Advance Wars Days of Ruin/Dark Conflict (DS)

2008-07-12 07:46:32 am

DS Dictator

I was watching Fire Emblem: Radiant Dawn speed run and I noticed that you can save halfway through the mission.
I thought with Advance Wars I can save part way through to abuse luck (little extra damage).
The newest one seems easier to speed run with because:

+Lack of character choice
This Advance Wars game you can't choose a character you want to play as like in the previous titles, therefore there is less worry to disscuss who is the most effective CO on the map.

+Good No. of Days target
Theres a website where it shows the list of missions in the least number of Days/turns as possible.
However In-Game time would be number of Days whilst Real time would be the time it took to complete the mission.

+No unfortunate CO Power abuse.
In the previous games CO power raises faster to the opponent that loses HP on Units. Usually CPU's 3-5 bombers getting destroyed = Super CO power. This time around you got to deal at least 60HP in the CO Zone to activate the power.

+Promoted Unit boosts
When units destroys other units they level up. Have a single unit to destroy 3 other units to raise to the max level.
Direct units are more likely to level up faster because:
The AI usually send in weak Bazookas to try and deal 1 Hp onto stronger units.
The indirect units usually take longer but are easier to level up with because the indirect range, (finishing off road blockers that you can use the remaining direct units to deal damage on the other CPU units).

Level 1: 105% Attack 100% Def
Level 2: 110% Attack 100% Def
Level Ace/Veteran: 120% Attack 120% Def

-Ally CPU
Unfortunately some missions you have a ally team and you can't control them, so luck from the CPU can either be both an advantage or even an disadvantage depending on the situation.

Thread title:

I was watching Fire Emblem: Radiant Dawn speed run and I noticed that you can save halfway through the mission.
I thought with Advance Wars I can save part way through to abuse luck (little extra damage).
The newest one seems easier to speed run with because:

+Lack of character choice 
This Advance Wars game you can't choose a character you want to play as like in the previous titles, therefore there is less worry to disscuss who is the most effective CO on the map.

+Good No. of Days target
Theres a website where it shows the list of missions in the least number of Days/turns as possible.
However In-Game time would be number of Days whilst Real time would be the time it took to complete the mission.

+No unfortunate CO Power abuse.
In the previous games CO power raises faster to the opponent that loses HP on Units. Usually CPU's 3-5 bombers getting destroyed = Super CO power. This time around you got to deal at least 60HP in the CO Zone to activate the power.

+Promoted Unit boosts
When units destroys other units they level up. Have a single unit to destroy 3 other units to raise to the max level.
Direct units are more likely to level up faster because:
The AI usually send in weak Bazookas  to try and deal 1 Hp onto stronger units. 
The indirect units usually take longer but are easier to level up with because the indirect range, (finishing off road blockers that you can use the remaining direct units to deal damage on the other CPU units).

Level 1:                105% Attack 100% Def
Level 2:                110% Attack 100% Def
Level Ace/Veteran: 120% Attack 120% Def

-Ally CPU
Unfortunately some missions you have a ally team and you can't control them, so luck from the CPU can either be both an advantage or even an disadvantage depending on the situation.

DTaeKim

2008-07-12 10:30:32 pm

I consider myself one of the best players at this game, and honestly, I would have a difficult time justifying a speed run for this game.

Greenalink

2008-07-13 04:14:00 am

DS Dictator

Most of the missions you can just Capture the HQ for an instant win. The mission against Carter/Forsythe is hard to capture the HQ, the same goes for Waylon/Finn's match. There is one Youtuber that has found the most effective tactics in the last set of missions.

DTaeKim

2008-07-14 12:00:35 pm

The record on C14 is 9 days.

The record on C20 is 10 days.

Those Youtube videos have been obsoleted since they were posted.

Greenalink

2008-07-14 01:12:43 pm

DS Dictator

C-14 in 9 days
Wow seems cool, I'll need to watch it if it's good for a real time setup.

Terri

2008-07-16 09:16:01 am

I love YaBB 1G - SP1!

I'd prefer to see a run on a other Advance Wars title. Maybe DoR is the easiest title to run in the series, but it's the worst title, too.

yoshmaster5

2008-07-16 09:35:09 am

Also tenkiforecast

I just started playing this yesterday, and... well... Undecided

I don't think this game is really 'speedrun-able'
It's just how the game is set up, and the way you have to play. While this game is very different than the previous games, it still doesn't make much difference to me. While I can be disproved, this is a bit of a tough game to run.

Wolfz

2008-07-16 05:07:26 pm

Let's go 2k11!

If someone actually decides to do a run of this game I would be interested. I bought this game a week after it came out and I still can't get past C21 =P

It will take a lot of patience and luck, but you should be able to run this game.

And if anyone could give me a good strategy for C21 that would be great =D

DTaeKim

2008-07-16 05:37:12 pm

Quote from Terri:

I'd prefer to see a run on a other Advance Wars title. Maybe DoR is the easiest title to run in the series, but it's the worst title, too.

I would disagree with you, but this is not the place for that.

mikwuyma

2008-07-16 08:25:16 pm

My feelings on The Demon Rush

I always thought this was a bad series to speed run (though great to play), but now I'm wondering what makes this series so much different from Fire Emblem speedrunning-wise (I've never played a FE game).

Greenalink

2008-07-17 02:59:39 am

DS Dictator

Yeah I've been trying to speed run through things but what loses loads of time are these things:
Thinking about your unit(s)' next moves (I would require a huge memory bank in my brain to do that).
Luck from both the A.I & extra damage.
I only saw a little bit of Fire Emblem and it's notable for saying that when you lose a soldier, it's gone for good. In Advance Wars there are bases where you can re-produce units.

