Profile image

Avalon Industries AZ Experimental Assault Vehicle ''Cerberus'' (fixed mk.ii)

6,288 ThomasRoderick  4 months ago
Auto Credit Based on AvalonIndustries's AST-104 ''Cerberus''

All credit goes to the original builders, Sadboye12 and Avalon Industries.


So two fine builders, one of them among the most distinguished builders of our time, Sadboye12 and Avalon Industries made a collab and created a monstrosity (in a good way) of a siege artillery platform, with the main guns so large that even the dreaded Schwerer Gustav looked like a toothpick in comparison: each of the three gargantuan mass drivers have a caliber of 1375mm (!), and is capable of accelerating those automobile-sized shells to over seven times the speed of sound, Every. Six. Seconds. Every inch of the magnificent construction exudes the rightful arrogance of a technologically advanced superpower, the malevolence of a supreme entity that can vaporize a continent by itself, th-
.
Tom! Quit drooling over that tank you disgusting pervert! I'm never allowing you to be within 30 feet from my car ever again! Who knows what you're going to do to it when I'm not watching!
.
... chillout, even I won't be attracted to that third-rate junker of y- - [punched mid-sentence] -
.
.
...Okay, back to the topic... I know I have no right to judge, but... the tertiaries, aka the "wing" quad-autocannon turrets beside the main turret, had some errors in its auto-targeting algorithm that made it drastically less accurate at ranges beyond 1 kilometer, while the settings on the rounds made them completely useless beyond 2.4 kilometers. With the fixed coding and revised setting, it can start knocking out important parts on enemy planes from over five kilometres away, and instantaneously vaporize targets that are unlucky enough to wander into the 2-kilometer range.


So What Happened:
Ever since update 1.9, people have been trying to use cannons to shoot down planes, from both surface-based and air-based platforms. And ever since update 1.10, people have been making auto-tracking AA turrets that can track targets faster than humans could react. The problem? There're two types of turrets, one cannon-based, the other gun based. What's the problem in that, you may ask? It turns out in-game cannon rounds have bullet drop and in-game machine gun tracers doesn't! So there are bound to be two types of auto-tracking codes, one to compensate for gravity and one that doesn't - and mistaking one for another meant not hitting jack * bleep * beyond point blank range. The two turrets used machine guns - yet the codes for elevation is for cannons. This resulted in the turret compensating for bullet drop that doesn't exist - which completely ruined the accuracy beyond anywhere more than half a mile or so (where drop from gravity is still minimal). Correcting the coding didn't completely solve the issue, either: perhaps due to an oversight, the guns, although massive by autocannon standards (they look like something at least 40mm in caliber, if not 76mm), yet had no bullet lifetime setting in overload, thus reverting to the default 3 second setting. 3s × 800 m/s = 2400m. So the shells disappear at 2.4km range - whereas most 35mm AA guns can comfortably punch above 5km range. This is addressed as well in this upload, where the projectile's lifetime is set to 8 seconds, allowing it to potentially hit targets more than six kilometers away - with a pretty high chance of actually hitting and taking some important parts out from 5 kilometers away.
.
.
... and those two turrets are locked with Ag5, not Ag4.

General Characteristics

  • Predecessor AST-104 ''Cerberus''
  • Created On Android
  • Wingspan 201.1ft (61.3m)
  • Length 247.3ft (75.4m)
  • Height 86.1ft (26.3m)
  • Empty Weight -4,734,385lbs (-2,147,483kg)
  • Loaded Weight 261,927lbs (118,808kg)

Performance

  • Power/Weight Ratio 0.205
  • Wing Loading 17.5lbs/ft2 (85.3kg/m2)
  • Wing Area 14,984.7ft2 (1,392.1m2)
  • Drag Points 24894

Parts

  • Number of Parts 321
  • Control Surfaces 0
  • Performance Cost 1,937
  • Log in to leave a comment
  • Profile image

    @Suubk27 Thanks!

    +1 3 months ago
  • Profile image

    @fleia262 Thanks!

    +1 3 months ago
  • Profile image

    @Sergio666 Thanks!

    3 months ago
  • Profile image

    @2008abdupro Thanks!

    +1 3 months ago
  • Profile image
    6,020 Grroro

    I am disappointed about the upvotes. I can 100% Guarantee that people who saw this on page were either in the mindset that you just reposed the thing or they didnt mind to come down here. Ive read the whole comment section and now i kinda feel sad about this. Here I am making so freakingly overpowered boats having no detail or no funky tress but still managing to get over 500 downloads and 20 upvotes. It honestly feels like i stole something from people who are 100x better than me. So good job on this! @ThomasRoderick its pretty awesome that you took the time and effort to make the system work correctly and then also took the time to write the description. this is underrated so ill share it with some people so that they find this and maybe even get help from it

    +1 4 months ago
  • Profile image
    6,020 Grroro

    @ThomasRoderick Welcome

    +1 4 months ago
  • Profile image

    @Grroro T h a n c c?

