Ship Balancing - Star Control II - ADG Pro 8
ฝัง
- เผยแพร่เมื่อ 5 ต.ค. 2024
- Why the heck is my team of Spathi Eluders, Supox Blades and Orz Nemesis ships getting annihilated by a single Androsynth Guardian?! D:
So... today on ADG Pro, we're taking a deeper look into the balancing of the twenty-five different ships present in Star Control II, mostly from the Super-Melee perspective, to try and get an idea as to not only how well balanced they all are in respect to the point values they've been assigned, but to also derive some extremely useful data to make it much easier to select the right ship for the right job! ...err... deadly space battle! ;)
------------------------------------------
Additional Information and Corrections:
Here's a link to download the PNG file I created! If this gets updated in the future you'll see the revision number increase as well! www.pixelships....
Again, because this was all compiled as AI vs AI fights, there may be Human factors which could completely alter some of these outcomes. If anyone finds any glaring discrepancies, let me know so I can test them out and revise the table! :B
So far, the Thraddash Torch seems to be the most underestimated ship by this table. After some additional testing it's definitely clear that Human control of this ship has significantly stronger results than AI control. When watching the AI fight itself the opponents almost never get hit by the Torch's flame trails, but this seems to be more a consequence of how the AI is using this attack rather than attempting to dodge it, as the AI doesn't attempt to dodge it when used at range, but the AI never uses the attack at range. Interesting...
BTW: Not many weeks ago as of my posting this video, the legal troubles between Stardock (current owners of the Star Control franchise) and Star Control's original creators, Paul Reiche III and Fred Ford, FINALLY, after way too many years, came to an amicable conclusion! You can read all about the settlement from the creators themselves at the following link: www.dogarandka...
------------------------------------------
Pixelmusement Website: www.pixelships.com
ADG on Pixelmusement: www.pixelships....
Alphabetical Index of ADG Episodes: www.pixelships....
Kris' Patreon Page: / kasick
@Pixelmusement on Twitter: / pixelmusement
This is actually something that was extensively investigated and discussed around 15 years ago or so on the old Star Control Discussion Board, and one prominent member there actually fixed up the balance a bit with something called the "Balance Mod" and later renamed to the "Improved NetMelee Mod". I don't think it's maintained anymore unfortunately, though I could be wrong. It was great.
Regarding vanilla UQM, netgames used to be common (organized over IRC), and I can tell you the basics of how all the ships balance pretty easily:
* Thraddash is broken, so much that the convention for netgames at least when I played them was to ban Thraddash entirely. The main reason is its long-range, low-damage gun can both draw out fights and make it nearly impossible to beat in the hands of a skilled player.
* Most powerful ships (relative to points value) are Androsynth, Kohr-Ah, Chmmr, and Utwig.
* Worst ships (relative to points value) are Mycon, Ur-Quan (which the maker of the Balance Mod referred to as the "Banana Boat" at one point), Yehat, Arilou, Spathi, Orz, and to a lesser extent, Chenjesu.
* Best ships to start with are Shofixti and Zoq-Fot-Pik.
* Umgah can also be a decent starter ship.
* Other ships are pretty reasonably well-balanced and equally good.
Competitive play involves strategic use of "counter" ships which are good against other, more expensive ships. The most noteworthy examples:
* Androsynth counters Orz.
* Druuge counters Chmmr.
* Ilwrath counters Earthling.
* Utwig counters Chmmr and Kohr-Ah.
* VUX counters Utwig and Yehat.
* Earthling counters Druuge.
It's also worth noting that Androsynth is easily the most powerful non-broken ship in the game, which is actually why Orz is so useless. Androsynth is very good against a wide variety of ships and nothing counters it, you in a competitive game you can basically expect your opponent to have one. Some other notes:
* Mycon is useless because while it's very good against Druuge, its cost is far too high to justify using it against that target (and it's not all that good against anything else).
* Yehat is useless because Utwig is cheaper, better, and does the same job.
* Kohr-Ah is a very good high point ship that, in skilled hands, can usually last several battles. It's a good ship to deploy against Androsynth for this reason.
* Chmmr is another very good hhigh point ship that can last several battles in skilled hands. It's a good ship to deploy against fast ships like Slylandro and Pkunk, but less good against Androsynth.
* Ur-Quan is useless because it's such a high-point ship, yet does not tend to perform anywhere near as well as Chmmr or Kohr-Ah.
* Spathi is useless because it only works defensively (and not very well at that), and because a common rule to avoid dragging out matches was that the fastest ship has to take initiative, which Spathi is bad at. It also has slightly too high of a point cost to justify its use.
* Chenjesu isn't useless, but its cost is a bit too high to justify its performance. Better than the Ur-Quan Banana Boat, but still likely to be more of a loss than a gain.
* Arilou _can_ be used effectively, but it's very difficult to do so and realistically doesn't work very well against many ships. Usually Slylandro and Pkunk are better. The strategy is to use the teleport to try to pop into a place behind your opponent and sneak an attack in. The risk is sometimes you teleport into the planet and blow yourself up.
* Against Chmmr, choose Druuge and use the cannon's recoil to stay out of range of the Chmmr's laser, and sacrifice crew as necessary to keep your battery charged.
* The planet is very important to a lot of combat strategy, especially for Kohr-Ah, VUX, Ilwrath, and Chmmr (basically the slow ones).
* If your opponent has Chmmr with zapsats and you don't have Druuge, Shofixti can also do the trick to get rid of the zapsats.
For going above and beyond the call of duty, Captain Gemini, you are hereby awarded the highest medal in the Alliance of Free Stars for this outstanding work of military research. Neither the wisest philosophers amongst the Chenjesu, nor the farthest-seeing Pkunk visionaries, could have achieved these results. :P
...But yes, the AI is very lopsided. See what that Kohr-Ah is doing against the Androsynth - its battery is non-existent 90% of the time, and it never even tries to GTFO and recharge. So it spends its time as a sitting duck trying to scrape together a few red pixels to fire another blade, while the Guardian is literally running circles around it.
