

As always, impressed with your work. I'm confident you've thought about this, so I'm curious to hear more about it...

What if drag was increased (artificially) across the board but this model was used?

I ask because:

1 - The stock atmosphere is so draggy, but the feel of flight is sooo wrong - it isn't the drag that's messing up the feel

2 - There is always this concern about how this becomes "cheaty" because of the lowered Delta-V requirements (is an alternative/combination option to extend the thickness of the atmosphere?)

3 - The stock engines are so overpowered for a real atmosphere that this cheaty feel is exaggerated (particularly with the jet engines)

I mean, I've gotten over it, it'd be neat (as long as we're already bending the rules of reality) if we could get real air dynamics and feel of flight without the feel of cheaty (in a single Mod - and to be honest KIDS never felt quite right to me). I realize that this still becomes a judgement call ... but a shift towards more drag, or thicker atmosphere, to compensate doesn't seem too crazy in my own head (isn't this what RSS does?). What have your thoughts been in this area?

[rynak]

Black-Talon: I think what you want is Kerbal ISP Difficulty Scaler, aka KIDS... also by ferram - which addresses exactly those issues, and more (like thrust changing with atmospheric density, instead of ISP).

[ferram4]

More drag overall would be bad, since that would make things more unstable as well; since a lot of rocket designs end up being "stable" only because thrust vectoring is sufficient to keep the rocket on course, those would suddenly become unstable. It would also make the atmospheric cutoff a lot more jarring as well. Finally, drag is just about right for reentering capsules with the current settings, and increasing drag too much would simply make returning to Kerbin damn boring. If you wanna find out, go into the NEAR directory in GameData, open up NEARAeroData.config, find the areaFactor value and set it to something above 1 and see what you think. Reentry becomes boring as hell (for which the only solution is to remove drag dependence on shape, and now we're back where we started) and rockets become less stable. Not fun at all.

I've always found the dV requirement argument to be lacking, especially in the face of the new stability requirements on rockets. That's always struck me as the argument of people who already build ideal aerodynamic vehicles who don't realize how difficult it can be for new players. It's funny, it's led to people arguing that FAR is both "cheaty" and "absurdly hard" at the same time, which leads me to the conclusion that FAR (and probably NEAR as well) are balanced, but different from the stock game. Then the judgments on difficulty make sense in the context of players with differing skills in one area or another.

The stock engines aren't really overpowered for a real atmosphere, they're about correct. The jet engines are broken though, but that's because they're really magical air-breathing rocket engines and changing those around too much really falls outside the scope of an aerodynamics mod.

[Creat]

A very simple counterargument (or rather counterexample) to rynak's reasoning why this fills niche that isn't there: This is exactly what I wanted, and it is what FAR isn't for me. There were already quite a few others in this thread with a similar sentiment.

Let me elaborate a little on that. The reasons I've never used FAR (in a serious way, apart from playing around with it a bit) are incredibly similar to the stuff listen under "What it doesn't do, that FAR does". Frighteningly so. I essentially want it to make a difference what end (if any) of the rocket I point towards where I'm going (or if I'm going sideways). I want nosecones (or aerodynamic shapes in general) to make a difference and not just add weight (and now cost, have you looked at the prices of those things!?). I do not want things to get shaky just because I'm crossing the mach boundary, and I don't want stuff to fall off because it happens to be where the shockwave of another part ends up. I know this comparison is a bit off, but in my mind I don't care about the mach boundary just like I don't care about relativistic effects when planing or executing my maneuvers.

I rarely build planes, but when I do the stock model immediately puts me off again and I go back to rockets (also in career mode, which is what mostly play these days, there's barely any point to it). In addition, I agree with all the stuff Geoff-AU touched on above, full-scale FAR (and DREC, and others) cause too many problems that were just not considered when creating the current contracts, for example. Or the TechTree. There are obviously replacements for the latter, but at least for the moment there is no contracts-mod that gives you usefull missions for specific mods (I'm sure they'll come, hopefully soon, but they are not yet and can hardly be here yet). For those reasons I also just removed DREC (for now, at least) because relatively simple things can become tedious, like weird contracts that want you to test some stuff at strange speeds and altitudes.

The only other realism-mod I'm still running at the moment is RemoteTech, and that's unfortunately another thing that goes (almost) too much against the grain of the stock design. I want to use it, my inner self almost dictates it, but setting up a (primitive, but reasonably realiable) relay-network just to get to the Mün with Science-limitations for available parts, and no Contracts that have suitable objectives is not always fun, but can be a bit of work (and not just fun).

KSP has all these awesome features, and there are all these awesome mods. I would love to use all those at once with the shiny new features we just got, but they just won't quite fit together (yet) as nice as I had hoped. It's like the toy for toddlers where you have some objects shaped like circles and squares and the corresponding holes in a box to push them though. I have all these shapes but not the perfectly fitting holes like the toy. Some can be squeezed through holes with a bit of wiggling, but it's not a real good fit, and others won't fit though any of the holes at all.

NEAR is finally a piece I can easily push though one of the holes, and it fits perfectly. I am VERY glad it exists, so thank you for creating it (or for repackaging it from your very capable FAR codebase).

[MainSailor]

@MainSailor & Renegrade: I've confirmed the issue. Seems to be related to a few changes in how PartModules are saved to work with the new contracts system, and it might take awhile to resolve the issue.

Edit: Was easier than I thought, forgot to bring some code from FAR over to fix that. Version 1.0.2 is up to fix that issue.

Awesome! I'm glad I could be of at least a little help!

