-
Posts
778 -
Joined
-
Last visited
Content Type
Profiles
Forums
Developer Articles
KSP2 Release Notes
Everything posted by Box of Stardust
-
A minor history note, Gunbrick came first to 'correctly' exploit the gun spam, which was fully designed in all aspects as a gun-based craft. TFD 2.1 was a response to that, which only proved that a huge amount of guns was a valid solution to almost anything in BDA 1.0. Basilisk wasn't really intended as a gun spam competitor, but just simply a plane with heavier gun armament than PEGASys at the time (though the Basilisk-D5 was intended to compete with 2.5km gun range to partake in the gun meta, but that was replaced with the full-on gun design, Gunbrick). That's why it's still around in BDA 1.2- because weight of fire is still valid... and why the new meta is mounting gun numbers higher than 4 guns. Do remember, standard gun armament in BDA 1.0 was 3 or 4. Because tagging a plane for half a second with one gun was good enough to make it explode. Not anymore though. In a weird sense, I've been building my craft for BDA 1.2's damage model even back in BDA 1.0. Basilisk-C3 from BDA 1.0 is... almost near-identical to Basilisk-C4 in terms of construction. Funny enough, BDA 1.0's heat damage model actually caused internal structures to kind of work against me. Okay, enough of that. You should probably link the BD Armory wiki page somewhere (I didn't see it anywhere). The entire page on AI tuning is still incredibly helpful. I should mention that my AI tuning method has always been to tune Damping first. With some experimentation, I finally understood what the BDA page concerning rotational input. Lower damping is 'looser' control by the AI, letting the plane rotate more in its axes more freely. Higher damping 'restricts' rotation. To have a plane that tends to act more super-maneuverable, lower the damping like nuts, and the plane will alwya flip over. It's difficult to find the right Steer Factor setting without knowing how much Damper is needed first, at least in my opinion. Damping defines how the plane will turn and maneuver. Steer Factor only influences how it'll make that turn. Also, 1 Sidewinder is actually perfectly fine, since it really all depends on if the AI sees a missile and launches countermeasures. In the majority of cases, firing two ends up wasting two. I still recommend having the setting on 2, but for craft with less missiles, 1 is a perfectly fine value. (Funny side note, in BDA 1.0, I had it set to 3 because the AI was dumb with using missiles.) A single Sidewinder's damage can absolutely wreck a plane as long as it hits right; you don't need two to do it. Just luck.
- 28 replies
-
- ai pilot
- fighter plane tuning
-
(and 3 more)
Tagged with:
-
Okay, so more battles have happened. Leaderboard updated. The HSC has started its run, and I can say that it now has THE BEST moment in ASC ever. Moreso than the Lancer vs P-5a. Funny enough, the P-5a stars in the new best moment too. But the poor P-5a is again, the victim in this event. Be excited for the footage. It's hilarious.
-
Just a note, gun jousting doesn't work anymore since it's actually been programmed out of BDA for the most part. The AI is just generally a lot smarter now. You may also want to quote my bit in the ASC thread that sums up good aircraft construction: 25% airframe, 30% weapons and utilization, and 45% AI tuning. Also- EditorExtensionsRedux. Everyone needs it. For building literally any KSP thing. Anyways, this is my rating system for BDA combat aircraft, with no specifics on how to do anything, just the core of what's what: Maneuverability: Maneuverability of an aircraft. Aside from general control authority, it also includes the ability to perform extreme dogfighting maneuvers, which may be influenced by factors such as thrust. Propulsion: Acceleration and speed of an aircraft. High thrust-to-weight is good, but can also be influenced by aerodynamics; know the ins-and-outs of the KSP aerodynamics system to help create less draggy aircraft. Endurance: Time an aircraft can remain in combat before running out of fuel. 15 minutes dry thrust according to KER readout is the lower bound I recommend. Size: Size and form factor of an aircraft. Influences ability to avoid damage, due to being a harder-to-hit target, as well as smaller radar cross section. Damage Mitigation: Ability to take damage and remain combat effective. To clarify: not simply surviving damage, but remaining able to fly effectively despite the damage and not be locked into basic defensive actions only, and if possible, continue fighting. Structural durability comes into play greatly. Offensive Capabilities: Aircraft armament and the effectiveness of the utilization of the weapons (weapon ranges and such). Flight Control: Ability of the AI tuning to both push an aircraft to its maneuvering limits and maintain stability and directional control. The AI must be able to perform extreme dogfighting maneuvers, evade incoming fire, pursue enemy aircraft, and be accurate when firing on a target. Additionally, the ability for the AI to accomplish this despite the aircraft having taken damage. (Note: this has become increasingly harder to judge with BDA 1.2.) Also, about missiles: it's not that planes are more durable necessarily (though it does help to prevent immediate aircraft combustion when it does take damage). They did something with proximity detonation and how damage propagates from the blast that tends to give missiles a 'hit-or-miss' effect. A really good missile hit (usually coming from the side or front, not chasing behind) will usually cause enough damage to blow a plane apart. Unless it's an ASPEN aircraft like the PEGASys, which gets pieces blown off, but otherwise won't be completely killed off thanks to its protected internal structure and redundant controls.