+x86VileR The flame circle attack isn't very powerful though for how much energy it burns. I did a test just now controlling the Marauder myself and going with a primarily flame circle tactic... and achieved about the same results as the AI going at it with the blades instead. :B
My tactic would be to stop firing for a bit and seek the planet (crew is somewhat expendable with the Marauder), do a gravity whip for speed, and give my battery time to fill up. Then while still whipping across the screen, face backwards, and quickly spam those puny clones with blades as they chase me. The Fiery Ring of Inevitable and Eternal Destruction is more or less a bonus if it gets close enough and there's red pixels to spare. ;-)
You did an amazing job! However you really underestimate the Thraddash Torch. In human hands it's a devastating ship. You can easily dispatch Ur-Quan Dreadnought with your flame trail for example. AI scripts for some ships seem to not be able to show their full potential.
+Wolod1402 Possibly... did some testing just now and yeah, the Torch does seem to have untapped potential using its flame trails that the AI isn't capable of handling, but I wouldn't call it "easy" to accomplish by any means.
@@Pixelmusement Well, I spent a huge amount of time in super melee back in a day so my feedback won't be fully suitable for an average player. Supox ships are amazing too if are controlled by a skilled player. I never was able to release their potential myself but I've seen some very good performance.
@@Wolod1402 Yeah the supox blade is a good counter to avatars, it has just enough fire power to get through the zap stats and is more than maneuverable enough to stay away from the avenger.
Thradash torch when controlled by a player is hilariously effective against AI controled avatars and dreadnaughts. The fighters the Ur-quan sends out make no attempt at dodgeing the afterburner, and the dreadnaught itself isn't that good at dodging either. Avatars seem to make no attempt at dodging the fire, so if you save up some battery and fire your after byrner right before the the avatar gets in range it will eat up a load of fireballs and take massive damage. Another hilarious matchup is against an AI controlled guardian, when the guardian tries to charge you it make no attempt to dodge the flames so if it come in from the rear you can just afterburner and vaporize it in like 2 seconds.
Even against faster ships that can dodge the flames the torch is still good as the afterburner can be used to get huge speed boost effectily making it the fastest ship in the game. It's main weapon seems weak at first but it's good range abilitypierce through a lot of defenses allows you can kite so many ships. combined with its increadable speed. There really is no hard counter to a player controlled torch.
Nice work, but I think you are massively underestimating of effect of the AI. Some ships have some very problematic AI and actually perform better with the "weaker" AI, as poor usage of the secondary fire modes can actually hinder the performance of several ships.
+Michael Mills Interesting... That's something that would be incredibly difficult to properly account for and would still be nixed by Human factors. Again though, if any specific discrepancies are dialed in, I can test them and update the table accordingly! :)
OTOH, his results matched pretty closely to how my friends and I felt about the various matchups, back when we were playing it religiously. Especially quirks like the Utwig being massively OP in the right hands, the Supox being utterly useless, and neither the Ur-Quan or Kohr-Ah actually being as badass as you'd expect. To me, this was watching all of our amateur non-scientific deductions be validated. :-)
There were only a couple places where I felt like the AI was really missing out on tricks a human player would know. (And the Supox doesn't count because the Supox just sucks.) Like, the AI didn't exploit the Arilou's gravity resistance like a human can. And most humans - in my experience - play the Pkunk as a strafing ship, rather than doing the Death Blossom thing. Oh, and the Torch can perform a bit better, with smart use of its afterburner. But still, mostly the results lined up with my own memories.
@@Pixelmusement
I remember doing some testing of my own at one point, focusing on the kzer-za dreadnought. Against a lot of ships, it's fighters are less than useless (Avatar in general and Marauder's FRIED both take them out very quickly) and will just result in it health being lost. From what I remember of my testing the dreadnought does much better against such ships on the dumber AI.
I also think this is at the root of the problem with the Blade you observed. The AI doesn't know how to use to advanced maneuvering system correctly.
+Michael Mills Definitely knows how to use it to some extent, given what I was witnessing. :P
@@Pixelmusement I feel the biggest one is the supox as the AI has no idea how to use its glide which an expert player can use to upturn most of those dingy matchups
The souls I've crushed with Pkunk in head to head over the years
AI Torches are usually different from Human Torches, which makes sense. As usually in a human vs human super melee fight the torch is frowned upon to use, as most people would go up to other ships with the torch and lay down a large line of fire for the enemy to run into usually causing damage while the new speed gained by the torch makes it escape any attack as it's now one of the fastest things in the game until it needs to do another round of fire. Against an Ur-quan dreadnought the main tactic is just go infront of them and lay fire behind you as you fly away, and the dreadnought can't turn fast enough to dodge it and the fighters are to stupid to move away dying instantly. These strats are very useful against the AI as they usually don't understand to avoid a torch's fire well enough. With people even if they do avoid them with the torches ability to become faster than almost anything at any time it just becomes a slog fest to even hit them until the torch will most likely win. Even against the avatar as the zap-sats won't take out the fire fast enough causing either them or the main ship to be hit for massive damage. Otherwise the AI just does weird spread out fire trails and main fire shots without much of the setting huge fire nets to catch massive ships or fighters.
Yeah the Torch is insanely good in the hands of a human.
Great job in compiling this - certainly will be a handy doc to have when diving into SC2. Even if it has some terrifying results (eg. every darn ship you're likely to have in the early game is rubbish against the accursed Sylandro Probes), it's amazing to be able to think about some strategies to try out for sure.
Well, that was undoubtedly deliberate, since it gives the player a LOT of incentive to put the Slyandro problem near the top of their to-do list. If the probes weren't a genuine nuisance/threat, they could just be ignored.
(Also, IIRC, they give a lot of RU when you do kill them, so it kind of balances out.)
The slylandro probes can easily be 100%ed using the spathi eluder in the solar system. The conclusions in this video are a long way away from accurate, but perhaps it would take some research within the sc2 player community to understand this.
Great video! Subscribed!
You're going to read the phrase "In the right hands" a lot in this comment section. Like right now!
In the right hands, a Cruiser is more than a match for any Dreadnought.
Also, GO! GO! Launch fighters!