[Black-Talon]

@ferram - thanks for the thoughts. I agree, I've come to terms with the "FAR (and probably NEAR as well) are balanced, but different from the stock game" and really enjoy it, but I wanted to understand more.

@rynak - thank you for the suggestion. PS - I think you read a little too quickly - I've used KIDS quite a bit ;-)

<edited to add the following> I particularly agree with your argument (that I read regularly) that if NEAR/FAR were stock, we never would've complained. The stock physics ARE hard. Man...the number of rockets I've had sideways and then into the ground...we learned to build and fly in that world and loved it. Realistic aerodynamics wouldn't make that learning curve harder...if anything it would make a lot of things more understandable (and planes feel less ridiculous - note again that no one can make a plane until they figure out the right way to use the center of lift and center of mass markers...but the forum isn't overwhelmed with rage on that).

Edited July 19, 2014 by Black-Talon

[rynak]

A very simple counterargument (or rather counterexample) to rynak's reasoning why this fills niche that isn't there

Where exactly did i claim that the niche doesn't exist? Strawman attacks and fanboys bore the heck out of me. Ferram at least had some sound and interesting arguments, like for example the question of where exactly is "forwards" (or up, or down, or any such directions). That is the stuff i'm interested in - plain honestly and logic - not if something is "the best" or other such normative popularity contests.

I understand that by someone apparently criticizing the portrayal of your favorite toy, you yourself do feel threatened.... and thus do have to defend the goodyness of it..... but, you know - bleh!

[AlternNocturn]

YES! I've been trying to do this with FAR but I gave up. I really like the improved aerodynamics but I didn't want to deal with the frustrating realism. Thanks Ferram!

EDIT: whoops, looks like I stumbled into the middle of a middle school argument. I'll just be off playing KSP then.

[rynak]

@rynak - thank you for the suggestion. PS - I think you read a little too quickly - I've used KIDS quite a bit ;-)

Well, you didn't mention so in your post, so i didn't know. I red quite carefully, and am aware of what you're describing.... but KIDS was designed for exactly this reason: In the stock game, drag is exaggerated - in turn, deltaV is cheap to compensate the high cost of reaching orbit. So, once the exaggerated cost is lessened (drag), reaching orbit and available deltav in general, becomes too "easy" - hence the "cheaty" impression. KIDS addresses exactly that. What did i miss?

[hcalves]

Well, that's interesting. Supersonic aerodynamics is more challenging for making spaceplanes but makes for strange behaviors sometimes!

[Creat]

Where exactly did i claim that the niche doesn't exist?

Here: "[...] then you are not targeting the playerbase you think you're targeting. You in fact are targeting a player who's skill, understanding and concern is almost on par with FAR!"

At least that is how I understood it, though it was apparently not what you meant. My apologies.

Strawman attacks and fanboys bore the heck out of me. Ferram at least had some sound and interesting arguments, like for example the question of where exactly is "forwards" (or up, or down, or any such directions). That is the stuff i'm interested in - plain honestly and logic - not if something is "the best" or other such normative popularity contests.

I understand that by someone apparently criticizing the portrayal of your favorite toy, you yourself do feel threatened.... and thus do have to defend the goodyness of it..... but, you know - bleh!

Despite the fact that I mention you, this post wasn't actually addressed at you (or meant as a reply to you), let alone an attack on you. I thought I made that pretty clear when I said I just wanted to elaborate a bit. It wasn't even in defense of ferram4, but it was somewhat addressed at him as a thank you for creating this (apparently favorite) toy of mine I've known about for all of a few hours. It just so happend to be your post that got me thinking why I was quite excited to see this be released - to my own surprise I might add - hence my mentioning your name. So I decided to put my thoughts it in a post, also to continue the discussion/reasoning from Geoff-AU's thoughts on realism-heavy mods, and why this fits much more comfortably into an otherwise mostly stock game than FAR or even DREC. I'm sorry if this bores you, but there's little I can do about that...

[Alshain]

Ran into a problem and now I can't fix it. I even deleted NEAR and reinstalled it. I was flying a plane just fine but well, I'm not a good lander yet. So yeah in the ground, reverted to launch, and then the plane wouldn't get off the runway. I eventually determined the control surfaces weren't working. So I went back to the SPH and saw this:

The log is full of Null Pointer Errors

https://www.dropbox.com/s/4vlczkua5g4lq4u/output_log.zip

[Black-Talon]

Well, you didn't mention so in your post, so i didn't know. I red quite carefully, and am aware of what you're describing.... but KIDS was designed for exactly this reason: In the stock game, drag is exaggerated - in turn, deltaV is cheap to compensate the high cost of reaching orbit. So, once the exaggerated cost is lessened (drag), reaching orbit and available deltav in general, becomes too "easy" - hence the "cheaty" impression. KIDS addresses exactly that. What did i miss?

You seem upset - I meant no ill will. I did mention in my post. Sorry.

[ferram4]

Are you using NEAR v1.0.2? That version was pushed to fix that error.

[Alshain]

Are you using NEAR v1.0.2? That version was pushed to fix that error.

Ah, you are updating faster than I am keeping up. Good job

EDIT: Well I delete the old folder, installed 1.0.2 from KerbalStuff and it crashed the moment I loaded the plane in the SPH (entirely stock craft but I have other mods installed). I can't tell if this crash was actually the fault of NEAR.

https://www.dropbox.com/s/cavn76g9yrcsflb/output_log_2.zip

EDIT2: Ok, maybe the bug from the previous version found it's way into the craft file. Fortunately it was one I copied over from 23.5 so after re-copying the crash stopped happening. But I'm still having the original problem. Not enough lift, control surface tweakables are messed up.

Edited July 19, 2014 by Alshain