- 28 replies
-
- 1
-
- ai pilot
- fighter plane tuning
-
(and 3 more)
Tagged with:
-
I'm actually just working on trimming down some of the recorded battles and then going to do the reports on those. There's going to be a few more tonight, but the videos won't be posted until later. Any results, however, will naturally be on the leaderboard. Next match will be interesting. Du-4A vs Mini Shark III (which none of you have seen footage for yet). Stats are up for the MS3 though, so place your bets.
-
We don't know. That's the fun of it. Lots of different things could happen. We have to learn it and experience it to really know what's what about it, and mess around with it to know what's really possible with it and what isn't. What kind of weird things we could do with it, and so on. It's about both the function of the item, and potentially how we use the item. So, for another example, the routing devices we use here. I'm telling you all this by the way so you can get an idea of what, I think, makes ASC so special compared to the other BDA competitions out there. It's not just about the battles, but doing really weird things with BDA and finding out all the little things it can and can't do, and how to do those things or how to cause those things. And all without the restrictions of special construction rules like 'must have a cockpit and radome and weapon type limits, etc...'. So some more fun reading if you want; the idea for the multiple missile launcher (which you may have seen on a few planes here) starts around here: We had a problem in the last major BDA patch with the AI generally being pretty dumb and doing jousting stuff. There was only one person that really realized this at the time, and that was the guy that built this thing: https://kerbalx.com/juzeris/Cookie That didn't have much of a run in the series though; only one or two battles, so no one else picked up on that quirk until a few more battles with @dundun92's Du-3, and then I guess I picked up on it? At that point in time, I'd noticed that the BD AI pilot had a quirk where it would light the afterburners above 300m/s regardless of the craft's actual thrust capability. Now this was a weird random thing in BDA's AI pilot programming that most people wouldn't have noticed as a problem at all, but I happened to by chance because I set my max speed to exactly 300m/s... for no other reason than it seemed like a good number at the time. It was at this time that I reported the weird AI programming to the BDA devs, and they fixed it up in the next major patch. So that's one great thing for us. But another thing I discovered during this time was how exactly heat worked with BDA, with the help of the debug labels setting. Afterburners, naturally, are hot. But you wouldn't really think that if all you're doing is speeding around and stuff. But I was testing my craft against another very similar plane, but that one had a max speed of 325m/s. So its afterburners were on, but its throttle was all the way down at like, 20% or something because it didn't need all that thrust. But it was still 'running hot'. So I was always getting my Sidewinders off first at a fairly significant distance. Not just this, but I was also setting my guns closer so my missiles could do more work before the AI switched to guns, whereas everyone was just leaving their guns at 2500m, since a hit with guns in BDA 1.0 pretty much caused instant aircraft explosion. This lead to the whole consideration of 'thermal stealth' and trying to get missile locks first to break up the enemy formation instead of doing a jousting match. But that darn Du-3 was still really good! Well, it turns out, even if Panthers are pretty hot engines, if they're not on afterburners, they're still kind of 'cold'. By the time Sidewinders could fire, both teams would enter the 2500m default gun zone and the Du-3s would start firing before they reacted to any missiles shot at them. And the AI was pretty dumb (still kinda is with weapons) and the delayed switch time between trying to lock+fire missiles and switching to guns meant potentially losing a plane or two to the Du-3s. So, the dilemma was: how do you keep your missiles to deal with the rest of the leaderboard, while surviving the Du-3 which didn't have weapon switch times and was really deadly because of it and was not guaranteed to be missile-locked in time to split up the formation? Then I realized that a really good solution might just be the Du-3 method- but run even colder. Tiger engines, which we've already been using, run really cold. Heat lock distance under 2.8km or something and another few seconds for the AI to confirm and fire missiles meant ensured gun range firing. So I came up with Gunbrick: https://kerbalx.com/Box_of_Stardust/Gunbrick It's a dumb idea that boils down to this: small frontal profile to help avoid damage, symmetrical design in case of damage, and 10 guns to throw a wall of 20mm at the other team. And the worst part is: it worked. Jousting matches apparently do happen if you don't bother with missiles or can avoid having missiles being fired. Here's another hilarious example of my other gun-jousting craft in action, from PFC, another BDA dogfighting competition, which had construction rules, so I couldn't spam guns, but the idea is still there: So yeah. That's partially the BD AI being really dumb, and the damage model from BDA 1.0 causing planes to have zero durability. But... it's still kind of a problem that we had to get rid of. Planes flying at each other with some random deaths in between... is pretty unfair and un-fun. We want actual dogfights, not cheesing with some overpowered thing that kills things before a dogfight develops. So someone else came up with that funky Modular Missile Manager device that just drops missiles and just fires them. The missiles aren't aimed or locked; they're just unguided rockets basically. But the AI reacts to that, meaning jousting was no longer fully a thing. And that's the fun story behind the things that happen in ASC and how we discover the weird ins-and-outs of BD Armory.