You are a *!?!ing hero! This is amazing! There are definite some ships who’s results are wildly different in the hands of a skilled human pilot, but obviously that’s a whole other set of tests. The Androsynth Guardian is definitely the most under-valued ship in the game, and I can eventually beat most ships with the Spathi Eluder given enough time, including the Avatar. It’s really neat to see how closely your results match the original point values, though. Great and deeply geeky work!
I've found some ships better suited for human players, and others - for the AI.
Some ship abilities require human planning and ingenuity to work well (Torch is the most memorable example). And some ships are just relentless with an AI pilot (like Avengers providing an unexpected difficulty spike in the early singleplayer campaign, while performing quite poorly in the multiplayer skirmish).
Yeah, like the Arilou is actually better than its stats suggest *if* you can properly exploit its gravity resistance. The key to winning with an Arilou is tricking your opponent into hitting the planet. But that's more about psychology than controller aptitude.
(Side note: It's cool that the smartest race in the game has a ship which requires brainpower to master. Nice symmetry!)
Thraddash fire attack is not utilized well by the AI
Avengers special ability is plain laughable when playing vs human because of how screen centering works.
I mean, the other ship is always on the opposite side of game screen -- so other than homing projectiles being non-homing now, human just knows where Avenger is, no matter if its "hidden" or not.
@@rjherche yeah I remember using the torch quite effectively to attack my friends!
I agree. There are so many parameters here that you need a quantum computer to calculate the `real` balance sheet.
This is one of those videos where I wish I could give multiple thumbs up
Ultima's Guardian would approve of these results: "You have been a thorn in my side for far too long, Avatarrr..." :P
This was really interesting! Thank you for all the work that you put into it, and for the video itself. ^_^
First ship on screen: Eathling, Pirx at command... someone was reading Lem. :)
A lot of pilot's names are references
this video is fantastic
I really like how the data supports the game's *story* surprisingly well
+Dodo I'm not actually a huge fan of the game so I've never played enough of the story to notice that. :o
@@Pixelmusement The biggest complaint I usually hear is that it requires grinding so much, but it would seem that doesn't bother you!
+Dodo Grinding in and of itself is not a bad thing, rather it's the quantity you have to perform per amount of progress made. If you have to do a little grinding to make a little progress, that's fine. If you have to do a lot of grinding to make a lot of progress, that's also fine. If you have to do a lot of grinding for only a little progress, that's when there's something fundamentally wrong with your design. :P
@@Pixelmusement You haven't finsished the story? You should definently do that.
+Tarponpet Secord I'm pretty sure I explained this in my SC2 review, but the key reason I don't enjoy the game is because it's on a time limit. I prefer to play games at my own pace, but because everything is progressing at its own rate in SC2 I constantly feel stressed out about missing things or doing things wrong or that I'll go beyond a point of no return and be completely screwed and that just completely breaks the experience for me. :/
Great work here. I think your results line up with my experiences. I didn't play that much against other humans, but against the AI, I noticed that they were really good at certain things that skew the results in the way you saw.
First, the AI was really good at tracking, so the comet mode of the Androsynth was amazing in its hands.
Second, the AI could always "see" where the Avenger was when it was cloaked, so it was really good at getting in close to you.
Finally, the AI was really good at using the shield, so the Utwig was amazing. It could use the Yehat shields well, but that ship has limited energy and would run out unlike the Utwig.
That said, I definitely think the Avatar was the best of the 30 point ships, but it had it's counters. The Mauler could beat it pretty easily. The Maurauder was a soft counter because it could burn away its satellites.
I'm not too surprised the Marauder and Dreadnaught underperformed. The AI doesn't seem that good at using the blades to set up minefields. I think with the Dreadnaught, they wanted to keep it as the 30 point boss ship but it feels like there are a lot more ships that have ways of dealing with its fighters compared to SC 1.
Hell yeah man! That was some great work.
This seems like a safe video to confess this on, but I never made it very far in Star Control 2 because I kept getting filtered by the Slylandro Probes.
I recommend practicing with the eluder and then getting some they aren't that hard
Awesome!
Thank you for your video!
I decided to run the pair-wise defeat table through the Ranked Pairs algorithm, which produces a ranking not just based on the number of victories, but on the idea that a ship is strong if it defeats strong ships, and weak if it's defeated by weak ships. In other words, it gives the best choices if you don't know what your opponent will choose, but you can assume they are also choosing to win (rather than randomly).
I included the rankings implied by Kris's scores and the SC2 scores for comparison. Reading guide: "T14 (18, T13): Druuge Mauler" means that the Mauler is tied for 14th by Ranked Pairs, 18th by Kris's score, and tied for 13th by the SC2 scores.
1 (2, 7): Utwig Jugger
2 (1, T1): Chmmr Avatar
T3 (T3, T1): Ur-Quan Dreadnought
T3 (T3, T1): Kohr-Ah Marauder
5 (T3, 5): Yehat Terminator
6 (8, 17): Androsynth Guardian
7 (6, 4): Chenjesu Broodhome
8 (7, 13): Slylandro Probe
9 (12, 8): Mycon Podship
10 (T10, T15): Arilou Skiff
11 (T10, T21): Ilwrath Avenger
12 (9, T5): Orz Nemesis
13 (16, T11): Spathi Eluder
T14 (18, T13): Druuge Mauler
T14 (T15, 19): VUX Intruder
16 (19, 18): Syreen Penetrator
17 (T20, T15): Supox Blade
18 (T12, 10): Mmrnmhrm X-Form
T19 (17, 9): Pkunk Fury
T19 (T12, T11): Melnorme Trader
21 (21, 20): Earthling Cruiser
T22 (T22, 23): Umgah Drone
T22 (T22, 25): Shofixti Scout
T22 (T24, 24): Zoq-Fot-Pik Stinger
25 (25, 21): Thraddash Torch
There are several interesting differences, but the most visible one of course is that RP gives top place to the Jugger, because it defeats the Avatar, which defeats everything else, making it the meta-winner. The X-Form also gets a big downgrade as it's defeated 5-1 by the otherwise-weak Supox Blade, and the Trader loses out as it's defeated 5-1 by the X-Form. Everyone agrees that Earthlings are weak.
