The Lunatic Club

[Interceptor]

Oh my bad. Actually what needs to happen us that she gets a DS with Virion so she can level up. I have a back-up save in which exactly that happens (and she gets a great level up too, the only stats that don't go up are Strength and Luck). So I got saved by luck even though I'm a bad player.

As long as the guy dies and you don't ruin your positioning for other moves, that's all you need. Whatever it takes, that's my motto.

I'm sure you've already been told this many times, but this clear is really amazing. I was impressed by your Luna+ walkthrough when I found it (that was why I started lurking here), but you've outdone yourself! How long did it take to figure out?

Glad to hear it; I mostly just write for myself, but it's always cool if people can benefit from whatever I figure out. It was nice to finally be able to kick "Shepherds" to the curb, since it was the last piece of the puzzle after the Water Trick for making earlygame L+ consistent.

It probably took about two weeks from start to finish, playing on and off, and I have no idea how many dozens of resets that involved. The process was sort of iterative, in that I'd try to solve a particular mini-problem with what I had available, until I either figured it out or ran out of options. If I succeeded, I'd see how far I could take it. There were a lot of dead ends that I couldn't find a way to make work (initially I did some stuff on the right hand side), or were too reliant on certain enemy skills or movements (which I was trying to avoid), so those went into the scrap heap.

There's probably a comparable way that doesn't require SPD asset or level-up rigging, but I'll leave that for some other masochist to figure out.

[Czar_Yoshi]

There's probably a comparable way that doesn't require SPD asset or level-up rigging, but I'll leave that for some other masochist to figure out.

I'm more interested in getting a slightly more consistent method for Cht.3 (or at least one that doesn't involve a ton of winging it) to go along with this. +Spd is a pretty good postgame asset, so I don't see too much of a reason for an alternate Cht.2 clear for anything else (except maybe Def if you're going for the whole game without grinding).

It'll involve a lot more rigging skills (thanks, PavGis) but should be at least semi-possible. I think. Maybe.

Edited August 27, 2014 by Czar_Yoshi

[Interceptor]

I'm more interested in getting a slightly more consistent method for Cht.3 (or at least one that doesn't involve a ton of winging it) to go along with this. +Spd is a pretty good postgame asset, so I don't see too much of a reason for an alternate Cht.2 clear for anything else (except maybe Def if you're going for the whole game without grinding).

It'll involve a lot more rigging skills (thanks, PavGis) but should be at least semi-possible. I think. Maybe.

I've thought about it, but I can't even wrap my brain around the hellscape of a flowchart that this would require.

The advantage of the P-2 stuff is that there are no defensive bonus skills... as you point out, PavGis brings an offensive-based strategy to a screeching halt if the wrong units on the LHS start with them. You also have a wider range of potential stats on Robin, and the possibility of either an epic or a fail level-up on Frederick, which changes what's possible.

At least if you figure out THAT part, the rest of the map is academic. The corridor turtle is really effective and pretty resilient against a variety of skills distributions. Maybe I'll take a swing at it some day.

[Bloo]

I've thought about it, but I can't even wrap my brain around the hellscape of a flowchart that this would require.

The advantage of the P-2 stuff is that there are no defensive bonus skills... as you point out, PavGis brings an offensive-based strategy to a screeching halt if the wrong units on the LHS start with them. You also have a wider range of potential stats on Robin, and the possibility of either an epic or a fail level-up on Frederick, which changes what's possible.

At least if you figure out THAT part, the rest of the map is academic. The corridor turtle is really effective and pretty resilient against a variety of skills distributions. Maybe I'll take a swing at it some day.

It would be nice if there was at least a guide saying what enemy skills to avoid and reset for (similar to resetting for the first Myrm on the prologue if he has Vantage+ and Luna+) to give people trying to come up with their own strategies the best chance.

[Irysa]

The inconsistent nature of the AI in this game is beginning to annoy me. I've been experimenting with CH2 Lunatic for a while now on a different file, and I cannot for the life of me figure out what messes with the enemy turn order/movement so much at the start.

What I want

What happens randomly in testing

The enemy stats don't variate from what I can see; Abilities do but not for the soldiers. I thought Fred or Chrom's hp might affect it but it doesn't seem to, I've had Fred at 0 hits or 3 hits and still had the soldier full move down instead, regardless of whether Chrom gets hit or not. Anyone more versed on the AI in this game have any input? To be clear, I've got plenty of other ways of beating this map, I'm not stuck on it. I'm just messing around to understand the game better.