I think that making this thread must have been my dumbest move yet Sad

in SDA.

DTaeKim

2008-07-18 08:14:11 pm

The problem doesn't lie in the movement or luck.

The problem is that once you determine the fastest possible route, it's set.

You can follow a walkthrough on GameFAQs and get a speed run that way.

Birkal

2008-07-18 08:48:28 pm

Handsome Squidward is handsome.

It's a great game, but I don't see it much as a speed run-able game.

There just doesn't seem to be enough variables.

GipFace

2015-02-14 01:30:16 pm

Huge necro. I am attempting a speedrun of this. 100% disagree with everyone's argument that determining the fastest possible route is boring. Isn't that what all speedruns are about?

There's a fine balance between going for the fastest possible route and getting RNG-screwed, versus taking a slightly slower but more reliable route. The day-to-day walkthroughs around the internet focus on highest score, and many of them require mass amounts of RNG abuse. There is also the challenge of actually memorizing all the strategies and executing them successfully.

There are 26 campaign missions, C1 to C26. I've routed through C21 thus far, and I've filmed a C1-C19 run in 31:20: http://www.twitch.tv/gipstream2/c/6097778
The C20-C26 splits were from a previous speedrunner (http://www.twitch.tv/deln/c/4288224). As you can see I've obliterated his times so far. It may be possible to speedrun this game in under an hour, and it's entertaining the whole way through.

ais523

2015-02-14 08:08:03 pm

I'd say a speedrun route is probably different from a highscore route. (Also, I thought your name looked familiar.) When you're turning on battle animations to know that you've done enough percentage damage to be able to finish the map, you know you have a speedrun-unsafe strategy. Are you aiming for IL or single segment? Those seem like the two most sensible categories for this game. (Segmenting inside maps might lead to a faster time, but I doubt anyone wants to watch that.)

A good first start might be to find the fastest strategies on each campaign map that are 100% guaranteed to work, then to work out how much gambling can be done on top of that (and what the backup strats are).

mike89

2015-02-15 04:38:47 am

SEGA Junkie

For some reason Twitch highlights aren't working for me right now, but I'm definitely going to give that a look pretty soon! I love this game and I'm kind of surprised at how quick a speedrun of it turns out to be. (Especially considering the best runs I've found of Dual Strike are close to two hours for around the same number of campaigns!)

GipFace

Edit history:

GipFace: 2015-02-18 05:32:03 am

GipFace: 2015-02-15 07:59:23 am

2015-02-15 07:57:22 am

My ultimate goal is to have this game featured seriously at a GDQ, thus a single-segment run. Advance Wars and Fire Emblem are produced by the same developer (Intelligent Systems), and AW fans have felt a bit slighted at times because FE gets all the attention. I feel that AW Days of Ruin (calling it AW4 because it has THREE sub-names depending on locale) is the best game out of the franchise to speedrun. To a viewer who has never experienced AW or FE, this is an easy game to follow. Unlike FE, there's barely any menuing, equipment, or skills, and it's visually obvious and intuitive which units are good against which. Tanks beat soldiers, bombers beat land units, etc.

Going for high scores is definitely different from speedrunning. Three things determine your score: speed, power, and technique. In AW4, the power score is determined by total damage divided by number of attacks, with KOs adding a bonus. This leads to the highest scores involving allowing the enemy to attack you first, sponging their attacks, dealing counterattack damage, then 1HKOing their damaged units the day after. This is slower than simply attacking first. The technique score is roughly determined by how many units are lost in relation to the enemy army. (Building more units does not help, unlike in AW1-3.) In certain missions, I avoid touching a number of units for speed purposes and don't care whether they survive, so technique takes a beating.

As AW4 uses the exact same engine as FE Shadow Dragon, RNG manipulation is the same. You just select a random unit and then path it diagonally 2 tiles with a stylus tap. However, this is much less prone to abuse than in FE because the RNG advances after every attack and does not reset. However, if you need the AI to move units a certain way, this is where RNG manipulation can come in.

While the speedrun doesn't go for highest score, it still gets all S-ranks, and the scores are generally higher than what most Let's Plays get on YouTube due to maxing or almost-maxing of the speed score. Some saving is inevitable in order to get favorable outcomes: many attacks deal X5% but the game stores HP in increments of 10%, so sometimes winning the luck roll is important. I have tried to minimize this wherever possible because I know people hate seeing FE speedrunners attempt a section over and over.

I've really enjoyed routing this so far. The highlights are C5, C7, and C10, which are all spectacular suicide speed missions, and C14, a difficult mission done only in 6 days. The suicide speed missions are what gives me a tremendous time advantage over the other speedrun attempt; at the C19 point, I am over an hour ahead of his pace. However, routing the last six missions, C20-C26, is a huge challenge. These are missions that normally take people over an hour each to clear, but I have just over 28 minutes total if I want to complete the speedrun in under an hour.

For in-depth planning and discussion, there is a thread on Wars World News, probably the only AW community hub still alive. So far, it only consists of two people, but if you're really interested and would like to help, you're more than welcome: http://forums.warsworldnews.com/viewtopic.php?f=28&t=13649

Thanks for all your support!

My ultimate goal is to have this game featured seriously at a GDQ, thus a single-segment run. Advance Wars and Fire Emblem are produced by the same developer (Intelligent Systems), and AW fans have felt a bit slighted at times because FE gets all the attention. I feel that AW Days of Ruin (calling it AW4 because it has THREE sub-names depending on locale) is the best game out of the franchise to speedrun. To a viewer who has never experienced AW or FE, this is an easy game to follow. Unlike FE, there's barely any menuing, equipment, or skills, and it's visually obvious and intuitive which units are good against which. Tanks beat soldiers, bombers beat land units, etc.

