Gaming MotorWings! A KSP + Crimson Skies Remake, Build your own plane FlightSim (with realistic physics)

[Moach]

Latest stuff:

---------- original post: ----------




Well, I've waited long enough... Boy, now I know how and why my brother felt nervous when he first unveiled KSP. I guess this is my turn:


Hello, world! This is MotorWings!


UnEdited Screenshot:

I found that in the right circles, the whole idea can be described best and simply as: KSP + Crimson Skies

Which is basically what the game is supposed to be. You create your own plane, and there's a wide open world (kinda alt-reality, a bit dieselpunk style) where you can go around doing basically what you'd expect Han Solo or Nathan Zachary to do with a patchwork airplane of arguably sensible construction.

This is a very early alpha release - actually, this is the first time I'm telling the internets about it. Had to start somewhere, so to keep in spirit with the tradition my brother Harvester started: Orbiter forum gets the first post!



Let me know if it works out alright for you and be gentle about any nastyness it may or not do from undiscovered bugs.


Oh well, that's pretty much it I guess - Enjoy!


May your airplanes always land on the wheels first!

 

[dbeachy1]

Nice work!! This will be fun to follow.

 

[Urwumpe]

You know Crimson Skies? We had a nice small group doing a lot of Crimson Skies at our last "Phoenix Con" BattleTech convention here. :thumbup:

 

[Moach]

New version up with a few tweaks and bug fixes!

https://www.motorwings.net/start-flying

 

[Fabri91]

I'll be giving it a try this evening!

 

[Face]

Hy Moach,

first of all: great work, great idea! Thanks for letting our community here get a first look of what might become another KSP-like hit.

I've tried out the available alpha, and here is what I think of it:

• The alpha state is really very early. In contrast to what was possible with KSP when it was first announced here, there is not much you can do with it. I get the overall concept to be like Crimson Skies, but well, I don't see anything new in this first cut besides yet another flight simulator, this time with the KSP construction concept bolted on. I'd love to see more of the "get your crew, salvage/trade/pirate" concept to get a better impression.
• The controls are pretty rough. I'm familiar with the KSP concept of constructing and controlling flight vehicles, but this here is even worse. The flight characteristics feel wrong to me. The missing crash simulation makes the experience a bit stale.
• An interesting part of Crimson Skies is weapons, right? I know it's early on, but it would help to sketch out the idea to have them already, perhaps with some dumb targets to shoot at. AI would be perfect for this, but of course this is probably waaayyy out for this stage.
• The stuttering is there, but not really as bad as I thought it could be on my outdated laptop with 7. Overall the world rendering and performance already looks good to me. Of course finer details are missing, but this is not so important at this stage IMHO.
• I had no crash or bug or problems to get it running. Just unzipped, started, immediately knew what to do. Very stable already, great work!

Keep it up!

 

[Urwumpe]

Just looked at the planned gameplay features - does "make your own crew" mean something like the storyline parts of the old Strike Commander game? I really liked this aspect there, it made the whole game far less technical and more human oriented without actually doing much... it was the early 1990s, after all.

Will also give it a try when my official 16 hours of allowed work for this week are over.

---------- Post added at 13:29 ---------- Previous post was at 12:13 ----------

OK, the multi-engine planes are very hard to get off the ground, this should be a bit more stable later. But the single engine example is already nice to fly, feels pretty good. Stall speed and take-off speed are a bit high. While I got the tailwheel off the ground at about 80 mph, take off below 150 mph is hard. Engine power seems to drop rapidly with altitude, barely managed to get above 600 m for a dive test. An indication of sideslip would be nice, I often barely noticed that I was already flying at about 40° sideslip after a maneuver.

The terrain freezes are annoying, they happen pretty often and take a while. This might be better get improved.

While I was able to fly through the large hangar roof, barn storming is possible. Its possible to collide with the airships, and I even managed to touch down on one, though too fast to land. I doubt this should be a standard use case.

---------- Post added at 14:39 ---------- Previous post was at 13:29 ----------

OK, while ground handling with multi-engine aircraft is poor and slightly erratic (while idling, it either turns to the left or to the right, but never has a constant tendency to yaw), it is possible with a lot of patience. Using T a lot helps, just like wheelbrakes to keep the roll speed low while taxiing.

With more power, its also possible to climb to 1500 meters in reasonable time. Pulling out of a dive at over 450 km/h was no issue. Doing a tight looping around an airship is possible.