AI vs AI is interesting, but the AI isn't "perfect" and data from it cannot be the final word. For example, when playing Torch against AI, AI has no chance because it doesn't avoid the afterburner. When playing human vs human, Torch can stay out of range of every other ship and take 1-point potshots every now and then, resulting in tediously long battles and being unwinnable in skilled hands against every other ship. This balance mod to Super Melee is great fun: wiki.uqm.stack.nl/User:Shiver/Balance_Mod
+Eino Keskitalo I wouldn't call the table I came up with the final word either; it can clearly be improved upon given Human control, but it's absolutely a good starting point! :B
i wish more games used the .mod format
most early Unreal Engine games used tracker formats. Notably Deus Ex, Unreal itself. Also Age Of Wonders 1 had MOD music.
Hey! Just found your videos. First, I have to give a 100 thumbs up for spending so much time analyzing this game. I totally agree on you in regards with ships points and balancing are not really meant to 'balance' each team. And yes, you'll get decimated by a better players, fair and square.
I've been playing this game for years and probably can write basic guide on melee, something for you to try out again in the next few years.
The main key here is that each ship has a counter! And I'm talking player to player. Let's start
Ur Quan DreadNaught - Any ship that can kite from far and fast enough to avoid it's fighter. That's X-Form, Skiff even the lowly Spathi has a chance. Earthling cruiser, Chenjesu and other slow ship are food for this ship. Kohr Ah Marauder can win unscathed. And my favourite battle will be against the overpowered Chmmr Avatar. The fusion blast can take out the satellites and then the fighters will finish the job. The key is to keep a mid range distance.
Kohr Ah Marauder is an overpowered ship. On random matches , this usually takes out 3 opponent ships :D . Get taken out easily by the Utwig Jugger because the shields absorbs the blade. Just go in absorb and shoot from mid range distance. Yehat Terminator does the job well too but you need more energy management and shield timing.
Thraddash Torch , terrible ship. Get's taken out by pretty much anything. As long as you can stay put and not give a chase, the torch flame is useless.
Chmmr Avatar - Another overpowered ship if you are too close. Any ship that can move backwards like Supox or fire from behind like Orz Nemesis. The perfect counter for this ship is the Druuge Mauler. The kickback from the main gun keeps the distance, but you got to have good aim and not let the Avatar pulls you into an irreversible speed making aim impossible. Also the trick in using avatar is to pull ship while chasing and keep it in satellite hitting range, not pull them towards you and then bump away.
VUX Intruder - Most annoying ship of them all. A match usually last 3 to 4 minutes as the enemy is pummeled with Limpets until it can't even turn. Then it get's pummelled slowly with a green lazer. Countered easily by Earthling cruiser or any fast ship with long range homing missile. The caveat on this ship is that, as soon as the match starts, move forward and shoot because you always start near your opponent with your laser already pre aimed at the opponent.
Ilwrath Avenger- This ship, in spite of being ludicrously strong in AI match, is always ludicrously outmatched in human to human melee because a) you know where it is, b) you can out run it. If the human opponent panics and shoots randomly, yes, you can win.
Arilou Lalee'lay Skiff - Crazy kamikaze physic defying ship with auto aim lazer. It's fast enough to kill a Dreadnaught or most ship slow ship that has no auto aming. Go in fire, fire a few shots and teleport out, hopefully not into a planet. Anyship with rapid fire or homing missile can take out the skiff.
Umgah Drone - Maybe you think this ship is useless, but the trick to using this is, 1) reverse and bump into your enemy to make it move. 2) reverse again into the enemy path a short while later with the antimater cone. Hard to master, extremely satisfying when you get a kill.
ok, one last one to rant about and maybe my all time favourite ship.
Earthling Cruiser - The first thing you need to do is to run for the planet and give yourself a gravity swing. As you get up to speed, make sure you to the back of where you are going, not to the front, as the missile can hit you back. You then time your missile shots in the general direction of the enemy, keeping energy full, while sustaining gravity swing and let hilarity ensues as your enemy do crazy maneuvers to avoid the missile. Takes a lot of skill to stay out of range of bigger ships, so the planet is your friend.
I can probably write for all the ships, must have played too much of this when I was young (I passed my exam decently well, so that's fine). If you need more, let me know.
Yeah, after getting this video out it was definitely made clear that there were certain tactics I never expected to try myself. When it comes down to it, I seriously don't play any better than the AI does, in fact, I play almost exactly on part with the AI, so that probably at least partly explains why I've never been able to beat the main single player campaign. ^_^;
I love this game. Nice work!
Interesting watch. I appreciate your dedication in making this as well as the other pro videos. ^ _^ _b
Also, if I were to adjust the point values before seeing this video, my intuition would’ve gone in a very similar direction as your results, especially regarding the biggest gains and drops. I’d bet many longtime sc2 fans would agree with most of your findings!
+MusicByproduct There's been a few disagreements, but my own attempts at confirming said disagreements by playing particular combinations myself have led to similar results compared to having the computer facing itself, so I think you have to be REALLY good at the game to skew this table.
@@Pixelmusement Yeah, speaking from experience you have to be either really accustomed to the AI's behavior or really good at handling the ship you've chosen. I used to use one Androsynth Guardian or Supox Blade to take down entire fleets of ships back when I played the game regularly. Nowadays I'm not nearly as good but my friends and I still remember the days of huddling around a computer, sharing the same keyboard and seeing who could win with the most amount of ships remaining.
Idk where I'm going with this, but despite my disagreements with things like the Earthling Cruiser (which the AI gimps by not turning around to face the other ship), I think this is a pretty solid video overall. Thanks for making it, it's always nice to see StarCon 2 videos being made :)
Shout out to the Utwig Jugger, a 22 point ship with the most reliable overall odds. A couple others have more A and B matchups... but it's the only ship that doesn't have any D or F matchups, it's worst opponents are still basically even matches. (In AI vs AI, of course, who knows how this all translates to PvP?)
Best combat ever. I just wish there was an upgraded SC2 version with smooth continuous rotation instead of stepped.
It's something I really wanted for Ur-Quan Masters HD. The higher resolution makes the lack of animation more apparent. But it would greatly affect balancing & required a complete re-write of the engine. Fortunately, they're making Ur-Quan Masters 2, which will include smooth rotations.