    4 months ago
  • Profile image
    2,212 Formula350

    1/1

    OK! I managed to refine it down today to make it ALMOST automated (ie we just pull the trigger). This is better than I've experienced so far :)
    Problem is, the Cockpit is inside the Flak Turret so it might have screwed up some other stuff... (The AA Machineguns are still just as deadly since they're spraying so much lead)
    .
    That being said... [important changes to the code are in bold]
    NEW Fuselage Block @ Coords: 0.0000027, 26.43127, -22.29239
    -------Rotation: 0, 0, 0
    -------Attachment Point: "Bottom" to the "Top" of Part ID 328 (the big fuselage on top of the triple guns)
    Cockpit Moved to Coords: 0.0000027, 26.43127,- 22.29239(yes, same spot)
    -------Rotation: 0, 0, 0
    -------Attachment Point: "Bottom" to the "Center" of New Fuselage at same coords
    -------(NOTE: Be sure to disconnect the cockpit from its original point; it will only be connected to this new fuselage part)
    UPPER Flak Cannon XML Changes (ID 173):
    -------explosionScalar: 3.5
    -------fuseInput:

    TargetSelected ? TargetDistance / (1250 - rate(TargetDistance)) * 1.01: 60

    LOWER Flak Cannon XML Changes (ID 172):
    -------explosionScalar: 3.5
    -------fuseInput:

    TargetSelected ? TargetDistance / (1250 - rate(TargetDistance)) * 1.05 : 60

    ------- CameraVantage -> viewMode: Chase (seems better, for when new targets are acquired)
    Flak ELEVATION Rotator (ID 168):
    -------invert: true (should be this already, but just in case)
    -------input:

    clamp((TargetElevation + rate(TargetElevation)*(((2000 * sin(asin((TargetDistance * 9.81)/(pow(2000, 2)))/2))/9.81) * 2)) + (asin(((TargetDistance + rate(TargetDistance)*(((2000 * sin(asin((TargetDistance * 9.81)/(pow(2000, 2)))/2))/9.81) * 2))* 9.81)/(pow(2000, 2)))/2), -12.5, 45) / 45

    Flak TRAVERSE Rotator (ID 165):
    (unfortunately its scale is 0,0,0 so need to edit this in the XML, or do as I did and edit its scale back to 1,1,1 so you can access it in game)
    -------invert: true (pretty certain it's already this)
    -------input:

    (TargetHeading - Heading + rate(TargetHeading)*(((1000 * sin(asin((TargetDistance * 9.81)/(pow(1000, 2)))/2))/9.81) * 2))/180
    (this could probably use to have those 1000 set to 1250)


    EDIT: Stupid Markdown codebox not using word-wrap... AND this site lacking the ``` code-block ability! Ugh

    +1 4 months ago
  • Profile image
    2,212 Formula350

    2/2

    (boy can I ramble, huh?....)
    EFFECTIVENESS:
    -In my tests --which were all performed at Wright's Spawn point-- I'm now able to, with ONLY the Flak Cannons (no AA machineguns), eliminate moderately-fast moving targets that are NOT performing elaborate acrobatics or have a high rate of climb/descent. Drone, Bomber, a few of my own builds, Astro's AS-15 Dagger, Enemy, were various targets I managed to take out.
    -Depending on direction of travel, I had to use a slight bit of main-turret rotation in order to lead or trail the target JUST A TINY AMOUNT so that the shells exploded closer to them; hence, thinking that Traverse "1000" may be fine to change to "1250" (or maybe it needs to go back to 800?).
    -RANGE though? In conjunction with my change to Fuse timing, that's now anywhere from <1 mile, to a peak of 3.5mi for slow moving targets (Drone), but an average Max of 2.2mi! :D If the Elevation with VTOL (noted below in Future Plans) were implemented, then this could be increased easily up to 5mi I think.
    -Targets coming in to land were the ones that were usually closer before being able to destroy (~1mi, AS-15), but not always, as slower planes landing were destroyed around 2mi.
    .
    Code Changes:
    Elevation - Changed the -10 at the end, to -12.5 (determined after trying -20 and -15)
    Rotation - Changed the two 800 entries to 1000
    .
    Additional notes:
    I tried a few things with the Elevation which didn't work out. Such as setting all the asin to sin, changing the 2000 to 1250, changing the few * 2 to various * 1.9x values, and combinations of those. Ultimately, they produced drastic changes and needed to be reversed, with finally testing changes to just -10 and having the results I was after; lower values drop the Point of Aim, so -20 is far too low (the obverse to -10's too-high impacts).
    .
    Future Plans:
    1) At least for use as a turret on a vehicle/plane that isn't already using the VTOL input, I plan to either...
    A) use VTOL as the "multiplier" in my Fuse code, so that you can adjust the fuse timing on the fly
    or
    B) use VTOL somehow, somewhere, in the Elevation code (maybe as a (-12.5 + VTOL)) so that the aim height can be adjusted for targets based on whether they're further or closer away.
    2) I need to add a damn camera to the turret body so that I can switch back and be viewing behind the turret, with the camera pointing at the new target! When mouse click is needed to rotate AND is assigned to fire cannon, it's... a waste of ammo to rotate the camera lol