Landing the second example aircraft works, if you take a lot of time for lining up and reducing speed. A really shallow approach is recommended.

The terrain freezes are much more annoying at low altitude. I get about 3 seconds here each time on mobile i7 here. Maybe using smaller blocks of terrain could shorten the time of the freeze, but make them happen more often.

Hardware setup:

Operating System: Windows 10 Home 64-bit (10.0, Build 18363) (18362.19h1_release.190318-1202)
Language: German (Regional Setting: German)
System Manufacturer: SchenkerTechnologiesGmbH
System Model: XMG CORE (M19, GTX 1660 Ti)
BIOS: N.1.04 (type: UEFI)
Processor: Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz (12 CPUs), ~2.6GHz
Memory: 16384MB RAM
Available OS Memory: 16240MB RAM
Page File: 7020MB used, 11652MB available
DirectX Version: DirectX 12

 

[4throck]

Gave it a try. Loaded a plane, pressed "Fly" and it freezes. After a few seconds it crashes to the desktop. No error report or any message. :hmm:

 

[Moach]

Gave it a try. Loaded a plane, pressed "Fly" and it freezes. After a few seconds it crashes to the desktop. No error report or any message. :hmm:

How very strange... Have you tried the "NT" version as well?

The terrain stuttering is a very well known problem - It will be addressed shortly as soon as I can migrate to the newer 2.0 version of the map generator plugin.

This newer version is supposed to have been more or less completely rebuilt, such that it would allow us to have far more flexibility with terrain tiles and resolutions (1.0 requires all tiles to have the same resolution, regardless of distance, which greatly limits visibility range and performance)
I expect to be able to completely remove the stuttering, or at least lessen it enough that it's barely noticeable after upgrading to the new system.



I'm constantly working on the physics settings (there are MANY variables, some with obscure side-effects) in order to find that perfect balance where PhysX is happy and flight feels like flying. Alas, it's not viable to go about it in a purely mathematical way - PhysX has certain "preferences" about orders of magnitude and is seriously limited by using 32 bit floats internally. We do have a floating origin system that helps cope with the "float noise" problem quite effectively. But if I were to use real-world units such as KGs for part masses, the thing goes completely nuts and feeds us to the Kraken.

This is something of an endless task, anyways - And every time I change something in the simulation itself, a lot of tweaking is necessary to make sure nothing spazzes out in return. It's a strangely inexact science though, and each time I test fly it I end up changing a few things here and there. It's come quite a long way - you wouldn't believe how weird it can get just by having the setup "slightly off"


Still I'm generally unsatisfied with the overall drag model (drag can be a bitch to simulate properly when you have a plane made of unknowably assembled parts, KSP also had lots of trouble with it) - Right now there is far too much guesswork involved in the part authoring end. This is what I suppose causes some strangeness to how planes handle.

I reckon I'll soon overhaul the whole thing and have it go more systematically about it... Lift is rather a lot easier to approximate satisfactorily, but part-based drag is definitely not a trivial thing to get working right



There are indeed some oddities around how planes handle on the ground - sometimes the same aircraft can be perfectly stable, then on the next flight it starts bouncing up and down like a caffeine-happy bunny....

That's PhysX for ya, it works when it works, but it has more moods and opinions than my ex wife



....

anyways, the "make your crew" feature is planned to have a TheSims-like customization tool for your "captain", and then you'd be able to hire more crewmen to fly as gunners, copilots, engineers, or as wingmen, if the crew owns more than one airplane.

These new characters would be made out of a random combination of the customization options, and will be named by a name generator similar to the one in KSP. You'd find them by visiting "Saloons" and similar watering holes at various locations around the world - Each character would have his/her own skills and specializations, and getting them to join might be as straightforward as paying up front or could require negotiating shares of your expected plunder

This whole aspect of the game is something I tend to think of as the "Nathan Zachary / Han Solo Simulator" - Except that your crew would be something you put together yourself as you go about your adventures. Crew members might come and go, or simply get dead if you go for that type of flying.


I also can't wait for this part of the game to be implemented - but there's a certain order to these things, and logic demands that the flight simulation sandbox core must be up before we can build this kinda stuff around it.



I'm gonna fire up the build here and see if I can reproduce some of these issues. The ground handling one is particularly elusive - sometimes its there, sometimes it's not....

We'll get to the bottom of it eventually...

 

[Urwumpe]

I'm gonna fire up the build here and see if I can reproduce some of these issues. The ground handling one is particularly elusive - sometimes its there, sometimes it's not....