I love this games, and it's remakes.
Honestly, I think the grading system underestimated the Torch. If used properly, it's able to do crazy damage against any of the slower ships. Get in front of them and fly directly along their flight path, then hit the burner, and you'd be surprised at how hard it is for ships to turn out of the flaming path you've laid down. You can also use the burner to make high speed passes in an arc in front of or directly across another ship's path. The only race this doesn't do some damage to is the Arilou, since they have an inertialess drive and can stop dead while every other ship uses Newtonian physics and has to kill or overcome inertia to change direction. Keep doing it repeatedly and use your speed to stay away from the enemy, and even the Ur Quan or the Kohr Ah will eventually bite the dust. And the Chmmr are helpless against this tactic, as the flame bursts will destroy their zapstats if they fly through them and the Thraddash ship is very fast and very good at accelerating, so the tractor beam isn't going to be much use against them.
Yeah, there's some definite differences in these results for expert players, but I'm not an expert by any stretch and my own ability to fight with these ships was comparable to the AI, so given all the comments I've received on the table I came up with I think it's safe to say these results are more useful for average players, not experts. :B
Great video. Love this game, still play it. Although you’ve already touched on this I’ll say it again; human players (good ones) handle many of these ships way differently from how the AI handles them.
+Sullivan Nix I've done a couple tests myself with a couple things people have pointed out and actually achieved similar results, so even in cases where a Human will play much differently the end results so far haven't been different enough to warrant any updates to the list... yet. Time will tell! ;)
@@Pixelmusement a skilled human player in a Fury can generate a good loss ratio against the AI Avatar, Dreadnaught, Marauder, and Jugger. Against the Avatar, you get the laser pointing away from you, then you charge in, fire continuously, bounce off, and retreat. The Fury fires 3 bullets per shot, and those saturate the Zapsats. With the Dreadnaught and Marauder, you hit and run. You can pick off the Dreadnaught’s fighters with your gun, but you may get hit. With the Marauder, it’s simply a game of dodging the discs, and you can exploit the fact that the AI will deploy the FRIED system at any incoming shot if it has energy. With those ships, you just want to keep land a long string of hits on the other guy and hen break off. With the Jugger, it pretty won’t often shield in time, but I think I use more short bursts against them.
Great analysis!
Kinda sad to see the awesome Supox Blade being rated so lowly. In competent hands it can defeat all the big guys like Chmmr, Chenjesu, UR-Quan and Kohr-Ah for a very decent price, as well as Utwig Jugger if you deplete it's shields first. Backwards propulsion is extremely powerful if used correctly.
Every ship gains new potential in the right hands of an expert player. The final chart is more of a baseline as to the relative value of everything and the type of battles to expect without any skill advantages. :B
I'm wondering why you were surprised for the Spathi eluder results. Eluder is a fearsome ship, able to take down both UrQuan and KohrAh with relative ease. I have a feeling that AI was not piloting them properly. While sure, back-firing missile may suggest defensive ship, isn't dealing damge to opponent while not taking yourself is the key to winning?
Spathi Eluder is not as good a ship against skilled human players. It works so well against the computer because the AI will have the ship run into the back- firing missles. Good human players won't behave like this and have found solid counters to it. A somewhat skilled human-controlled Kohr-ah Marauder, for example, will not be losing to a Spathi Eluder despite what the AI would have one expect.
I used to love trying to take out random enemy squads with a full squad of Zoq-Fot-Pik stingers for a fun challenge. I always loved that tongue attack for some reason. Yeah, they are correctly one of the worst rated but I've managed to take out several slower moving beasts like the Dreadnaught with sting-and-run tactics before. The Avatar on the other hand, lol yeah that aint happening.
Also, I know this doesn't apply to super melee but the Pkunk fury was the ONLY ship I could beat the Sa-Matra with. Their speed, rotational speed, and most importantly side firing attack lets you easily just fly by it (rather than directly, which would force you to slow down and change course more often) it while hitting the shields and avoiding the projectiles.
Another related-but-not-quite-the-same judgement is what I call "staying power," which is how often a given ship survives a random battle before finally going down, i.e. how many ships it can kill before it dies. Sometimes a ship is really good against most others but has a real weakness against one or two sometimes weaker ships. A good example is the Yehat Terminator vs. the Vux Intruder. Another ship might not be as good on average against another random ship, but doesn't have the Achille's Heel and therefore might have more staying power. The Chenjesu Broodhome and the Khor Ah Marauder both fall into this category for me. If I had to stake my life on being able to beat a certain number of unknown ships, I'd choose them over the Avatar. With really skillful playing, they can kill a nearly unlimited number of other ships until very small mistakes and chance asteroid hits finally end things.
With this i can make the accurate assumption that the Melnorme, Kohr-Ah and Chenjesu AIs are really bad. The Melnorme one in particular just doesnt feel like it needs to charge up shots and it doesnt use its secondary often. The Kohr-Ah AI just fires off sawblades way too much, leaving it without enough energy to use a FIRED which could help in a lot of battles. The Chenjesu AI just throws out a bunch of DOGIs but pretty much doesnt attack.
That's interesting, I very much agree! In the sc2 competitive community, the melnorne was one of the very best ships. It was one of the most common picks.
There again, this is only where the thraddash torch was banned (which it usually was), which this video rated at the very bottom!
The melnorme trader was an incredible ship in the competitive scene. It has very few terrible match ups (an important feature when you get to pick your ship to counter the one that just beat you), and it's special weapon could deactivate the shields of the jugger and the terminator : ships that usually required 'big' ships to get the upper hand against)
The kohr-ah marauder was also an extremely good ship. Top tier and selected in most competitive teams, for sure!
I know you said 'accurate' prediction, and I think you're dead right! The ai did not play those ships to their potential! At least 2 of them were top tier.
In fact, generally speaking, using the ai to assign combat values to ships is never gonna be nearly accurate!
The chenjesu broodhome was, indeed, pretty bad for its cost, but it was my personal favourite, and it certainly wasn't unusable, it could still pull its points cost in the right hands, but was never 'top' in a competitive match.