    +1 4 months ago
  • Profile image
    2,212 Formula350

    @ThomasRoderick Yea the assumption I had was as you said, the first set of numbers are the velocity, right? Except the coding in this thing totally contradicts that?? :( The only place that it turns up is in the Fuze timing :\
    .
    Traverse code:
    (TargetHeading - Heading + rate(TargetHeading)*(((800 * sin(asin((TargetDistance * 9.81)/(pow(800, 2)))/2))/9.81) * 2))/180
    ... uses 800.
    .
    Elevation code:
    clamp((TargetElevation + rate(TargetElevation)*(((2000 * sin(asin((TargetDistance * 9.81)/(pow(2000, 2)))/2))/9.81) * 2)) + (asin(((TargetDistance + rate(TargetDistance)*(((2000 * sin(asin((TargetDistance * 9.81)/(pow(2000, 2)))/2))/9.81) * 2))* 9.81)/(pow(2000, 2)))/2),-10,45)/45
    ... uses 2000
    *throws papers in the air*
    So what the Eff, Jeff?! lol
    Anyways, since I have a moment, I'll throw some of today's thoughts at the code right now and see what it results in!

    +1 4 months ago
  • Profile image

    @Formula350 So the point is that I tried putting the cockpit on the edge of fuselage blocks, nothing weird happened beside the tank now yawing/turning a bit while the gyro-locked turret didn't thus spoiling the aim. But for the coding part... I'm equally bad (if not even worse) at math, but IIRC the first number after pow should be somewhere near 1250 (the muzzle velocity of the guns)? Other than that the coding is equally incomprehensible to me as it is to you.

    4 months ago
  • Profile image
    2,212 Formula350

    1/2

    @ThomasRoderick Yea I had a system going because I wanted to get through my tests quick, and so I wasn't doing anything unnecessary heh
    I'd spawn in, then press R to switch to A2A targeting, and press F twice (usually) to switch to Flak Turret. Then press Tab to select a target provided one wasn't already. (Sometimes I'd go through that sequence so quick nothing would've spawned in yet)
    .
    I will gladly test again to make sure, as indeed my SP had been running for hours by that point and who knows if there was some other anomaly happening. As there were certainly times when NO aircraft were spawning for me, and I'd have to click on Restart.
    .
    But yea, what I did first, after you mentioned cockpit's location, was I relocated it to directly in the middle of the Flak turret, between the two barrels Y position.
    Then, first, I just pressed R to let it reconnect. On that test, it was 180deg off, so I thought I did something wrong. Manually reconnected and tried again, same scenario. Then it clicked, that I needed to Invert the Rotator for the traverse.
    .
    Ok so step back for a moment. This is before I'd realized what's happening, but this to me seems like the first clue: If the part you're connecting to, its rotation, were to have no impact on anything... why would I have needed to toggle Invert on that rotator? (Not just an in-hindsight-question, but a legitimate question, in case I'm overlooking or missing something... @Sadboye12 you have extensive knowledge/experience, is there a reason you had inverted that rotator in the first place?)
    .
    After inverting, it was now pointing the right direction, but...... was still not properly aiming AT the target that was selected. So I went back into the designer and was setting up a new manual selection.
    NOTE: Since I had my cockpit buried inside a fuselage part, I was therefore selecting that part's Center attachment point, because I didn't want the game thinking that the cockpit was on an edge and that causing for some odd calculation mishap.
    It's when the connect-to-part's point selection came up that I noticed the fuselage I was selecting the Center point on, its arrows was not pointing Up like usual. That's when it hit me that "Oh? Is fact the part is rotated also impacting things??"
    So I opened Fine Tuner and sure enough, that part was 90, 90, 0. (could've been something slightly different, but X was 90)
    .
    Searching around the neighboring parts, the only one that was 0, 0, 0 was the back tapere