Edited August 28, 2014 by Irysa

[XeKr]

I suspect there's simply some randomization thrown in, wrt to the pathing or whatever. The AI will follow some of the obvious trends like ruthlessly targeting your squishy units but it's not as predictable/manipulable.

This is most obvious if you play a map where the enemy rushes you, end turn immediately without moving anyone and you'll see the enemy adopt different formations. Can be seen in the first wave of prologue too iirc.

[Interceptor]

It's almost definitely just randomization, not just attack priorities.

One of the chronic issues in the reliable L+ Shepherds clear, is that Turn 2 sometimes finds one of the Barbarians getting in the way of the other. There's room for them both to get in a swing (one on Robin, one on Fred), but occasionally the close one will attack Robin, and the second one can't reach Fred as a result.

It's not a question of stats: nobody levels-up during this sequence. It's also not a question of health (everyone is at full, because Fred heals himself), or equipped weapons (always the same), or terrain (likewise). Gamble varies, but there's no causal relationship between Gamble and targeting.

Sometimes you can manipulate this by forcing enemies to go through terrain that slows them down (which effectively makes them closer to some units than others), but it's not consistent.

It would be nice if there was at least a guide saying what enemy skills to avoid and reset for (similar to resetting for the first Myrm on the prologue if he has Vantage+ and Luna+) to give people trying to come up with their own strategies the best chance.

There aren't many hard-and-fast rules that I think you could come up with. Almost definitely you want to avoid Counter on the Soldier, and probably Aegis+ on the Knight, but that's about it other than "minimize Luna+".

Edited August 28, 2014 by Interceptor

[Irysa]

Ah, dissapointing. Thanks for the quick feedback though. In FE12 I found I could basically near 99% of the time get the AI to move in the way I wanted them to via some measure of positioning or low hp bait in some general area. This was a huge part of what made Lunatic of that game feel very fair and tolerable, so it's sad to see that they regressed like that.

Gamble does appear to make enemies break turn order just like FE12 if they can get a chance of kill via crit however, but the positioning I have in my screenshots has no influence on that whatsoever.

Edited August 28, 2014 by Irysa

[XeKr]

It’s a tradeoff that can be framed a few ways. FE12 is well known for being extremely predictable such that you can adopt silly formations that feel like they shouldn’t work but for the quirk of the AI to attack in a specific manner (but certainly more tactical, in a sense). Whereas you’re expected to adapt a bit more in FE13 (or alternatively, enjoy artificially inflated replay value).

Both are understandable, but I prefer less rigid flowcharts myself. (edit: incidentally, I still don't really like L+ for other reasons. >_>)

Edited August 28, 2014 by XeKr

[Irysa]

Well generally speaking, a bit of randomness adds to variety for sure. But within the context of situations that literally force a reset when things don't line up, it's just more time wasted spent on things you already grasp. Situations such as those have a low amount of flexability to compensate for such randomness, "adapting" isn't always possible. I mean sure, I can accept things like calculating for a miss, getting hit on low % hitrates, but inconsistent movement from immediately YOLO-ing enemies at a Lunatic level difficulty is hardly comparable. It has more merit on the lower difficulty settings than the higher ones really.

Obviously that being said I've not actually played past Chapter 6 yet, so I can't really comment on whether it's a serious problem or just this map in particular, even FE12 CH2 is a little erratic sometimes. Although I mostly put that down to enemy stat variance, which appears to be gone in this game, hurrah! But then they have random skill distributions even on lunatic, boo.

Edited August 28, 2014 by Irysa

[XeKr]

With clever positioning/decision-making, you’re more likely to die to say a 30 hit/10 crit gamble crit (and not that likely) or w/e than because the AI did something unexpected movement wise. That’s the consequence of planning for contingencies; it adds depth.

All plans are subject to RNG in FE (gba games don exits), but you can design something as robust as possible. RNG will most probably screw you elsewhere before pathing as different formations can mostly be played around, unlike “oh look, a 2% chance to do an ohko crit.” So that kind of RNG in AI gives the impression of more interactivity and player control to me, as you’re more obviously forced to adopt a different formation in return. It’s arguably more challenging to deal with enemies approaching from different angles and such than if they just have differing amounts of health (from misses, etc). But, crucially, still within the player control.

Admittedly, pathing will screw up things if one is adamant about ltc/rta/etc (but so will other things more significantly, ie dual strike/guard). Otherwise, I can’t think of where the ultimate success rate is significantly dependent on enemy pathing, and outside of player control (besides C2 in L+ which I think is well-documented).