The default value for the Ilwrath Avengers makes more sense from the view of a human player instead of AI. Humans users wouldn't know the exact positioning of a cloaked ship and thus would fare much worse in combat - one of the few cases where AI is usually superior to even a skilled player.
The camera scrolling also ruins the avenger, It's pretty easy to get a general Idea of where an avenger is, so you can throw a few shots in its direction and if you get scared you can just fly away. Any ship that is faster than an avenger can counter it pretty easily.
You're batshit insane. I like that.
I prefer the term "Inspired". ;D
@@Pixelmusement Batshit inspired then!
Damn, I haven't played this in a long time...of course, I never had the chance to get good at it...
The computer uses the Thraddash wrong. I can take out almost any ship with the Thraddash. it also says on your chart that Yehat won 5-4 against the Ilwrath, I beat 10 Avengers with 1Terminator. but this is A.I. vs A.I. and sometimes the A.I. does not fight as well as real people. I love the Yehat and Thraddash. 5 Torches can take out 10 Dreadnoughts no problem. The Thraddash if you know how to pilot them, they are unbeatable. except against the Yehat and Utwig. Thanks for all your hard work. I did the same thing when this game came out, but I lost all my data.
Yeah, I'm discovering from all the comments that experts can really skew the results with certain ships. I tested all of the ships myself but I'm not an expert at the game and found my own results with each ship to be similar to the AI results, so I just sort of assumed the AI was playing well enough; I didn't expect some of the extreme edge cases that are showing up! :o
@@Pixelmusement It's all good. I loved the video. I remember when I did the same thing ah nostalgia. Again thanks for the video.
Of course, almost all fights can be won, if you try REALLY hard, but skill cannot be measured. Lets just say that Avatar is a LOT weaker without the orbitals, and very skilled Blade users are able to take down most of the ships. Anyway, awesome effort!
Surprised you said earthlings weren't good against any strong (or big) ships... but in my experience, they counter ur quan very well.
Their point defense prevents fighter launches, so the general strategy is to gravity swing and lob nukes from oblique angles. It's a grind but in my experience you win 60% of battles, considering the points difference, that's impressive.
I know Ur Quan are not great for their cost, but i always thought it was by thematic design that the humdrum humans were best against the main villains of the game
In the end I could've clarified better that there might be Human factors I'm failing to take into account, since this was all based on having the AI fight itself, but my own experience suggests a player with AVERAGE skill at this game will follow the rankings I came up with pretty closely; Experts are the ones who can skew the math by doing things the AI or an average player can't. :B
This is all well and good but a Player can exploit AI pretty well, at a certain skill level:
for the campagin mode:
The Spathi is aquired at the very beggining, and you can kill a Probe (which takes 3-4 minutes of bait and switch), with only a 1 hit, or no hits at all.
The Thradash is grossly OP. Your go-to mid-game ship.The AI doesn't seem to have a clue as to how to dodge the fire attack, and it's quite overpowered against the Green Ur-Quan. It is even usefull against the Kohr-Ah. Also good against the Mycon and the Illwrath. Doesn't perform that bad against the VUX and the Yehat as well.
For End-game I've actually found that the Chummr isn't really all that great, because it's hard to kill an enemy ship with no hits (which is my goal usually), but the Utwig can clean down almost any enemy ship, although, it's a lot harder to get it (compared to the Thraddash).
It IS fun how this game kind of predates a lot of the old "hard counter" unit selection I've seen in more RTS/fighting/RPG/etc. games as the years went on, where a character/unit selection that you wouldn't think could win a fight against way stronger opponents could instead easily kick the crap out of specific ones. Of course, I imagine this game wouldn't do something so silly as, say, making the Pkunk Fury an absolute engine of death against the final boss or something :U
Nice work, however since this seems to be AI vs AI, with some ships even the advanced cyborg uses non optimal solutions with some ships, and that can affect ratings. With this game, as a human player you can become more experienced with a type of ship and that can lead to completely different difficulty match tables, since the AI doesn't exploit its own strengths in battle sometimes, and other times you can just become more expert than the AI. As well sometimes you can exploit the AI as a human. For example Spathi vs Slylandro you have as a D. However that is an A+ battle for me due to an AI exploit. There is a strategy there the AI doesn't use where you can fly away from the Slylandro turn yourself at an angle and shoot out BUTT missles until the opponent gets closer than speed away and repeat where you can almost do a guaranteed win, heck I did 150 Slylandro kills in a row with 1 Spathi in the no Starbase challenge no problem. Other times it is AI using weak strategy, so the Thraddash Torch is actually a very good ship against the Ur Quan Dreadnaught, Kohr-Ah, and Chmmr however you have to ignore using the primary weapon and only try getting the enemy with your secondary weapon as you get near to use it and thrust away with it and get them going into it, and avoid their weapons. You can get skilled enough easily as a human to use this strategy very effectively against most ships. Since the AI only uses those thrusts as a means of speed and not killing power, and tries using that weak front weapon to kill, it has a serious disadvantage. As well some ships can see huge difference with massive skill increase Supox Blade backwards and side to side ability when honed in can decimate a Kohr-Ah or Ur-Quan and a lot of other ships no problem, and an expert Arilou player, which I am not, I've seen do wonders against other ships you'd normally be at a disadvantage with due to being experts at avoiding enemy weaponry using the warp ability or just getting in between firing spots very well.
Very cool
Great video - is there any reason it's better to run original SC2 in dosbox instead of just playing "ur Quan masters"?
Nope. In fact, it's better to run the Ur-Quan Masters on a modern system. The trick for me is that I focus on the original DOS titles with this channel because that's just the way I roll! ;)
AI is not a good player. All references according to AI player will not work in human-human deathmatches.
But your effort is A+++
by random chance do you happen to have the raw data for ship versus ship kills available or just the averages?
I still have the raw data, though I think I may've made corrections on the fly during editing without updating the data itself, so fair warning there: www.pixelships.com/adg/files/pro8rawdatat.pdf
@@Pixelmusement This is cool, thanks! I'm also curious if you have data for individual matches used to calculate the means in this chart? The day before I found this video I was basically recreating the same results because I was also curious, THEN I found your video. A few results still interest me (eg. my results show Arilou vs Kohr-Ah had a mean of 3.422:4.608 with wild variation) and I was hoping to look at more data driven results such as the standard deviations or modes of different match ups.