    4 months ago
  • Profile image
    2,212 Formula350

    2/2

    Unfortunately, the aim was still too high, but the Fuze timing now was much better. *(NOTE: The verdict of the Fuze was that for craft heading AT me, the 1370 [top cannon] code was dead on but the 1250 was detonating early; if traveling horizontally the detonated too late and too early, respectively, where probably 1310 is needed,)
    This high-aim was what I would end up being plagued with it seemed. The lead on the aircraft (traverse) was now roughly where it needed to be, but the elevation was off, which I'm not exactly sure what I need to adjust in the Elevation code to address that. Feels like the Gravity is off, yet touching that any makes it WAY off so that's not it. Those 2000 numbers don't seem to do anything when using 2000 or 1000, and the '2' in the pow(2000, 2) seems to have the same massive detrimental impact that adjusting the gravity does (just in a different way).

    TL;DR -

    What I've found is that moving the cockpit to the center of the Flak turret, between each cannon barrel's Y coord, is indeed best.
    That mounting the Cockpit directly to a part that is rotated, impacts how SP references how its spawned in and therefore its calculations. (I did not test with the cockpit rotated in the opposite direction of the part it's connected to, though)
    The Fuze timing is pretty good, but maybe could use a Randomizer to select a value between 1250 and 1370. Would lerp or pingpong be possible?
    And that at the end of the day, its always shooting too high, by quite a bit, causing a miss. Roughly an amount equal to .... 20-30meters? (I'm an Imperial, so yea, but it's >5 plane height's worth).
    .
    Lastly, as I quickly glanced at Snowflake's page, the only conclusion I can now come to on what might solve it, is using sin instead of asin? (I'm guessing, I have no math abilities, but plan to try this in a bit, after I get back from doing some stuff!)
    Problem is, I see it uses BOTH, and so, maybe removal of the asin? *epic shrug* :\
    [/massive wall of text]

    4 months ago
  • Profile image

    @Formula350 For some reason I didn't manage to replicate the effect of "If the part that the cockpit is attached to is rotated".
    You sure the hull itself didn't move/rotate after loading (which I did notice) and that the your craft/cockpit is pointing exactly at 000 degrees, with no bank and/or pitch angles?

    4 months ago
  • Profile image
    2,212 Formula350

    @ThomasRoderick Yes you told me, which is what motivated me to do more testing. It was that, which all the below were based on, as I went back and started testing over from scratch (ie, this upload of yours).
    .
    If the Cockpit is rotated, things are off.
    If the part that the cockpit is attached to is rotated, things are off.
    .
    And I feel like if they didn't insist on re-rotating things, and would keep what we apply for angles, this wouldn't be a problem lol
    (which by that I mean, whatever causes, for instance, if I input something akin to 0, 0, 180 and then when I go click back on that part, it's rotation is now suddenly 90, 90. 0 [or of that sorts] *sobs* whatever's behind that, I hate it, but I think might be at fault here too).

    +2 4 months ago
  • Profile image

    @Formula350 I think I've already told you about the cockpit thing.... The thing is, Jundroo's system is simplistic - and perhaps overly so. One would've thought the target data would be relative to the plane doing the aiming? Wrong! The coordinates are absolute for it's based on the relative coordinates of the plane(s) as "viewed" by the computer! And thus I'm pretty sure there will never be any localized FCS....
    .
    "High" as in "10 meters higher than the target"? 'Cause the gun is mounted about 10 meters above the cockpit.
    .
    Wait, I haven't noticed the part about a cockpit WITH rotation on all three axis being zero causing any guns to aim off by the rotation angle of the part the cockpit is mounted on...

    4 months ago
  • Profile image
    2,212 Formula350

    OK new findings after moving the cockpit...
    1) Screw you Jundroo! :D
    2) That SORT of addresses some of the problem; it's STILL shooting high, I'll get back to you on leading or trailing too far; Fuze timing seems pretty good now though.
    3) See 1. The part you attach it to, or at least how you attach to that part, seems to impact how everything functions. For example, I tried manually attaching its Rear-Bottom point, to various part's (fuselages) Center point... Result: ANY part that was oriented any way other than X:0. Y:0, Z:0 would cause the auto-aim to be "off" by that amount.
    4) Naturally, at least in hindsight, you can't attach it to any part ON the turret because... it moves, and as a result of the cockpit moving, the FT script can't calculate, and it freaks out. It'd always rotate 180 backwards, and then wobble a couple degrees left and right since it was basically stuck in a loop heh