We'll get to the bottom of it eventually...

If you need any test flights just say so. :lol:


So far I can tell, that the behavior can even change during a 360° turn already. While pointing south, the aircraft turned port, so I kept a port turn, then, it suddenly decided to prefer turning starboard...



Almost as if the aircraft is turning on a slope - without there being any slope.



Maybe some odd feedback somewhere with parts of the aircraft being too elastic?



Again, generally, the behavior of the aircraft especially at speeds above 100 km/h is good, I can't complain, even on the ground, with just two wheels. Running out to a full stop does also not cause problems. But then taxiing back to take off makes everything go mad again, as if being considered airborne resets something.

 

[Moach]

Ok, so far I've discovered that not all planes have issues handling on the ground

The RaptorHawk seems most bugged for whatever reason, yet the Learoch FN-100 and the SwiftEdge sat down nice and pretty and gave me no trouble at all on takeoff or taxi...

The development version I got here has already been tweaked enough that the handling feels quite different - even though I made that build only yesterday....

I'm gonna keep making new builds as frequently as I can stop to make sure nothing's badly broken - For now, the main advantage of being in such an early stage is that builds are relatively quick and simple to make

---------- Post added at 13:47 ---------- Previous post was at 13:37 ----------

If you need any test flights just say so. :lol:
So far I can tell, that the behavior can even change during a 360° turn already. While pointing south, the aircraft turned port, so I kept a port turn, then, it suddenly decided to prefer turning starboard...

Almost as if the aircraft is turning on a slope - without there being any slope.

Maybe some odd feedback somewhere with parts of the aircraft being too elastic?

This "pull to one side" preference is something I have noticed, and it does seem to be mostly random as to what side it wants to go and how hard it pulls...

I suspect it's something related to the configuration of the physics joints that hold the plane together - there's a tolerance-like parameter to them which has to be set just right, and in various conditions it causes pulling one way or another...

Another guess would be that perhaps some of the parts have been made with one or more connectors slightly off center - I've double checked this already, but that kinda thing is sneaky enough to slip by as soon as I'm not watching.

I've found that whenever a plane starts pulling like that, if I go back to the hangar, pull out the wings and/or adjacent parts then attach them again, the pulling tends to stop (or rarely, starts up in the opposite direction)

Again, that's PhysX, doing it's "thing"... whatever that might be....

That's the kinda thing that makes me wanna make my own physics engine from scratch - again - I'd rather avoid reinventing that big a wheel this time around.

I recall that at one point a while back I had this mostly sorted out - but then I changed some things and gradually it started happening again...
There's definitely some setting that makes this more or less apparent. One of these days I might even find out what it is

 

[4throck]

Gave it another try. The problem was just a long loading time.
After about 20 seconds the plane appeared on the runway.

Once it loaded, it paused around 5 seconds each time as I moved the camera around.
After about one minute it was smooth and I managed to get flying.

I went towards the zeppelins (with a distant few trees visible) and it stopped again for about 10 seconds. It then resumed with some more pauses about a second apart.

When it's running it's rock steady at 60fps, but those loading pauses are really odd.
Hope this information helps!

My specs:

- Intel(R) Core(TM) i5-7400 CPU @ 3.00GHz
- NVIDIA GeForce GTX 1050 Ti
- 16 GB RAM
- Windows 10

 

[Urwumpe]

That's the kinda thing that makes me wanna make my own physics engine from scratch - again - I'd rather avoid reinventing that big a wheel this time around.

I also think, it is better for all of us (including the test players), to keep things as fast and simple as possible. Maybe it would be better as approach to change the inputs to PhysX and use a different input model, than to reinvent the wheel completely.



When the model that is simulated is so mind-gobbling complex, that even the developers start resorting to praying before starting the engine, it might be better to cut down complexity or calculate many parameters in advance.


I think nobody expects a second DCS here. For me, some simplified flying physics and many opportunities to customize the aircraft would be more interesting. Like adding Methanol-water-injection to the engines of the aircraft for a performance boost.



And of course adding lots of non-flight-model changing stuff into MY cockpit... like a pot of flowers or a art deco window in my back. :thumbup:

(Sadly the world building does not permit me to even dream of having a Dancing Elvis figure on the dashboard.. maybe something else.)