I'm testing on the most recent Ur-Quan Masters release and since I wasn't finding anyway to speed the gameplay up instead have ~15 instances running at once but it is still taking a very long time. I've seen at most only minor discrepancies between my results and what your video shows so I suspect all our results will likely be very similar. If you could help speed up this process for me that'd be amazing! I also totally understand if those results are lost to time; this is probably one of the stranger requests anyone could get.
@@Snowfats Because I had sooooooo much data to cover in a relatively short amount of time I streamlined the process and thus the data you see is the only data I tracked, though you'll note some of the boxes are coloured. I believe I chose colours to indicate notable battles, recorded battles, and battles which took FOREVER.
@@Pixelmusement Dang; that's still really helpful though!
star control 2 source code is available here github.com/videogamepreservation/starcontrol2 so you can directly look up what is going on under the hood
The only way to make SC2 balanced is to force always AI for all arcade battles. When you have a human onboard a ship, you can win any encounter.
At least theoretically.
That would depend completely on how good the ai is.
In fact I'd lean the opposite way: if the ai was programmed well enough, a human could never win.
This is problematic. Each ship has a number of available strategies to use vs opponents AND a skill level required to operate it effectively or optimally. The AI is flawless when skill level is being measured BUT only uses ONE strategy all the time. This massively hampers the AI on ships which require changes in strategy depending on the opponent and massively boosts the AI versus humans in fights with ships where skill level matters a lot.
For example: The Arilou is lethal if piloted with a highly skilled pilot but useless if your timing is even off by a fraction of a second. And of course my FAVORITE ship the Thraddash, This ship can slaughter nearly every ship in the game including the Chmmr but only if you carefully match your strategy against the weaknesses of your opponent. Or of course luck based ships like the Pkunk... I kid you not I lost a chmmr once to a Pkunk that refused to die, while everytime I fly one it jut dies the first time.
+T Sisko This doesn't surprise me. Again, this table is more-so a starting point and a means to gauge base potential. It can be tweaked over time. :B
I remember I used to believe that holding down shift when the Pkunk died it would increase the chance of revival. Don't know if it's true but it felt like it made a difference, lol
@@joonasfi Yeah I remeber spamming 'special' whenever a pkunk died, it's literally just a 50% chance evey time. If you playing pkunk you're probably just holding down special the whole time anyway though if you know what I mean.
Is there a Super Melee that allows battle via players on internet?
Yep! The Ur-Quan Masters has support for netplay, though I've never attempted it so I have no idea how it works. It's probably done as a direct connection to your opponent's system, meaning the person hosting would probably have to do some port-forwarding stuff to get it to work right and the opponent would need to know the host's IP address in that moment, but that's just a guess based on how basic netplay tends to work in projects like this one. :B
Eh the thradash kills the chmrr avatar so easily. In fact the thradash I think is the most power full ship.
The computer has a lot of bad habits in SC 2. Really even, Human vs Human, Computer vs Computer, and Human vs Computer are ALL completely different. With the obvious being how in Human VS computer, you can merciless Kite infinitely with almost all of the "light" ships. Even something like the Drugge's knockback is easily enough to kill computer controlled avatars any number of times in a row. You can also AI abuse how the computer tracks you when you repeatedly screen wrap the top/sides of the battle area - Earthling Humans can trick either Ur Quan into constantly turning around even though if it kept moving in 1 direction it would out-accelerate you. Ships that are extra reliant on using planet gravity slingshots like the Intruder and the Podship play obviously different even in human vs human matches. The game's balancing of ship cost for human play is a longer topic, but unmodded… the Androsynth and Shofixti are well known to be brokenly effiecent against an equal "point cost" of enemies. - so I like the intent of the video, but it isn't really adding much to the conversation for either singleplayer against the AI / in story mode or using UQM to play people online with or without Shiver's balance mod.
+Edward Oakley It was still neat to see how this turned out, but yeah, adding in the Human factor to a table like this would take far too long to do for a weekly show, so we can rebalance this table over time for sure. :B
Hey, have you looked into the black spathi squadron?
wiki.uqm.stack.nl/Black_Spathi_Squadron
There should be something in the game code if it does exist
I think you are half rigth, yes the data its ok and all but is A.I. made
So, the ships balance in the A.I are in the video, and not the actual human results, ye it would be a long run to do all that with humans, but i think it would be wayyyyyyyy diferent
Like the spaty boyz, they are amazing in singleplayer for humans to use cuz u can just cait around the other ships and there is not much the A.I do against you but on your table it looks horrible
Idk good video tho nice work
Generally speaking, after all the comments I've seen in response to this, compared to my own testing, I think the results I got are most comparable and reliable for the AVERAGE player. Experts are definitely going to skew these results as they will have much different ideas about how to take advantage of each ship's special features. :B
Ah, since you mention that you think the AI isn't much worse than humans, I can tell you from experience that you're wrong on that. In fact the AI is completely incapable of piloting basically any ship remotely well. I think the only ships the AI pilot well are Slylandro, Yehat, and Umgah. A lot of this comes down to the AI never using the planet strategically (which is important for a lot of ships), and a lot of it also comes down to the AI just being too restrictive and simplistic in how it uses ships (e.g. it won't take advantage of Druuge's recoil or Kohr-Ah's long-range capability). Really, playing against an AI is nothing like a competitive game between two human players.
EDIT: I see this coming up a lot in this video, so let me elaborate about things the AI does poorly:
* Doesn't shoot with Earthling efficiently at all (often shooting the missile while facing away from the opponent).
* Fails to use Kohr-Ah as a long-range ship.
* Doesn't shoot Pkunk's weapons as much as it could (Pkunk is actually a very powerful ship when used correctly and only ever so slightly overpriced).
* Often fails to use Androsynth comet when it should, or uses it in situations or ways that it shouldn't.
* Never takes advantage of Druuge recoil (useful especially against Chmmr, which it is very good against).
* Never takes advantage of Thraddash's long range.