Edited August 28, 2014 by XeKr

[Irysa]

I don't actually disagree with any of that, but I also fail to really see how it impacts on what I was very specifically targetting in my post. I already admitted I have no insight on whether or not it is actually a problem at any other point in the game, but I discussed Ch2 as an example where pathing randomness can result in situations that are actually NOT workable around.

In Ch1 it's largely a non issue because the enemies do not start close enough, you have enough flexability to lure them around and have a general strategy that can easily compensate for an archer moving a little closer than you expected. Ch2 is just notably bad because even a tiny bit of variance at the start likely means that somebody is going to die. Now, saying that, I also actually came up with one or two other ideas for doing it that, insofar as I've tested, are pretty much immune to that problem (Fred/Virion instead of Fred/Avatar, lets Chrom and Avatar snatch stragglers instead, so I don't get rammed into one corner.). So, I would admit it isn't like, a game crippling flaw or anything of the sort.

Perhaps a better way of putting it, is that different types of map design benefit differently from those different approaches to AI behaviour (observation of the century). And that in the case of maps where you get very little initial breathing room such as Ch2, random AI behaviour is workable, but somewhat undesirable. By the same thread, it's one thing to give strong enemies crit weapons and bonuses when you have the ability to do something about it, it's different when they just randomly dump crit skills onto early enemies where you're going to be relying on a unit who has somewhat low luck to weaken or kill them. See where I'm coming from?

Edited August 28, 2014 by Irysa

[XeKr]

I discussed Ch2 as an example where pathing randomness can result in situations that are actually NOT workable around.

I largely agree with you; my point is I don't think the above quote is true (to the extent of randomness that is generally considered "acceptable" in FE). In Lunatic at least.

Again, the only place where I would consider that issue significant is C2 in Lunatic+ and there are clearly other factors besides AI there as well (Luna+ density, ofc). I haven't played C3 in a while, but for now I'll maintain it felt reliably doable whatever skill configuration and without necessarily an exceptional Robin.

Edited August 28, 2014 by XeKr

[Interceptor]

It'll involve a lot more rigging skills (thanks, PavGis) but should be at least semi-possible. I think. Maybe.

It would be nice if there was at least a guide saying what enemy skills to avoid and reset for (similar to resetting for the first Myrm on the prologue if he has Vantage+ and Luna+) to give people trying to come up with their own strategies the best chance.

Alright, I took a swing at this. Here's something quick and dirty that has decent reliability once you get the skill distribution that you need. As a special bonus, it's basically generic and will work with any asset/flaw.

Requirements:

• Thunder with at least 2-3 uses left
• Knight cannot have Aegis+
• Soldier cannot have Pavise+
• Archer (near) cannot have Luna+
• Archer (far) cannot have Luna+... maybe. This might be OK.

The "eureka" moment here was a two-parter. I abused two things to make it work. First, Robin participates in three combats against two enemies in the first Turn (like a boss). Second, Miriel is only exposed to one Archer, which she barely can survive, with the help of cock-block placement and a +2 DEF support. Not even Pass will get the second Archer in range.

So assuming equal skill distribution, there's a 5/7 chance that any one enemy will be missing a particular skill. So with three or four enemies that each need a specific skill gone, it's either a 36% or 26% chance that you have a map you can work with. Statisticians, feel free to correct my math. There's no DS rigging here, although there's one kinda-chancy hit. It's super-easy to reset for skills, because you can save on the formation screen and there is no dialog or cutscene to blast through.

Anyway, set up your shit like this:

The pairings and placements are actually important, although Lissa and Chrom can do whatever. Miriel and Robin need to split up the tomes, and it doesn't matter who gets which. Steps:

• Move Miriel/Robin 4W, attack the Knight. Thanks to Pair-up, Miriel doubles and does a ton of damage.
• Move Virion/Kellam 4W, Switch, Transfer Virion for Robin, and attack the Soldier with a Javelin for 8 damage. Robin gives Kellam the +STR/HIT he needs, and Virion lets Miriel survive an Archer on Enemy Phase.
• Move Fred/Vaike 6W, kill the Soldier with Silver Lance. Vaike's +4 STR to Fred guarantees the finisher (8 + 25 = 33... the Soldier's HP).
• Move Sully 1W of Kellam, Transfer Robin, Switch, finish off the Knight at 2-range with a tome.
• Fly Sumia west and Pair with Kellam. This gives him enough +SPD to avoid being doubled.
• Note that the units are placed such that the far Archer can't get to Miriel, not even with Pass. Take that, asshole.
• Move Chrom and Lissa a little to the left while staying out of range of the Archers. Turns ends.
• Near Archer will move next to Kellam and take a shot at Miriel, bringing her to 1 HP. Far Archer will attack someone else; Robin and Kellam will retaliate if it's either of them. He never seems to go for Fred for some reason.

