|
Post by L5Resident on Sept 28, 2016 23:34:16 GMT
|
|
|
Post by blothorn on Sept 29, 2016 1:43:09 GMT
Try putting it in a low-angle turret; that should allow you to distinguish between target prediction errors and maneuver errors.
|
|
|
Post by the_Demongod on Aug 7, 2017 18:49:28 GMT
I'm going to resurrect this thread because the issue is still there. As far as I can tell, the issue stems from a problem with the broadside command.
Essentially, broadside is trying to put the ship into a fixed orientation which allows the weapons to aim at their targets. This is great for gimballed/turreted weapons because this minimizes the footwork the targeting systems have to do in terms of calculating a firing solution, especially with a moving target. However, this is extremely suboptimal for fixed mount weapons for anything but a perfectly fixed target. The ship aims the gun at the target and begins to fire, but as soon as the target moves out of the acceptable error range of the solution, the ship has to recalculate and reorient, and stops firing during this process. Once it has aimed back towards the target, it fires again, and the cycle repeats. The higher the angular velocity of the target, the less time it will spend in the acceptable firing solution before our ship has to reorient.
This contrasts sharply with the turrets, which move smoothly while firing, enabling fire as accurate as the weapon will allow. While an accelerating target does confuse turrets sometimes, they can handle non-accelerating moving targets just fine, but spinal weapons can't. I couldn't even hit a space station with low relative velocity with my gun, despite having a ring of 2MN RP-1 resistojet RCS thrusters for aiming, with considerable control authority (at least considerable for a capital ship as large as mine). I suspect that drones aren't as hindered by this problem because their weapons are far less accurate, engage at much closer ranges, and have very high directional authority and can turn fast enough to negate this problem.
As far as I can tell, the best solution is simply for the broadside algorithm to detect the presence of spinal or fixed-mount weapons and employ a targeting routine more similar to the turrets (where the entire ship will maintain a calculated rotational motion to continuously track the target) instead of the regular broadside (which kills rotation once aligned).
I hope this can be fixed at some point, it cripples ships that use internal mounts to protect their weapons; they are already penalized by being limited by the large rotational inertia of capital vessels and the added mass required to effectively aim such a large ship.
|
|
|
Post by AdmiralObvious on Aug 7, 2017 19:16:28 GMT
In addition to the above, nose forwards commands barely help either.
|
|
|
Post by jtyotjotjipaefvj on Aug 7, 2017 20:17:30 GMT
I can verify that this happens. It looks like spinal mounted weapons don't take relative velocity of the firing ship properly into account. If tangential velocity is 0, they do hit pretty well.
|
|
|
Post by the_Demongod on Aug 7, 2017 20:38:02 GMT
In addition to the above, nose forwards commands barely help either. I'm guessing this is because the point of aim is off the target if they're moving at all, where aiming directly at the target is only going to have your rounds land where the target was x minutes ago (where x is projectile flight time to target), so the gun just doesn't fire since it doesn't have a valid firing solution. I can verify that this happens. It looks like spinal mounted weapons don't take relative velocity of the firing ship properly into account. If tangential velocity is 0, they do hit pretty well. I was shooting a station with very very low relative tangential velocity and it was still struggling to hit. It would stay on target for a few seconds and then slowly deviate until the rounds started missing, then it would stop, readjust, and try again. This is only at 700 or so kilometers where the turret-mounted variant of the gun will be incredibly accurate and powerful against targets as small as drones.
|
|
|
Post by Rocket Witch on Aug 8, 2017 0:43:58 GMT
Simply implementing a fudge factor of 1° where the gun can still fire slightly off from where it is actually aimed would make the problem go away immediately. Given the space inside ship noses, there would no doubt be a toroidal tank qswitched-sama pls an internal gimbal mechanism where the muzzle is hinged at the armour join (or ideally just behind it with the armour overlapping) and the breech-end moves around freely to tilt the weapon. This could be explicitly implemented as an extra part of module design, but the simple fudge factor solution would satisfy me.
If you watch the trailer for the game you'll see Stinger drones turning and immediately beginning to fire again as they pass their target. This feels like it has progressively gotten worse with updates, the finishing blow being the update that brought actual dodging and aim prediction.
|
|
|
Post by the_Demongod on Aug 8, 2017 19:52:57 GMT
I had the same idea when I first discovered the issue and it seems like a design that would be implemented in real life for added precision while maintaining the defensive bonus of an internal mount.
I still think it's reasonable to expect my original request to be implemented though, spinal weapon mounts are something that are talked about all over Atomic Rockets and is a no-brainer method of adding significant protection to your weapons. All it would take on QSwitched's end would be to take the logic gimballed weapons use to orient themselves, and allow the vessels themselves to use it. I'm a programmer so I know such a task is not necessarily trivial, but shouldn't be too difficult to implement.
While it's difficult to aim an object as large as a capital ship accurately, the large inertia makes the relatively weak RCS thrusters highly precise (especially thrusters with as easily variable throttle as a resistojet) so I'm sure the ship could handle aiming itself accurately just fine.
|
|
friedriceistoolazytologinfromm
Guest
|
Post by friedriceistoolazytologinfromm on Aug 12, 2017 0:01:23 GMT
qswitched is of course the final arbiter of things, but wouldn't a fuzzy targeting percentage be a pretty straightforward way of doing things? At extreme ranges, even a tiny amount of deviation means a pretty big shot group.
might be easier to code, might be harder. No clue.
|
|