Going for high scores is definitely different from speedrunning. Three things determine your score: speed, power, and technique. In AW4, the power score is determined by total damage divided by number of attacks, with KOs adding a bonus. This leads to the highest scores involving allowing the enemy to attack you first, sponging their attacks, dealing counterattack damage, then 1HKOing their damaged units the day after. This is slower than simply attacking first. The technique score is roughly determined by how many units are lost in relation to the enemy army. (Building more units does not help, unlike in AW1-3.) In certain missions, I avoid touching a number of units for speed purposes and don't care whether they survive, so technique takes a beating.

As AW4 uses the exact same engine as FE Shadow Dragon, RNG manipulation is the same. You just select a random unit and then path it diagonally 2 tiles with a stylus tap. However, this is much less prone to abuse than in FE because the RNG advances after every attack and does not reset. However, if you need the AI to move units a certain way, this is where RNG manipulation can come in.

While the speedrun doesn't go for highest score, it still gets all S-ranks, and the scores are generally higher than what most Let's Plays get on YouTube due to maxing or almost-maxing of the speed score. Some saving is inevitable in order to get favorable outcomes: many attacks deal X5% but the game stores HP in increments of 10%, so sometimes winning the luck roll is important. I have tried to minimize this wherever possible because I know people hate seeing FE speedrunners attempt a section over and over.

I've really enjoyed routing this so far. The highlights are C5, C7, and C10, which are all spectacular suicide speed missions, and C14, a difficult mission done only in 6 days. The suicide speed missions are what gives me a tremendous time advantage over the other speedrun attempt; at the C19 point, I am over an hour ahead of his pace. However, routing the last six missions, C20-C26, is a huge challenge. These are missions that normally take people over an hour each to clear, but I have just over 28 minutes total if I want to complete the speedrun in under an hour.

For in-depth planning and discussion, there is a thread on Wars World News, probably the only AW community hub still alive. So far, it only consists of two people, but if you're really interested and would like to help, you're more than welcome: http://forums.warsworldnews.com/viewtopic.php?f=28&t=13649

Thanks for all your support!

Greenalink

2015-02-15 04:08:49 pm

DS Dictator

Been watching the twitch highlight and awesome stuff btw.

The biggest reason why this game is very quick compared to Dual Strike is because every single mission has pre-deployed units and building CO power works a little different this time around so the A.I can't abuse it as much.

GipFace

2015-02-18 04:17:32 pm

I have now planned out everything except C25. Below are the missions I've had difficulty planning. There are two ways to win in most maps: destroy all enemy units (rout); or capture the enemy headquarters (HQ). HQ is the preferred method almost every time.

C20: Waylon Flies Again
This is unfortunate in that my current strategy require 5 save points. Antiair vs. duster is 75%, so the antiair may deal 7HP or 8HP damage. The AI reacts differently if its dusters are at 2HP or 3HP. Estimated time is 4:30, making it the 2nd-longest mission in the run. I don't know if I can make it less luck-reliant, but it's currently a D10 semi-suicide speed win: units built are 1 each of fighter, duster, t-copter, and antiair.

C22: The Great Owl
1 save point is needed on D2. The problem is that the HQ is located at the NE rockets. You can see that it's heavily defended by a cluster of defensive indirect units. For this one, I've planned a D10 rout. Someone has claimed to do a D9 HQ capture, but I'm unsure as to whether this is actually any faster. Due to the unskippable nuke animation, this is the longest mission in the game, clocking in at an estimated 6:00. Still an improvement over the YouTube video that shows a D14 rout in 16:30. The WR speedrun of 3:40:25 has a C25 split of almost 59 minutes!

C25: Lab Rats (The transparent units appear on D3)
This is a suicide speed mission that has a D9 win. I did it a while ago, but I can't get it to be consistent. It involves rushing the HQ with your air units and ignoring the war tanks. I need help with this one.

C26: Sunrise
DTK/Swifty's D2D strategy requires 3 save points in order to win on D5. Estimated time is 3:30 if you're lucky enough. Currently looking for a more reliable strategy.

---

Playing fast
One thing I have noticed in the other AW speedruns is that they use the d-pad. Using the d-pad is a no-no. The fastest method is a combination of stylus and button presses. This is conceivably a game where it's enjoyable to see the stylus and button presses on camera, so that is how I record it.

On an actual DS/3DS, there is a considerable speed advantage for left-handed players. If you have the stylus on your left hand, that frees your right hand for the important A, B, and START buttons. A righty could possibly mimic this by crossing arms. SELECT is a guaranteed way to bring up the menu regardless of where the cursor is placed, but I've found it to be cumbersome in practice. I'm too used to finding an empty tile, then tap+A+tap to end turn. As I don't own an XL version of a DS/3DS, I'm unsure of whether they result in faster or slower play.

Fun stuff
It's possible to complete AW4 without either side invoking a CO power, a trademark of the franchise. If speedrun at a charity event, invoking a CO power could be a donation incentive. There is an opportunity to use one without disrupting the run too much: C18. There is also an opportunity in C22, though I'd rather use the CO unit as a suicide battering ram there.

C16 can be won without attacking once. A pacifist run results in a B-rank because the power score is 0. This is actually faster than getting the S-rank. Could be a silly donation incentive. However, I attack once to keep the streak of all S-ranks alive.

I have now planned out everything except C25. Below are the missions I've had difficulty planning. There are two ways to win in most maps: destroy all enemy units (rout); or capture the enemy headquarters (HQ). HQ is the preferred method almost every time.