---------- Post added at 19:50 ---------- Previous post was at 19:12 ----------

No big ground handling issues with the Cloudrider now. Did two take-offs and landing, the first landing had too much bank and I spinned to a rapid stop after a lot of fighting, but otherwise it was OK (Not unexpected result after touching down with about 15° bank). taxiing to a new takeoff was a slight fight because it liked to keep the turn rate longer than I liked... lack of nosewheel steering?

Second takeoff was also fine, second landing was perfectly level and VERY smooth at 180 km/h. No ground handling issues there, braked to a stop without yawing.

---------- Post added at 20:10 ---------- Previous post was at 19:50 ----------

All canard aircraft seem to react very poor to the terrain generation freezes.

 

[Moach]

....Like adding Methanol-water-injection to the engines of the aircraft for a performance boost.

And of course adding lots of non-flight-model changing stuff into MY cockpit... like a pot of flowers or a art deco window in my back. :thumbup:

Naturally, all of this is surely planned! :thumbup:

I wanna have engines installed as internal parts that go inside the cowling (as long as they fit, based on block shape, such as: radials require radial-type cowlings, and so and such)

Then these would have system subcomponents that can be replaced, like a standard block-mounted oil cooler could be replaced with a separate more advanced ducted unit which exploits the Meredith Effect - Then you'd have to route piping for these.
Or carburators could be swapped for direct injection systems, and turbochargers could be attached and all these things... There's a whole world of possibility there

With the career mode, each of these components would also be subject to persistent wear and tear, and the more advanced types would only be available in specific parts of the world where they are made.

I got a whole load planned for that - cockpit editors and all


Of course, I'm not going for DCS-level complexity -- I find that much detail actually detracts from gameplay in many cases. The key is to find that balance where each detail feature actually has some impact on the game. Simulating a detailed startup procedure only so that people can learn once to never do it wrong again adds nothing to the experience, besides raising the learning curve - Such a procedure could just as well be just animated (as in IL2:GB) so that all steps are realistically followed without actually having to simulate all the "what if" conditions that pilots would invariably learn to always avoid.

On the other hand, it'd make sense to simulate detail such as oil temperature effects on oil pressures; As that is something that can lead to cumulative damage to the engine if proper warm-up is not observed. And with persistent part reliability on career play, that "detail" makes some difference as to how thing could eventually turn out for an incautious pilot


So that's my guideline for simulation depth here: I'll go for as much detail as is actually relevant to some aspect of the game. Complexity only for the sake of itself is generally undesirable.

---------- Post added at 18:24 ---------- Previous post was at 18:10 ----------

lack of nosewheel steering?

Nosewheels can unlock and caster freely, same as tailwheels - they do not actively steer though. It works like the landing gear on the Me262, the B-25, or the Lancair Legacy (they all use differential brakes to turn, despite having tricycle gears)

---------- Post added at 18:32 ---------- Previous post was at 18:24 ----------

When it's running it's rock steady at 60fps, but those loading pauses are really odd.

Yes, this is a fault of the soon-to-be-replaced terrain generator system. This is a third-party package for Unity called "Map Magic" - It is currently being superseded with it's 2.0 version. Once I manage to migrate to that, the stutters problem should be more readily dealt with.

If you find you're having too much lag so that the game isn't even playable, you can try the "NT" build.

This has the terrain auto-generator disabled and runs smooth as silk all the time. Unfortunately, it also has only a limited area around the home airfield to fly around. But it is pretty much lag-free.

 

[Urwumpe]

Simulating a detailed startup procedure only so that people can learn once to never do it wrong again adds nothing to the experience, besides raising the learning curve - Such a procedure could just as well be just animated (as in IL2:GB) so that all steps are realistically followed without actually having to simulate all the "what if" conditions that pilots would invariably learn to always avoid.

Well depends.

1. if the sequence has no effect at all beyond starting the engines, it is well enough to skip it.
2. If the sequence is needed for initializing highly customized things, it makes sense and should not be repetive - like setting up navigation gear or fire control systems.
3. If it makes sense to discover maintenance problems before taking off, it is useful.

I think the points 2 & 3 could apply here, but I don't think it should be a 10 minute sequence. Something not more complex than starting a real Fw-190 or P-51D should be enough, and even there, the actual engine start could still be mostly automated, since there is little human decision needed.



There is also a weak point 4 possible - to avoid having to write code to handle too many special cases.



Again, I don't think it would make any sense to put a DCS level of detail on system modelling. Especially since DCS already exists. But an open world is planned, so the software can't know everything that goes on in the mind of the player. And it might not be easy to know every aircraft configuration in advance in this case. Why not leave this to the player to handle it?