-
If/when BDA implements those, then most likely no. We tend to ban things that break the spirit of creating the ultimate dogfighting machine. But of course, we'll test them out first and see what really is what, and if it warrants banning. Though we are the least limited in all BDA dogfighting competitions I know of, we still limit super crazy things because, well... here's the first time the rules got pushed, all the way 2 years ago in one of the first big BDA competitions I remember of: Which was awesome and hilarious, but completely shutting down the enemy team with not much effort turns out to really make the idea of the competition pretty pointless. We like to push the boundaries of what's possible with BDA combat aircraft here, and there are times when we realize we've crossed a line that would be considered no longer acceptable. This was the last one we've come across: It's kind of a silently agreed upon thing to just not put a lot of guns on a plane, because that just got really dumb; Gunbrick was pushing the limit, and worked because of BDA 1.0's quirks at the time. I've actually re-made Gunbrick for BDA 1.2 and it still sort of works, actually, but I'm going to hold off on it for now until I see what the HSC does... If you want a good read, I suggest you read through the previous thread (link in main post) to see the tale of Gunbrick and the progression of that train of thought. And just lots of fun battles.
-
Well, I'm off to do a lot of battles, because I really want to see the Du-11 and Basilisk-C4 slug it out. I've looked at the Du-11 in the SPH, and so far everything I can rate off of quantifiable analysis has it equal to the Basilisk (stats sheet updated to reflect these values). Except durability. So, we've yet to see which has better maneuverability and flight control.
-
I've just run a number of AI dogfights- - and there's a weird behavior where one team may not turn in to engage the other, or won't turn in to engage on time. It's very consistent on which planes it do it... or I think rather which planes cause it to happen to the other. Maybe. What might be causing this quirk? Altitude differences or something?
-
So a lot of battles have happened, but I just haven't had the time to edit/speed them up and upload them until now. First up, @ZLM-Master's X-Fighter Hunter goes on re-advancement against the PEGASys-D6. Can it win in the 5v5 rematch format? Battle: After Action Report: Next battle is two manned fighters trying to see who stays on top. @dundun93's ASW-MI(D) goes up against @53miner53's Fighter 3 Mk4: After Action Report: Next battle, we have a newcomer. @MDZhB's Alice Doll drone goes up against @dundun92's IV-7F to start. After Action Report: Next battle: After Action Report: @ZLM-Master's Support Light Guard is next, going up against the Lure Target. Small, light drone, vs big fish lure with lots of missiles. After Action Report: Next battle: After Action Report: Next battle: After Action Report: And that's all I've recorded! More to come, with a few heavy hitters on the bottom end of the queue...
-
@DoctorDavinci is it possible that something is messed up in the latest patch? A reinstall hasn't changed anything. Though, upon testing with my little test rig, I've found... some inconsistencies. These are the values in the editor: The Wing Connector Type C is given two different hitpoint values on different instances of it, and the tail fin apparently has 4000 hitpoints. However, when the rig is actually launched: The hitpoint values between the WC Type Cs normalize, and the control surfaces have a more normal 1000 hitpoints, and react accordingly. Still, the second hitpoint bar doesn't seem to make sense to me; I'm not sure what that is, and it doesn't seem to have any effect on the part (though it does retain an identical value between the editor and in-game). Also, the values for the Delta Deluxe here are at odds with the values in my other screenshot, so I'm just really confused (though, that was before I reinstalled, so maybe it makes a difference).
-
Well, the competition isn't too stiff at this point. Plus being flimsy only matters once it starts getting hit. It's really maneuverable, so it generally avoided damage in the first place. Like the Lure Target, it has the fundamentals down that make it effective despite having some flaws that can be beaten with a more heavily-designed plane.
-
I thought I did re-download the HSC when you said to. Maybe not. I re-download before sending planes to fight anyways. Technically speaking, the PEGASys is a biplane itself, and Basilisk is kind of a triplane... Really though, I just say my stuff either has double or triple redundant control systems, lol. The Du-4A battles are going up later by the way.
-
Drone leaderboard and roster comparison: Going from left to right: Corner: Armor Bud (which has too many hitpoints for like no weight disadvantage to be fair, so it's disqualified for the normal leaderboard) 1st row: P-5a Flash, X-Fighter Hunter, PEGASys-D6, Du-4A, Lure Target 2nd row: Alice Doll, IV-7F, Lumia Doll, mini drone (not very aptly named) 3rd row: D-1 Gen 1, Support Light Guard, Mini Shark III, HSC 4th row: Basilisk-C4, Du-7, Chinese Knockoff PEGASys Du-2A Block 30 The stats sheet has also been updated for size ratings for all these planes.