[img]http://www.warsworldnews.com/images/dorc20.png[/img]
[b]C20: Waylon Flies Again[/b]
This is unfortunate in that my current strategy require 5 save points. Antiair vs. duster is 75%, so the antiair may deal 7HP or 8HP damage. The AI reacts differently if its dusters are at 2HP or 3HP. Estimated time is 4:30, making it the 2nd-longest mission in the run. I don't know if I can make it less luck-reliant, but it's currently a D10 semi-suicide speed win: units built are 1 each of fighter, duster, t-copter, and antiair.

[img]http://www.warsworldnews.com/images/dorc22.png[/img]
[b]C22: The Great Owl[/b]
1 save point is needed on D2. The problem is that the HQ is located at the NE rockets. You can see that it's heavily defended by a cluster of defensive indirect units. For this one, I've planned a D10 rout. Someone has claimed to do a D9 HQ capture, but I'm unsure as to whether this is actually any faster. Due to the unskippable nuke animation, this is the longest mission in the game, clocking in at an estimated 6:00. Still an improvement over the YouTube video that shows a D14 rout in 16:30. The WR speedrun of 3:40:25 has a C25 split of almost 59 minutes!

[img]http://www.warsworldnews.com/images/dorc25.png[/img]
[b]C25: Lab Rats (The transparent units appear on D3)[/b]
This is a suicide speed mission that has a D9 win. I did it a while ago, but I can't get it to be consistent. It involves rushing the HQ with your air units and ignoring the war tanks. I need help with this one.

[img]http://www.warsworldnews.com/images/dorc26.png[/img]
[b]C26: Sunrise[/b]
[url=http://www.warsworldnews.com/wp/aw4/campaign-aw4/c26-sunrise/]DTK/Swifty's D2D strategy[/url] requires 3 save points in order to win on D5. Estimated time is 3:30 if you're lucky enough. Currently looking for a more reliable strategy.

---

[b]Playing fast[/b]
One thing I have noticed in the other AW speedruns is that they use the d-pad. Using the d-pad is a no-no. The fastest method is a combination of stylus and button presses. This is conceivably a game where it's enjoyable to see the stylus and button presses on camera, so that is how I record it.

On an actual DS/3DS, there is a considerable speed advantage for left-handed players. If you have the stylus on your left hand, that frees your right hand for the important A, B, and START buttons. A righty could possibly mimic this by crossing arms. SELECT is a guaranteed way to bring up the menu regardless of where the cursor is placed, but I've found it to be cumbersome in practice. I'm too used to finding an empty tile, then tap+A+tap to end turn. As I don't own an XL version of a DS/3DS, I'm unsure of whether they result in faster or slower play.

[b]Fun stuff[/b]
It's possible to complete AW4 without either side invoking a CO power, a trademark of the franchise. If speedrun at a charity event, invoking a CO power could be a donation incentive. There is an opportunity to use one without disrupting the run too much: C18. There is also an opportunity in C22, though I'd rather use the CO unit as a suicide battering ram there.

C16 can be won without attacking once. A pacifist run results in a B-rank because the power score is 0. This is actually faster than getting the S-rank. Could be a silly donation incentive. However, I attack once to keep the streak of all S-ranks alive.

ais523

2015-02-19 07:11:59 pm

My experience is that stylus is much faster than D-pad in AW3, but not by as much in AW4, because they screwed up the stylus inputs somehow. (I always used the stylus in Time Survival in AW3, and often elsewhere too, but it's so frustrating to use in AW4 for reasons I haven't figured out yet.) Note that pressing L automatically selects an unmoved unit; in the maps where you don't care which order your units move in (which is common, but not always the case), this is probably the fastest way to play. (Also, I find that A-up-A, or when using stylus A-tap, is faster than Select because you don't need to move your hands around as much.)

If pacifist is faster, then on a serious speedrun, you should probably go pacifist. The occasional low rank is quite common in an Advance Wars speedrun; I think there are plenty of As in the AW2 TAS, for example. (Pity there are no Cs; getting a C rank takes serious effort in this game.)

Also, I'd say CO powers aren't much of a staple of AW4. They're more like an emergency panic button in this game, or something for finishing off when you're winning, than they are a part of normal play. In a speedrun, you're unlikely to even charge one, and if you do, in most cases you want to keep it around for the radius, unless you have a very specific use for the CO power.

The different translations also make the game quite hard to watch. (In the UK translation, for example, the units have different names; some are the same names as in the US version but assigned to different units, e.g. Artillery (US AWDoR) is called Mech Gun, or Mech for short (UK AWDC), which makes things very confusing.) I'd recommend you use the US version, because the localization is better in most respects (I prefer the CO and mission names in the UK version, but nothing else).

One final important thing is to ensure that there are no points where you need to turn battle animations on to see exactly how much fractional damage was dealt. That's likely to cost a lot of time, and is probably going to be avoidable (especially as you typically aren't going for rout).

All this said, I'm not convinced that rout vs. HQ capture is as unbalanced as it seems. Playing for a fast rout can reduce the amount of time the opponent spends moving units around, given that the actual act of an HQ capture takes two turns.

My experience is that stylus is much faster than D-pad in AW3, but not by as much in AW4, because they screwed up the stylus inputs somehow. (I always used the stylus in Time Survival in AW3, and often elsewhere too, but it's so frustrating to use in AW4 for reasons I haven't figured out yet.) Note that pressing L automatically selects an unmoved unit; in the maps where you don't care which order your units move in (which is common, but not always the case), this is probably the fastest way to play. (Also, I find that A-up-A, or when using stylus A-tap, is faster than Select because you don't need to move your hands around as much.)

If pacifist is faster, then on a serious speedrun, you should probably go pacifist. The occasional low rank is quite common in an Advance Wars speedrun; I think there are plenty of As in the AW2 TAS, for example. (Pity there are no Cs; getting a C rank takes serious effort in this game.)