* Often controls Supox terribly, as you saw. Supox is a decent ship and can even be quite powerful in the right human hands.
* Never takes advantage of planet's gravity, which in particular makes it worse at piloting Ilwrath, Kohr-Ah, Chmmr, and VUX.
* Never uses Arilou teleportation offensively (which is the only way to use Arilou effectively).
* Doesn't systematically slow down the enemy ship with VUX, which is essential to using it correctly and highly effective against both Utwig and Yehat.
* Pretty much always fails to react correctly to Ilwrath's cloaking, which affects Ur-Quan and Orz especially badly. Ilwrath isn't even remotely capable against Orz when piloted correctly.
* With Yehat against Orz, attempts to use the shield against marines. This is much less effective than just shooting at them.
* This is a weird one, but the AI is bad at playing against a Mycon which is turned away because of how it decides whether to dodge. This allows AI to lose against Mycon with ships that should beat Mycon easily.
* Pretty much always fails to adjust for Chmmr.
Chmmr is actually a lot less powerful than you observed. It's powerful, but most of the perception of power is that it's highly unusual and requires specific strategies. Androsynth should immediately turn into a comet and ram into it as much as possible, Shofixti should immediately go in and try to blow up next to it, Kohr-Ah should immediately try to blast it with the FRIED, Druuge should use its recoil to stay away while blasting it to smithereens, VUX and Mmrnmhrm should just try to do as much damage as possible with their lasers, Orz should try to stay away while shooting shots backwards (it has just enough speed to not be sucked in), Chenjesu should just throw crystals and try to hit the Chmmr dead-on.
The AI's flaws also gave you an incorrect perception of Ilwrath, which is a mid-range ship, not above-average in power but a perfectly good fleet addition. it counters Earthling (also a mid-range ship), and other than that, it's somewhat decent against Androsynth. It doesn't have much utility beyond those two battles.
And while you're right to say that Androsynth is very powerful, you're actually underselling it a bit. Androsynth is so powerful that it's practically a necessity for every fleet, easily the best ship in the game.
The point value you got for Ur-Quan is hilarious, and still way too high. Its actual value is closer to 18. Ur-Quan seems a lot more effective than it really is against AI simply because AI doesn't do anything to play around Ur-Quan's obvious flaws. While there aren't true counters against it, it's also just in general quite easy to beat with most ships (Orz, Chenjesu, Mmrnhmhrm, Slylandro, Pkunk, and Supox come to mind) while still being weak to Utwig.
When all is said and done, it's definitely been made clear to me by this point that the best SC2 players out there need entirely different balancing from both what the game shipped with and the results I ended up with here; The results I ended up with here seem better matched up with players of average or above-average skill and I would've clarified that in the video had I realized some of the expert tactics got way deeper than I could present. :P
@@Pixelmusement Fair enough! I do think your representation is quite inaccurate even at that level, though. In particular unskilled players are unlikely to play Slylandro, Umgah, Ilwrath, or Yehat effectively (with Yehat's effectiveness being especially exaggerated by the AI), and even unskilled players will likely be much more effective with Kohr-Ah, VUX, and Supox than the AI is. Other than the fast ships, I think the only ships whose power is likely to seem significantly different for unskilled players is Chmmr, simply because the tactics you have to use against it are somewhat specialized.
You should try UQM's net-melee, it's great. 🙂 I don't know how they organize it these days (or how much) but I do know a form of the Star Control Discussion Board still exists. Maybe try some of the tactics I mentioned in my other post against some friends. It's really a lot of fun when you have two players who know all the tactics, and with a few exceptions, it's very well-balanced. The rules we always used back in the day on #uqm-arena were 200 points max, no duplicate ships, faster ship must take initiative to avoid drawing out fights. Cheers!
its not supposed to be balanced lol
I enjoyed watching this video and liked it, but I can't resist mentioning that if this approach is ever used for multiplayer balancing, giving balance decisions to flawed 90's ai, is a bad idea.
In fact balancing based on any ai is a *terrible* , and I mean *terribly* *terribly* terrible approach to game balance / game design.
Don't worry, I'm not criticising this video, I know it's not primarily focusing on actual game balance, but the introduction of the video at the start makes it sound a lot like it's trying to 'fix' the game balance. In fact it specifically mentions player vs player balance, and questions the original balance, but then uses ai to reach far worse balance conclusions! You say 'just how well balanced was it? Let's get down to it', but then do a far worse job of balancing it! You go on to mention that the ai 'has some shortcomings, but doubtful enough to vastly misrepresent balance'. How wrong you were! In fact I'd say that when you originally mentioned you and your friend banning the avatar, and you concluded 'did I just suck', I'd argue its more likely than the game balance being bad, given how much worse your conclusions would have made it.
As someone who has both been in charge of multiplayer balance in games, and someone who played star control at a competitive-enough level (ie. I did play in tournaments) , I can safely say this video was as disastrous an approach as I suspected.
Believe it or not, Star control had a slightly healthy online multiplayer community back in the day. It has also been balanced properly, rather than this disastrous ai approach.
Spoiler: the thraddash torch is by far the most powerful ship, not using its thruster, but rather its pea-shooter main gun for damage. The fact that this video gives it the joint-lowest score of 4 just demonstrates how bad this 'balancing via maths and ai' approach actually is. The (4 point here!) thraddash torch was actually the only ship regularly banned in un-modded competitive play.
Also of interest: *tons* of ships have a favourable match up against the avatar, to the point where it was sometimes considered the worst of the expensive ships (though more often that was the dreadnought).
The mauler is an essential part of most competitive player's lineups, rarely excluded from any team, due to its ability to counter many top ships including the avatar and the jugger.
Verdict : don't ever try to balance anything based on ai performance. You will end up with a horribly broken multiplayer component to your game. Game balance involves far more than just watching 16 hours of bad ai fighting one another, and even with humans there are huge implications of counterplay and unit selection above and beyond costs, which themselves are terribly assigned here.
16 hours of watching ai is, effectively, 16 hours of completely wasted time in terms of balance and game design.
At least it's only 16 hours though! Less than a day of what is, ultimately, a completely wasted effort.