After that, clean up the Archers (easy, since both are probably injured). The guys on the right hand side haven't even made it halfway over to you yet, so you have approximately three f'n years to open the door, kill the Knight, heal people, and set up a chokepoint on the north side of the corridor. Chapter is academic at this point.

So once you get the skill distribution, the toughest move is the Javelin from Kellam. It's only about 85% True Hit. It might not be strictly necessary to avoid Luna+ on the far Archer, because it seems like they won't ever double-team someone (the close one always goes for Miriel in every attempt I have tried).

Fin.

[Irysa]

I largely agree with you; my point is I don't think the above quote is true (to the extent of randomness that is generally considered "acceptable" in FE). In Lunatic at least.

Again, the only place where I would consider that issue significant is C2 in Lunatic+ and there are clearly other factors besides AI there as well (Luna+ density, ofc). I haven't played C3 in a while, but for now I'll maintain it felt reliably doable whatever skill configuration and without necessarily an exceptional Robin.

My phrasing could have been better, I apologise. Perhaps more stress should have been put on the "can", obviously the correct choice of moves results in very consistent results; I've already done that myself.

That being said, there are quite a few misleading first turn moves that on the whole, would appear to be reasonable choices (like the one I screenshotted) that can just result in an instant reset. The chance of something like that happening is rather a lot higher than the chance to get RNG screwed by attack calculations, which we all generally accept as part of the game. As such, in my eyes, this is going a bit over the line. I don't mean to infer that the randomness in C2 is an issue that significantly distorts successful clears on regular Lunatic; only that I consider it to be unreasonable from a design overview, in the same manner that I consider chucking something like gambit on early enemies randomly to be unreasonable. Or, for another example, the same way FE11 H5 earlygame is pretty unreasonable because of how much Javelin and Handaxe chip at terrible hitrates needs to connect to get through it.

This entire point just stems from the fact that I felt that FE12 H3 was incredibly fair by comparison, which I'll accept, is a totally subjective observation, and you rightfully pointed out the negatives of it's rigidity. Having one or two unfair scenarios within the game on the whole is not beyond most games in the series, if that's what you're getting at, but I don't think that gives them all free passes.

And yes, I know you can totally negate the Prologue Boss Gambit w/bonuses, that's totally different and actually pretty cool.

EDIT: Gosh, now I've gone and undermined a great post by Interceptor with this irrelevant nonsense.

Edited August 28, 2014 by Irysa

[Interceptor]

Otherwise, I can’t think of where the ultimate success rate is significantly dependent on enemy pathing, and outside of player control (besides C2 in L+ which I think is well-documented).

Even in the case of L+ Chapter 2 it's not really that bad; if the Barbarians block each other, you can still salvage the run: your worst case scenario is a Barb at full health, which isn't fatal. Once the enemies are in that neighborhood, you have complete control, because you can kite them counter-clockwise around the mountain range. It's more difficult, but it's not impossible. You are still going to kill the Merc and possibly the wounded Barb on Enemy Phase, leaving only three units to deal with and a huge part of the map open for maneuvering.

I don't actually disagree with any of that, but I also fail to really see how it impacts on what I was very specifically targetting in my post. I already admitted I have no insight on whether or not it is actually a problem at any other point in the game, but I discussed Ch2 as an example where pathing randomness can result in situations that are actually NOT workable around.

Pathing randomness can also result in bailing you out of a situation that you had no business surviving in the first place. I personally prefer the ad-hoc nature of dealing with random movements, because that's something that takes skill to deal with (whereas perfectly predictable movements can be defeated just by memorization and iteration).

FWIW: Gamble is a double-edged sword. It does give Barbarians listed crit on a lot of your guys, but it also cuts their HIT rates by 5%, which can be significant when they have WTD.

EDIT: Gosh, now I've gone and undermined a great post by Interceptor with this irrelevant nonsense.

It's not a big deal. That was just a rough cut; I still need to re-format it, adjust for any feedback that people might have (particularly about that top Archer), and then post the final version in my Chapter 2 thread.

[isetrh]

I beat Chapter 2 using Interceptor's method! Going to try Chapter 3 bow.

[Interceptor]

Alright, I've done a few more trials of Chapter 3; I'm pretty confident that it doesn't matter what skills the far Archer starts with.