Also, I'd say CO powers aren't much of a staple of AW4. They're more like an emergency panic button in this game, or something for finishing off when you're winning, than they are a part of normal play. In a speedrun, you're unlikely to even charge one, and if you do, in most cases you want to keep it around for the radius, unless you have a very specific use for the CO power.

The different translations also make the game quite hard to watch. (In the UK translation, for example, the units have different names; some are the same names as in the US version but assigned to different units, e.g. Artillery (US AWDoR) is called Mech Gun, or Mech for short (UK AWDC), which makes things very confusing.) I'd recommend you use the US version, because the localization is better in most respects (I prefer the CO and mission names in the UK version, but nothing else).

One final important thing is to ensure that there are no points where you need to turn battle animations on to see exactly how much fractional damage was dealt. That's likely to cost a lot of time, and is probably going to be avoidable (especially as you typically aren't going for rout).

All this said, I'm not convinced that rout vs. HQ capture is as unbalanced as it seems. Playing for a fast rout can reduce the amount of time the opponent spends moving units around, given that the actual act of an HQ capture takes two turns.

GipFace

2015-02-21 06:45:16 am

Thanks for the response ais523!

- The main difference between stylus control from AW3 to AW4 is that you can tap your unit, then immediately tap an enemy unit to bring up the damage box. This is a great time saver for indirect-fire units. With direct-fire units, it will assign a path with the least MP usage. If there are multiple such paths but results in attacking from various directions, the path is randomly chosen. Therefore, for precision direct-fire attacks, it's still best to do it the traditional way.

- For now I'll keep with getting the S-rank in C16 because the extra attack literally only takes an additional 1.5 seconds, and we're not yet at the point where times are super competitive.

- CO powers are certainly useful in multiplayer, but that's irrelevant to the speedrun. In C18, Will's CO power (+2 MP) is only used to reach enemy units that have retreated to the corner of the map. It may be possible to prevent this but the AI is quite random here.

- As you can see from the Twitch highlight, I'm using the NA version for the speedrun. (I'm from Canada so "US" bothers me.) The NA version is superior because it keeps the naming convention from the older games. EU/AU unit names are awful and confusing: The duster is called a "fighter", and the real fighter is called an "interceptor".

- There are a few maps in AW's history where rout is indeed faster than HQ capture, but the only one that may apply here is C22.

---

And now some great news: AW4's planning is finished! Hopefully I'll be able to record a <1-hour run this weekend and smash the current WR of 3:40:25.

I'll still be using a DS lite with a custom stylus. As the start button is crucial, only the New 3DS/XL are viable alternatives. The regular 3DS/XL have the start button below the touch screen, which is bad.

Thanks for the response ais523!

- The main difference between stylus control from AW3 to AW4 is that you can tap your unit, then immediately tap an enemy unit to bring up the damage box. This is a great time saver for indirect-fire units. With direct-fire units, it will assign a path with the least MP usage. If there are multiple such paths but results in attacking from various directions, the path is randomly chosen. Therefore, for precision direct-fire attacks, it's still best to do it the traditional way.

- For now I'll keep with getting the S-rank in C16 because the extra attack literally only takes an additional 1.5 seconds, and we're not yet at the point where times are super competitive.

- CO powers are certainly useful in multiplayer, but that's irrelevant to the speedrun. In C18, Will's CO power (+2 MP) is only used to reach enemy units that have retreated to the corner of the map. It may be possible to prevent this but the AI is quite random here.

- As you can see from the Twitch highlight, I'm using the NA version for the speedrun. (I'm from Canada so "US" bothers me.) The NA version is superior because it keeps the naming convention from the older games. EU/AU unit names are awful and confusing: The duster is called a "fighter", and the real fighter is called an "interceptor".

- There are a few maps in AW's history where rout is indeed faster than HQ capture, but the only one that may apply here is C22.

---

And now some great news: AW4's planning is finished! Hopefully I'll be able to record a <1-hour run this weekend and smash the current WR of 3:40:25.

I'll still be using a DS lite with a custom stylus. As the start button is crucial, only the New 3DS/XL are viable alternatives. The regular 3DS/XL have the start button below the touch screen, which is bad.

ais523

2015-02-21 11:14:24 am

I'm actually a little surprised that Canada and US use the same translation. I'd have thought it would be more likely that Canada would use the European translation, so that it worked in French too. Glad to hear you've got the actually good translation. (All the cutscenes with actual text are skippable, right? So there's no speed difference due to different lengths in the translations.)

Interesting about the rout vs. HQ capture thing. My prediction for "most rout-friendly map" would have been C03, but looking up the actual layout, it doesn't actually have an HQ, so no wonder you didn't mention it.

I play the game on the DS original, but I can't offer much more than encouragement, because I know that you're much much better at this game than I am. (I did manage to S-rank every level by myself, and get a handful of 450s on maps where they're particularly easy, but that's about it.)

GipFace

Edit history:

GipFace: 2015-02-27 03:53:24 pm

2015-02-27 03:53:06 pm

Got the sub-1 hour run! Did AW4 in 57:02:

Thanks for all your support. I'll keep pushing and refine it. 55 minutes is definitely possible!

ais523

2015-02-27 10:15:06 pm

I take it what happened on C2 was bad luck rolls?

In C4, you could have retreated the advance infantry one space east rather than NWW; that should save some time on animation. (I'm not sure how deterministic the AI is on that map; moving it two spaces north might be more consistent.)

What's the save for in C7? It looks like you're manipulating luck for the bike to take 0 from the duster, but if it took 1 instead, I can't see any reason why it couldn't just complete the capture a turn later (which would be faster than resetting). Is there some AI pattern (e.g. rockets moving so that the HQ's within their minimum range) that would kill the run and that didn't happen here?