    +1 4 months ago
  • Profile image
    2,212 Formula350

    @ThomasRoderick That sounds plausible...
    I wonder what is more important (seriously heh): having the Flak being accurate, or having the various side-guns, which also use auto-aim code....
    .
    Because if that's the problem, then the solution would simply be to move the Cockpit up to the Flak turret. The main triple-guns don't use it (I don't think?). so there doesn't seem to be much point having it centered near them.
    I'll have to give that a try, thanks. I didn't think the cockpit's LOCATION also had an impact on stuff *facepalm* Jundroo is really kneecapping us on what all can be done, despite giving us so much power with FunkyTrees coding :(

    +1 4 months ago
  • Profile image

    @Formula350 Oh, about the secondary guns (the big AA guns above the main turret)... the entire spaghetti code is WAY beyond me... Although according to the original builder (@Sadboye12), the problem is still caused by the gun being mounted too far away from the cockpit - * beep *, he actually said that part of the reason why the flak guns rotate with the main turret is because then you can manually correct that target lead... I think the problem is caused (at least partially) by the fact that the turret is mounted around 10m above and 10m behind the cockpit, which means the shots would constantly miss by about 10 meters, depending on which direction the plane comes from.

    +2 4 months ago
  • Profile image
    2,212 Formula350

    I'm seriously about ready to give up... In addition to being unable to time the fuze well, there's more!
    There seems to be a serious bit of screwiness, like with the Exhaust Scale when using asymm values, where it matters which direction you're facing for whether your jet exhaust looks correctly. or oblong shaped. :|
    .
    In this case, I'm having it....
    - lead craft too much when they're traveling S to N.
    - aim too far left when they're traveling W to E
    - trail craft too far when they're traveling N to S
    I assume, too, that E to W would to the inverse, as well.


    Unfortunately, I fear this is either a core SP bug, or a much worse bug that relates to Unity Engine....
    .
    I at least managed to get the Fuze timing a bit closer to detonating when it's properly near the plane, but it's hard to refine it further when I have these aiming anomalies that require a consistent aircraft spawning system heh
    I can hit stuff that's under a mile from me, since the accuracy is trivial at that point, but we already knew that.

    +1 4 months ago
  • Profile image
    2,212 Formula350

    @ThomasRoderick I can confirm what Woody was saying, that the Flak AA turret's aiming is still way off and missing almost everything unless it's super close. The one was only 2.5mi (just over 4km) out, flying horizontally in front of me, and every 5 shots exploded high above it (about at the top of its targeting square, visually at that range).
    .
    After shooting some more and tinkering, I realized there's a bullet cam setup on the Flakk heh Which is showing the round exploding WAY too early actually. So I wager that the visual effect of going "high" is actually because the round has only barely reached its peak height. That's seemingly confirmed by the bullet camera, as for craft flying at ~3000-5000ft alt seem to be below the round still just as it explodes, because it hasn't had time to arc back down towards the target.
    .
    I'm still tinkering to try and figure this out, but I'm --pardon the pun-- shooting blindly here since not only is the FT coding out of my league, but so is the math (yea I'm that bad at math shush lol). But it's also important to me, since I need to use this code for myself! :}

    +1 4 months ago
  • Profile image

    @Sadboye12 Thanks!

    +1 4 months ago
  • Profile image

    @Formula350
    IIRC it's more about the turrets being placed too far away from the cockpit, and the distance between the cockpit and the turret is directly translated to the distance between the supposed point of aim and where the bullets will hit. All cockpits would work the same, as the only thing that mattered would be the relative angles between the cockpit and the target.

    +1 4 months ago
  • Profile image
    2,212 Formula350

    @ThomasRoderick @woodypan666
    I don't know if this at all pertains to the aiming being off that was being discussed, but I just personally discovered that the Primary cockpit's orientation determines where the on-screen crosshairs point at.
    .
    I had my primary set at X +5deg (355 in Fine Tuner), which in turn caused my Machineguns to shoot "low", since they were at 0deg. Rotating that cockpit back to 0 solved that.
    I was using Cockpit-4 aka "Cockpit Block", the angled one that is 4th down in the list. Just in case it happens to be at all specific to only that cockpit variant, as I've yet to test whether it's also the case on any of the other types...

    4 months ago
  • Profile image

    @Formula350 Neither do I... If there's anything it would be having only one gun with an actual firingDelay, and everyone else set to have a delay of zero...
    .
    ... and let's just say there's a reason why I'm still sticking with miniguns and wing guns, and is desperately trying to find a way to simulate AOE splash-damge effect.

    4 months ago
  • Log in to see more comments