Il2 had the big advantage, that it never was an open world game and the game play was divided in missions. The aircraft was pretty much always in perfect shape when you started it. The concept for this one sounds like this will not always be the case and in some situation, the player might even need to take off with a broken component, compensating its loss.

 

[Moach]

Well, yes - You get the point pretty much exactly, I think.

Any level of complexity that has no purpose other itself is what I'd consider a candidate for "abstracting away".

If there's any consequence that could result from doing something or not, and that is something that causes effects that make the game more interesting, then by all means, it should be there.



I've seen what happens when developers err on both sides of this fine balance. The clever ones usually listen to players and end up getting it right.

I don't expect myself to get away without blundering either way various times over this kinda thing, but hey, that's just part of the job.

We mess it up, get yelled at, fix it, then people get happy and find something else to yell about. Then the process repeats until the project is "done", (if there can ever be such a thing) or somebody goes insane and gets put in an institution with a nice comfy padded room...


Yet a lot of what comes across as "simple" in game design, is most often just the visible end result of "unthinkable arcane confustication packed tightly into an interface of such genius device that players think it wasn't any trouble at all to create"

Interface design is the key to making the unfathomable appear obvious.

 

[Urwumpe]

We mess it up, get yelled at, fix it, then people get happy and find something else to yell about. Then the process repeats until the project is "done", (if there can ever be such a thing) or somebody goes insane and gets put in an institution with a nice comfy padded room...

Well, from my experience, there are luckily also the projects, where nobody yells at all, all understand how things go on... and still, you can measure the velocity of the development team in curses per minute, just to prevent the latter result. :lol:


And yes, people are rarely aware that something that is easy for a human brain is hard for a CPU. Still, I prefer the games that can explain their rules: When I press button X in situation Y, the following will happen...


Its more interesting to put the complexity into storytelling, especially for open-world games, which quickly feel empty, if nobody interacts with the player.

 

[4throck]

Tried the NT version. On the first run it still takes 20 seconds for the plane to appear on the runway, but after that it's smooth.

Still it's a long loading time for a 310MB game... Specially compared to online stuff that loads in 3 seconds on my machine: https://www.geo-fs.com/geofs.php



Here's my gameplay feedback:

» The control system is a bit awkward.
It's odd to press the left/right arrows and A/D at the same time
I guess most players won't even understand the reason for that.
So I suggest having "auto coordination" on by default (check "World of Airplanes" for a nice arcade control).

» Free camera makes flying confusing. Camera should default to a top back view of the plane (again, check "World of Airplanes").

» You also need an attitude indicator on the HUD.



I think you need to add some gameplay fast.
Some simple goals to start like "fly straight for 1 mile; climb to 10000 feet".

I get the build your airplane angle, but that's not enough.
And whatever you build should be toward whatever goal gameplay dictates.
Ex: if you have a cargo mission, you need to build a cargo plane.



Just my 2 cents, hope it helps!

 

[Moach]

Tried the NT version. On the first run it still takes 20 seconds for the plane to appear on the runway, but after that it's smooth.

Still it's a long loading time for a 310MB game... Specially compared to online stuff that loads in 3 seconds on my machine: https://www.geo-fs.com/geofs.php

(...)

I think you need to add some gameplay fast.

I think the long startup time was Unity compiling up shaders for you, since it was the first time you ran that build. It should load up faster on future runs, I think.

At least for me it loads up considerably faster than GeoFS - Though it does feel clunky occasionally when shaders need to be recompiled. But after that it gets pretty fast again.


Hard to say - the PC I got here has a crummy old Core i5 5200 CPU, so anything that needs heavy processing takes a while - Yet it curiously also has a dual SSD Raid0 setup, so that it boots from zero to desktop in less than 20 seconds.... it's a strange frankenputer, I know. I made it mostly out of spare parts. (kinda like the planes we make in MotorWings, though perhaps the lack of textures doesn't convey the "2nd hand metal" look properly)

 

[4throck]

I have a SSD for the OS, but the game is on a standard HDD
It's not a fast drive. Kerbal takes about 2 minutes to load, Orbiter 2016 with terrain about 45 seconds. But those have gigabytes to load...

My suggestion is putting a loading message, perhaps cycling through some screens illustrating gameplay (see World of Tanks while you wait for other players).

:thumbup: