Fixed Still Broken Flight Mechanics

Version: 11.0 2705
Mode: All
Mode: All

SERVER NAME:
SEED-ID:

MODIFIED PLAYFIELDS: No
Reproducibility: Always
Severity: Major

Type: Flight Controls
Summary: Flight mechanics are broken.

Description: New flight mechanics are still broken. Top speed calculation for vessels not working correctly.
Vessels that turn left or right but not both, vessels that turn up or down but not both, and vessels that roll isn't working correctly.

This is happening with all vessel classes from time to time. Vessel can built 100% symmetrical or not and it still happens. Sometimes it happens with one blueprint one time but not the next. Sometimes it will turn left but not right and the next time you spawn the same blueprint it will turn right but not left. Etc Etc.

Steps to Reproduce: Spawn the attached blueprint and it happens every single time. Again, this is happening with all vessel classes and not just SV's. It's happening with all kinds of blueprints but not all of them are reliably reproduced.

1. Spawn the blueprint.
2. Get in and hover off the ground.
3. One at a time using keyboard only, press various movement buttons one at a time and observe how it behaves.
Watch how it behaves and how it's opposite behaves (left arrow key vs right arrow key, up arrow key vs down arrow key, Q vs E)

Please also take another look or 100 at how your top speed is being calculated for vessels.
Covering a giant block with 1,000 spikes shouldn't make it go faster, and in fact would cause more area for drag to be caused than a giant block would.
A 10x10x10 cube vessel should have less drag than a 10x10x10 vessel covered in spikes on the front simply because the area between each spike/pyramid on the front would act like a mini parachute of sorts creating drag.
The way you are reducing top speed based on some arbitrary cross section look and some magical minimum acceleration is ruining enjoyment in the game. It makes no sense from a gameplay perspective in how it's implemented.
Read your feedback threads to find examples of this. People have been telling you this for over 2 weeks now.

 

It's complicated, though, and very likely speed-dependent (the turbulence created by the spikes will evolve differently depending on the ship's velocity relative to the fluid). For Empyrion's purposes, the two cases should probably just be treated the same, since the ship has the same overall size and shape.

Worth mentioning that max speed is also a function of mass right now, when it should *not* be.

 

Here's two more blueprints. The Test2 one is a bit more substantial in it's size. Each one uses a slightly different thruster location as well.

I have witnessed this behavior of turning in one direction in even larger vessels as well but it's not always consistent. One time I spawn it it functions correctly and the next time it doesn't. I can't find any T4 CPU vessels where I can reproduce it 100% reliably in order to provide you the blueprint, but it DOES happen none the less. It's not just micro vessels affected.

 

To clarify, it's not an issue with certain shapes at all, it's entirely due to the local nature of the drag coefficient calculation. This can't be solved just by tweaking the coefficients for certain blocks, it's a fundamental issue with the algorithm itself.

 

The problem is the way it's being calculated. This is a problem the developers need to fix. I'm not going to be forced into only placing thrusters in very exact spots.
We were specifically told that these features work no matter where the thrusters are placed, but that depending on where they are placed they will have a bigger or smaller impact on turning. This is simply not true based on what you are now saying.

No, I'm not going to add 8 thrusters. The whole point is to make a starter craft for scenarios where resources are scarce. Your suggestions goes against the whole point of the craft.
Besides that I've already experienced this with more than 6 thrusters as well. This is a bug plain and simple and needs to be fixed by the developers to where it always works.

If they aren't going to resolve this then just mark it as not a bug and we'll let the Steam reviews speak to the devs when this gets pushed to the public.
I'm sorry, but the devs need to start taking this stuff seriously.

It's not that I want to use these exact blueprints at all. The issue here is that these blueprints clearly show that the feature isn't working correctly and it NEEDS to always work correctly no matter what. If you and the devs can't see that issue then we have bigger problems.

Edit
They need to remember something very important here,
Until a player finds promethium they don't have the luxury of picking a thruster up and placing it down somewhere else. They are going to lose 66.666+% of that thruster if they need to move it.

 

Did you check any of the blueprints I provided?
The Test2 blueprint has thrusters placed many blocks away from the center of mass and it still won't function at all in certain directions.

This isn't about the ones you have that do work, this is about the very real blueprints I provided that don't function at all when they should.
I can create 10/20/10000 more blueprints if you want. The issue is still the same. They don't work.

You aren't addressing the fact that the ship is built perfectly symmetrical and yet it functions one direction but not the opposite. How can it turn left but not right when it's built perfectly symmetrical? How can it turn down but not up when it's perfectly symmetrical? It's bugged.

 

So you tell me I'm just building too small, so I'll build much bigger.
So you tell me I'm just not using enough thrusters, so I'll add a lot more.
So you tell me I'm placing thrusters too close to the center of mass. Not anymore.

Yes, this blueprint I'm linking is extremely impractical in every sense. None the less it's exactly the type of stuff a new player is likely to build when they are learning the game (just scaled way up in size). It's built symmetrically once again, which is what a new player is most likely going to do.

It is capable of rolling to the left, but can't roll to the right.
Hold on, let me use "O" to level off and check that a few more times......Now it will roll to the right but can't roll to the left.
Checks a few more times......Now it will roll to the left again but not the right.

How about turning up and down. Nope, I level off and hold up arrow for 30 seconds and it doesn't budge an inch.
Down arrow? Same exact thing.
Now let me tap one of them instead of holding it. It moves a fraction of an inch every time. Still does nothing when holding the button.

How are you still not seeing how bugged this is????........
I'm doing what you said and I'm moving thrusters to the outer edges so they have the highest impact on turning.

Here's a screenshot. For size reference my other blueprints are parked on top of it.


No, I don't have any intention of using these blueprints at all. These are specifically for the purpose of showing you the bugs in the system.
I'm not about to spend 30 hours designing a nice ship only to have it be incapable of turning one direction but not the other. I'll quit the game forever after an experience like that, as would most players.

 

Sorry, I don't really get what you want to achieve here. I tested your ship (Test2) and Test3 doesn't look any different, just larger. Your thrusters are placed asymmetrical (back left facing up, back right facing down). Thruster-only torque doesn't work that way and of course shouldn't.

If you want to use thruster-only torque:
You need of course up- AND downward facing thrusters on EACH side to make it roll in both directions. And also left AND right thruster BOTH in front and in the end of the ship to get good yaw in both directions. Isn't that common sense?

If this design would roll in both directions, then there would be a massive bug. You can of course still work with only six thrusters as in 10.5, but then you will need a RCS again.

You could even make your test3 design a non-VTOL ship (just with drag & lift) by just placing a single thruster facing back (in the middle) and one RCS.

/jmc

 

I just loaded up the Test1 blueprint, and I can confirm that, indeed, the Yaw torques are borked. Despite the placement of the left/right thrusters, the ship rotates quickly to the right, but not to the left. The statistics screen confirms this, as the ship thinks it has 29 Nm (also, so small . . .) Torque+ and 1 Nm Torque-. The roll torque seems to be ok (it indicates that the ship can only roll in one direction, which is true). The pitch torque also seems to be wrong, as it shows only +2 and -1 Nm, when it should be +/- ~10 Nm.

On a hunch, I built a perfectly symmetric ship, with the fuel tank and generator inline. In this configuration, the CoM is precisely centered left/right. And, the torque generated by symmetric left/right thrusters is indeed the same. . . .

Ok, here's something interesting:

If the thrusters, core, and cockpit baseblock aren't all on the level, the yaw torque is greatly reduced. In the first image below, all are on the same level, and the yaw torque is +/-40 Nm. In the next image, the cockpit has been moved up a block, and the yaw torque suddenly drops to +/-2 Nm. This happens as well with a similarly-built CV (shift the cockpit up, and the torque drops). So, I don't know how the thruster torque calculation is being done, but it's clearly including factors that it shouldn't, as all it should care about are the locations and directions of thrusters relative to the center of mass.


This is why it'd be really great if you'd tell us exactly what physics equations you think your code is implementing, so that we can help check to see if they're actually being done properly, and if not, why not. Because all we can tell you right now is that the calculation is wrong.

EDIT: There's some kind of symmetry issue here, as, for the CV case, if I place another cockpit above the inline cockpit, the problem occurs, but if I also place a another cockpit below the inline cockpit, the problem goes away.

 

Attached are 3 CV blueprints that illustrate the problem (or, at least, the part of the problem that I've been able to fully isolate). The first one ("inline") has the cockpit inline, and, with just the front side-facing thrusters active (turn off all the other ones when testing; they're just there to keep the CoM in the same place), and has +/- 1.49 kNm torque. The second one ("1-above") adds a second cockpit above the first, and the torque goes down to +/-186 Nm. The third one ("symmetric") has a cockpit both above and below the inline cockpit, and the torque is +/-1.48 kNm (slightly less than in the first case due to the extra mass of the cockpits, of course).

 

Hmm, strange I can't really confirm that. I'm not saying there's no bug, just trying to help finding the underlying issue, if there is one.

I tested around again now with test2 (but added both up and down to left and right to make it roll in both directions), but placing the yaw thrusters on pillars or into the baseplate (as they were originally) does NOT make any difference as far as I can tell (+-1 deg/s). Roll on the other hand is different, which is to be expected, when putting thrusters on pillars.





But If there is a difference in some builds, this sure would mean some strange behaviour at least that needs being looked into.

/jmc

 

But, did you move the core as well? The fact that the core and cockpit are on different planes is also part of the problem there.

 

Also, just as a point of interest, only 6 thrusters are needed in order to have full rotation and translation for a ship, if they are place appropriately. Having 12 symmetrically-placed thrusters improves things a lot, but is not strictly necessary.

 

Ok, did another test with moving the core.

Moving the core HAS an effect indeed (moved the core up on a pillar 3 blocks and yaw was down to 63.

Then I did a clearpivot on the vessel and it changed again (55 deg/s²). So the torque calculation is somehow based on the pivot point (whatever that is exactly)

/jmc

 

Ok did clearpivot again and I'm almost back to normal (65 deg/s²). However, this difference could also be due to deviations from the mass and COM change and the shape of the pillars the core is resting on in high position (as it was under the cockpit and I added two pillars left and right of the cockpit).



Although there's something a little strange going on, I can't really reproduce this reliably. My guess: It has something to do with the center of mass (pivot) calculation which does not always reset itself, when placing blocks.

/jmc