I had a couple of runs where the far Archer had Luna+, and the near one didn't. Kellam with a Sumia support takes 8 damage from a normal Archer, and 15 damage from Luna+. Since Kellam only has 21 HP, they could have comboed him for a kill, but the AI apparently doesn't recognize that fact, because the near Archer will consistently go for Miriel every time. There was even one run where I put Chrom in range of the Archer, who then completely ignored him (despite the fact that Chrom couldn't even counter-attack).

So you only need to rig three skills for this strategy, which is honestly pretty easy to do. Even when the Soldier starts with something nasty like Luna+/Vantage+, he won't do enough damage to Fred for the Archers to be a threat.

[isetrh]

I beat the lower half of Chapter 3 using Interceptor's method and made a back up save. The entire upper right has Counter. No Luna+ though. I don't know how to deal with this. My Avatar has 13 Speed, 12 Def, and C supports with Chrom, Frederick, Lissa, and Miriel. Fred has 15 Def. Help?

[Interceptor]

Unequipped FrederickxKellam Paired in a choke point, fire over their heads with tomes/bows. Kill the Archers first if possible, makes it easier (with no Archers, Lissa can easily heal Fred). Use the corridors as your choke point, and fill the area behind you with trash un-Paired units so that enemies with Pass can't make it through.

EDIT: and read my playlog write-up for this Chapter. There's a couple of general tips in there.

Edited August 29, 2014 by Interceptor

[isetrh]

I've been doing most of what you said, though the Archers don't get in range; instead the Hammer guy just kills Fred. I looked at your playlog again (I've actually read through it several times in the past, it's helped me out a lot) and what I got out of it was that I should kite if there's too much Luna+. I'll try this since it seems like the same principle, too much damage to tank. Thanks!

[Tables]

Just curious about something for Luna+. In later chapters, many bosses (and some generic enemies) can have 4 or 5 skills on Lunatic. Which skills get overwritten with the Luna+ skills? Is it random, or is there a pattern to it? I would expect it's the last two listed skills, but I dunno.

[Interceptor]

I've been doing most of what you said, though the Archers don't get in range; instead the Hammer guy just kills Fred. I looked at your playlog again (I've actually read through it several times in the past, it's helped me out a lot) and what I got out of it was that I should kite if there's too much Luna+. I'll try this since it seems like the same principle, too much damage to tank. Thanks!

Oh yes, HammerGuy, I forgot about him. You'll either have to kite or tank him with Robin. Kiting is pretty easy, so I'd recommend that against non-Counter enemies. If you tank with Robin, just have someone else fire over his/her head or take away weapons after attacking and before healing. You sort of need to adjust depending on the sort of make-up you are facing.

Just curious about something for Luna+. In later chapters, many bosses (and some generic enemies) can have 4 or 5 skills on Lunatic. Which skills get overwritten with the Luna+ skills? Is it random, or is there a pattern to it? I would expect it's the last two listed skills, but I dunno.

It's not entirely random, and I didn't analyze the pattern extensively, but the "intelligence" of it is limited.

For example, Walhart will always get the + versions of Pavise, Aegis, and Luna. He keeps Conquest, but he loses Rightful King in place of a random L+ skill. Grima only swaps Pavise for Pavise+, and keeps the other four skills.

On the other hand, L+ will also do things like give Hawkeye to a skill-stuffed unit with HIT +20, and notably will also give both Pavise and Pavise+ to Cervantes in Ch16. I've also seen instances where a Swordmaster had Vantage replaced with something entirely different, rather than having it just be upgraded to Vantage+.

So there's that. I don't have a ton of data on it WRT generics, but it appears that they possibly intervened with exceptions in certain circumstances.

Edited August 29, 2014 by Interceptor

[Tables]

Can someone verify the reinforcements for chapter 16 (Lunatic/+ of course) match up with what the wiki says? I thought there were only four Falcknights on turn 5, and a second wave of Falcoknights around turn 7 or 8.

Also, I'm missing chapter 19's reinforcement data. IIRC the reinforcements are turns 4-8, with 8 units on each of turns 4-7 and 16 units on turn 8 (total 48 reinforcements), but I don't know specifics. If someone is approaching that part of the game and can list down what spawns each turn, that'd be great.

Edited August 29, 2014 by Tables

[Interceptor]

^ Are you looking at the JP wiki as well? The data there was pretty good whenever I had to look something up for reinforcement data in my L+ run.

http://www18.atwiki.jp/fireemblem3ds/