I like C8 because it really shows off the difference between speedrun and casual play. The "infantry in T-copter escorted with other units to surround it" strategy is one that I enjoy casually, but you can't pull it off on that map without near-perfect knowledge of where the enemy units are.

On C9, I don't think you have to kill the Missiles near the end. It's probably faster anyway, though, because otherwise you'd have to block the anti-air from doing chip damage to your capturing infantry.

C11: Seriously? That works? I think this is my second-favourite Advance Wars strategy (after the strategy for "Two-Week Test" from the TAS of AW2, which works by never building any units so that the opponent cannot score a rout, and luck-manipulating her to move her units in the wrong order and fail to capture the HQ as a result).

I don't see why you needed the second Seaplane in C13 (that's what they're called in this translation, right?). It cost you a few seconds to build it, and you don't use it for anything (and I don't think it manipulates Greyfield's orders either).

On C15, I like how different this strategy is from the usual ones. The important goals are keeping the blue recon safe (you made sure it never got into rockets range, for instance), and moving your units as little as possible, because this is basically an autoscroller. A pity the yellow recon didn't die with on your turn near the end, costing you a turn; is that KO possible?

C16: Wow. I'm in awe at this even though I managed something similar myself once; this version is just so much cleaner (not to mention faster).

The second attempt at C17 (let us not speak of the first) brings back memories. This is one of the first levels I really started trying to optimize, and I came up with exactly that strategy. The 450 strategy is basically the same too (the only difference is that it manipulates enemy unit building for a few extra technique points). Thus, I guess what this means is that that map is so easy to solve that even I can do it :-)

C20 is a really tedious map normally. That shows even in the speed route you're using, which is quite slow compared to most maps. It's still crazily fast compared to a normal completion, though. (Also, I assume you had to restart it in the attempt your splits are comparing to? A 3 minute saving is quite a lot.)

What was the purpose of the Fighter moves you did on turn 1 of C22? I can't figure it out offhand. (Also, Recons are a pretty weird unit to build near the end. I guess movement range is more important than any other consideration, because other units simply wouldn't get to the battle in time.)

C23 is an AI exploit, right? Normally the AI loves to attack meteors, but the fact that it leaves that meteor alone (presumably due to a special case for that level?) seems to be the only reason this strategy works (otherwise the anti-air would have saved the HQ in time).

How preplanned is your strategy for C24? This is one of my favourite levels, but it must be pretty crazy to route. It feels like there's probably a faster route available, at least.

The strategy you use on C25 makes me think that there must be something predictable determining the AI unit movement order. If it had moved the fighter before the duster, it'd have surely got a hit in on the infantry capturing the HQ. Perhaps not?

Finally, congratulations on finding a fast solution to C26. Are you manipulating the lasers, and if so, how? I came to the conclusion that they weren't random, but couldn't figure out what the actual pattern was.

Your run really brought home to me just how much of this game is cutscenes. At first, I couldn't figure out why your hands were where they were, and then I realised that your right hand was almost entirely on cutscene-skipping duty (and occasionally confirming an attack). I'm beginning to think that ending turn with Select+tap may be faster when it's in the middle of a sequence of cutscenes.

Another control suggestion: it seems to take you some time and thought to find Wait in the menu for Rigs. I'm thinking it might be faster to hit Supply (or even Build, if there's nothing to supply) instead and then move your stylus on to the next unit during the animation; even though the game's taking more time in animations, it's not much, and you might save more time that way than you lose.

Finally, an observation I made trying out some 450 routes recently. I saved at the end of a turn (i.e. I had no unmoved units), in the hope of manipulating an enemy's turn (a few turns into the C16 route on World Wars News, in case it matters). They did the same thing something like 10 times in a row (quickloading each time). Then I tried changing options on a whim, and they did something different. In fact, each time I went and changed options (probably just opening the options menu is enough?), I saw one specific set of AI actions; each time I didn't, I saw a different set of AI actions. I'm wondering if there's some sort of luck manipulation possible outside TAS via this method. (Or perhaps I'm just crazy.) Come to think of it, it looks a lot like you were doing a more efficient version of that after your reload on C20.

I take it what happened on C2 was bad luck rolls?

In C4, you could have retreated the advance infantry one space east rather than NWW; that should save some time on animation. (I'm not sure how deterministic the AI is on that map; moving it two spaces north might be more consistent.)

What's the save for in C7? It looks like you're manipulating luck for the bike to take 0 from the duster, but if it took 1 instead, I can't see any reason why it couldn't just complete the capture a turn later (which would be faster than resetting). Is there some AI pattern (e.g. rockets moving so that the HQ's within their minimum range) that would kill the run and that didn't happen here?

I like C8 because it really shows off the difference between speedrun and casual play. The "infantry in T-copter escorted with other units to surround it" strategy is one that I enjoy casually, but you can't pull it off on that map without near-perfect knowledge of where the enemy units are.

On C9, I don't think you have to kill the Missiles near the end. It's probably faster anyway, though, because otherwise you'd have to block the anti-air from doing chip damage to your capturing infantry.

C11: Seriously? That works? I think this is my second-favourite Advance Wars strategy (after the strategy for "Two-Week Test" from the TAS of AW2, which works by never building any units so that the opponent cannot score a rout, and luck-manipulating her to move her units in the wrong order and fail to capture the HQ as a result).

I don't see why you needed the second Seaplane in C13 (that's what they're called in this translation, right?). It cost you a few seconds to build it, and you don't use it for anything (and I don't think it manipulates Greyfield's orders either).

On C15, I like how different this strategy is from the usual ones. The important goals are keeping the blue recon safe (you made sure it never got into rockets range, for instance), and moving your units as little as possible, because this is basically an autoscroller. A pity the yellow recon didn't die with on your turn near the end, costing you a turn; is that KO possible?

C16: Wow. I'm in awe at this even though I managed something similar myself once; this version is just so much cleaner (not to mention faster).

The second attempt at C17 (let us not speak of the first) brings back memories. This is one of the first levels I really started trying to optimize, and I came up with exactly that strategy. The 450 strategy is basically the same too (the only difference is that it manipulates enemy unit building for a few extra technique points). Thus, I guess what this means is that that map is so easy to solve that even I can do it :-)

C20 is a really tedious map normally. That shows even in the speed route you're using, which is quite slow compared to most maps. It's still crazily fast compared to a normal completion, though. (Also, I assume you had to restart it in the attempt your splits are comparing to? A 3 minute saving is quite a lot.)

What was the purpose of the Fighter moves you did on turn 1 of C22? I can't figure it out offhand. (Also, Recons are a pretty weird unit to build near the end. I guess movement range is more important than any other consideration, because other units simply wouldn't get to the battle in time.)

C23 is an AI exploit, right? Normally the AI loves to attack meteors, but the fact that it leaves that meteor alone (presumably due to a special case for that level?) seems to be the only reason this strategy works (otherwise the anti-air would have saved the HQ in time).

How preplanned is your strategy for C24? This is one of my favourite levels, but it must be pretty crazy to route. It feels like there's probably a faster route available, at least.

The strategy you use on C25 makes me think that there must be something predictable determining the AI unit movement order. If it had moved the fighter before the duster, it'd have surely got a hit in on the infantry capturing the HQ. Perhaps not?

Finally, congratulations on finding a fast solution to C26. Are you manipulating the lasers, and if so, how? I came to the conclusion that they weren't random, but couldn't figure out what the actual pattern was.

Your run really brought home to me just how much of this game is cutscenes. At first, I couldn't figure out why your hands were where they were, and then I realised that your right hand was almost entirely on cutscene-skipping duty (and occasionally confirming an attack). I'm beginning to think that ending turn with Select+tap may be faster when it's in the middle of a sequence of cutscenes.

Another control suggestion: it seems to take you some time and thought to find Wait in the menu for Rigs. I'm thinking it might be faster to hit Supply (or even Build, if there's nothing to supply) instead and then move your stylus on to the next unit during the animation; even though the game's taking more time in animations, it's not much, and you might save more time that way than you lose.

Finally, an observation I made trying out some 450 routes recently. I saved at the end of a turn (i.e. I had no unmoved units), in the hope of manipulating an enemy's turn (a few turns into the C16 route on World Wars News, in case it matters). They did the same thing something like 10 times in a row (quickloading each time). Then I tried changing options on a whim, and they did something different. In fact, each time I went and changed options (probably just opening the options menu is enough?), I saw one specific set of AI actions; each time I didn't, I saw a different set of AI actions. I'm wondering if there's some sort of luck manipulation possible outside TAS via this method. (Or perhaps I'm just crazy.) Come to think of it, it looks a lot like you were doing a more efficient version of  that after your reload on C20.

GipFace

Edit history:

GipFace: 2015-02-28 09:25:33 am

GipFace: 2015-02-28 09:22:03 am

2015-02-28 09:17:50 am

I'll go mission by mission.

C2: Yes, bad luck. Thankfully Beast suicided on his turn so that I didn't lose a few more seconds.

C4: The stray infantry moves 1N2W because there's a possibility I lose the luck roll and don't 2HKO that recon. That recon takes a minimum of 4+35+52 = 91 health damage, which means I need 9 luck to KO it. I get 3 luck rolls: 3, 4, and 8. I actually have less than a 50% chance to 2HKO that recon. Then infantry would then be used for sight. Looking back, it's probably just better to suicide the infantry to bring the recon to 95 health damage, though there's still a slim chance that the recon survives.

C7: The save is to ensure that the duster is brought to 2hp. Antiair vs. duster is 75%, so there is a 5 in 11 chance that the duster is brought to 3hp instead of 2hp. A 2hp duster can't take 1hp off the capping bike, while a 3hp duster can. Saving and reloading if the luck roll fails is far faster than capping a day later. An extra day means that every single one of Tasha's units move again.

C9: If the missiles isn't destroyed, it'll target the weak fighter walling your capping inf.

C11: You might wonder why the enemy gunboats don't attack. That's because they're on defensive mode and don't consider your gunboats to be a threat! I head right towards the HQ, then bait the artillery with another gunboat to prevent it from going back. You can actually do this mission pacifist style (just like C16) but it takes longer.

C13: That was a mistake. You need the second seaplane if Greyfield calls md tank or battleship, because the first one will need to be brought out early and then get shot down. Force of habit.

C15: Bad luck roll. I should actually put in a save point there because losing the luck roll costs many more seconds.

C20: The mega time save was because that strategy was new and wasn't in the previous run. This new strategy only requires 2 or 3 save points compared to the previous run's 5. In addition, there is no longer a double luck roll requirement (wanting 2 things to happen on the AI's turn), which crippled the previous run. I had to reload because Waylon's north duster retreated. At 3hp it won't retreat, but at 2hp it may or may not.

C22: The enemy fighters and bombers will not go into range of your fighters. In order to lure them into the nuke's blast zone, you must retreat your own fighters. Penny crippling her own air force is the secret to beating this mission quickly.

C24: This completely fell apart. I forgot to attack the war tank, then I mistakenly used my md tank on a crippled tank instead of the fresh tank. I usually complete this a minute sooner on day 6. Both of your rockets aren't supposed to get destroyed.

C25: Land units always move before air units, and newer units move last. You can see that this AI peculiarity is also exploited in C20: Waylon's two antiair attack my COU fighter before his fighter moves.

C26: Lasers aren't special. They fire every day in order of left, middle, right. As for the mortars, they will always fire on the most expensive cluster of units, but the barrier columns for each mortar are where The Nest spawns new units. Any units on or between the two columns won't be targeted by the mortar. (You can actually check the range on the mortar, try it out) For example, the right mortar only hit my infantry on D2 because that was the only unit I had beyond the right barrier. The left mortar hit my missiles+antiair+infantry cluster because they're valued at F20500, while the COU war tank + 2 infantry are valued at F19000. The CO being in the war tank doesn't add any additional value to the calculation.

RNG Manipulation: Right before confirming an attack, you'll see a damage box pop up. The RNG will advance every frame that the damage box bounces up-and-down, so it changes every 1/60th of a second. The amount shown on the damage box is the minimum damage that will be dealt. Units may deal up to an additional 10%, though this number can be reduced with either less attacker HP or additional defender DEF.

As for attempting to change AI movement on the following turn, you can manipulate the RNG with a single unit that can move diagonally. The RNG determines the pathing for a unit if there is more than one way to arrive at the destination using the same amount of MP. This is similar to Fire Emblem; Intelligent Systems made both games, after all! As you can see in C20 at the 31:05 mark, the AI didn't behave initially, forcing me to reload. I then set a diagonal path with the mech and cancelled. The AI then moved the way I wanted it to.

I'll go mission by mission.

[b]C2:[/b] Yes, bad luck. Thankfully Beast suicided on his turn so that I didn't lose a few more seconds.

[b]C4:[/b] The stray infantry moves 1N2W because there's a possibility I lose the luck roll and don't 2HKO that recon. That recon takes a minimum of 4+35+52 = 91 health damage, which means I need 9 luck to KO it. I get 3 luck rolls: 3, 4, and 8. I actually have less than a 50% chance to 2HKO that recon. Then infantry would then be used for sight. Looking back, it's probably just better to suicide the infantry to bring the recon to 95 health damage, though there's still a slim chance that the recon survives.

[b]C7:[/b] The save is to ensure that the duster is brought to 2hp. Antiair vs. duster is 75%, so there is a 5 in 11 chance that the duster is brought to 3hp instead of 2hp. A 2hp duster can't take 1hp off the capping bike, while a 3hp duster can. Saving and reloading if the luck roll fails is far faster than capping a day later. An extra day means that every single one of Tasha's units move again.

[b]C9:[/b] If the missiles isn't destroyed, it'll target the weak fighter walling your capping inf.

[b]C11:[/b] You might wonder why the enemy gunboats don't attack. That's because they're on defensive mode and don't consider your gunboats to be a threat! I head right towards the HQ, then bait the artillery with another gunboat to prevent it from going back. You can actually do this mission pacifist style (just like C16) but it takes longer.

[b]C13:[/b] That was a mistake. You need the second seaplane if Greyfield calls md tank or battleship, because the first one will need to be brought out early and then get shot down. Force of habit.

[b]C15:[/b] Bad luck roll. I should actually put in a save point there because losing the luck roll costs many more seconds.

[b]C20:[/b] The mega time save was because that strategy was new and wasn't in the previous run. This new strategy only requires 2 or 3 save points compared to the previous run's 5. In addition, there is no longer a double luck roll requirement (wanting 2 things to happen on the AI's turn), which crippled the previous run. I had to reload because Waylon's north duster retreated. At 3hp it won't retreat, but at 2hp it may or may not.

[b]C22:[/b] The enemy fighters and bombers will not go into range of your fighters. In order to lure them into the nuke's blast zone, you must retreat your own fighters. Penny crippling her own air force is the secret to beating this mission quickly.

[b]C24:[/b] This completely fell apart. I forgot to attack the war tank, then I mistakenly used my md tank on a crippled tank instead of the fresh tank. I usually complete this a minute sooner on day 6. Both of your rockets aren't supposed to get destroyed.

[b]C25:[/b] Land units always move before air units, and newer units move last. You can see that this AI peculiarity is also exploited in C20: Waylon's two antiair attack my COU fighter before his fighter moves.

[b]C26:[/b] Lasers aren't special. They fire every day in order of left, middle, right. As for the mortars, they will always fire on the most expensive cluster of units, but the barrier columns for each mortar are where The Nest spawns new units. Any units on or between the two columns won't be targeted by the mortar. (You can actually check the range on the mortar, try it out) For example, the right mortar only hit my infantry on D2 because that was the only unit I had beyond the right barrier. The left mortar hit my missiles+antiair+infantry cluster because they're valued at F20500, while the COU war tank + 2 infantry are valued at F19000. The CO being in the war tank doesn't add any additional value to the calculation.

[b]RNG Manipulation:[/b] Right before confirming an attack, you'll see a damage box pop up. The RNG will advance every frame that the damage box bounces up-and-down, so it changes every 1/60th of a second. The amount shown on the damage box is the minimum damage that will be dealt. Units may deal up to an additional 10%, though this number can be reduced with either less attacker HP or additional defender DEF.

As for attempting to change AI movement on the following turn, you can manipulate the RNG with a single unit that can move diagonally. The RNG determines the pathing for a unit if there is more than one way to arrive at the destination using the same amount of MP. This is similar to Fire Emblem; Intelligent Systems made both games, after all! As you can see in C20 at the 31:05 mark, the AI didn't behave initially, forcing me to reload. I then set a diagonal path with the mech and cancelled. The AI then moved the way I wanted it to.