Phantasy Star IV

2015-09-22 03:32:01 am

I did some more experimenting with the Sand Worm fight, especially what to do if Hahn dies before the fight is over. I was able to defeat the Sand Worm with Hahn dying after only 2 Tri-Blasters, which obviously isn't ideal, but is doable. I ensured Alys and Gryz both died as well, and limped along with Chaz and Rika. It was bad luck to lose Hahn so early, but my luck was ok for the rest of the fight, in that the Sand Worm missed a lot, which let me keep up on heals. In a better fight you could get all the Tri-Blasters off before Hahn dies. With Hahn dead before the end however, it is more important to Shift Chaz before Alys dies.

With only two party members alive you receive 12579 XP which is enough to get Rika to at least level 13 regardless of what level she was at the start of the fight (but she's usually around lvl 4-5) What this means is that she learns Gires from the Sand Worm fight, so you don't have to do any grinding in the Passage. It also means that Rika is guaranteed to learn Deban from Juza fight.

Also, in a funny turn of events, did you know it's possible to beat Gy-Laguiah with no TP for Rika, and no Crosscut for Chaz? Because it is. I forgot to heal before going to Ladea Tower.

I did some more experimenting with the Sand Worm fight, especially what to do if Hahn dies before the fight is over. I was able to defeat the Sand Worm with Hahn dying after only 2 Tri-Blasters, which obviously isn't ideal, but is doable.  I ensured Alys and Gryz both died as well, and limped along with Chaz and Rika. It was bad luck to lose Hahn so early, but my luck was ok for the rest of the fight, in that the Sand Worm missed a lot, which let me keep up on heals.  In a better fight you could get all the Tri-Blasters off before Hahn dies.  With Hahn dead before the end however, it is more important to Shift Chaz before Alys dies.

With only two party members alive you receive 12579 XP which is enough to get Rika to at least level 13 regardless of what level she was at the start of the fight (but she's usually around lvl 4-5) What this means is that she learns Gires from the Sand Worm fight, so you don't have to do any grinding in the Passage. It also means that Rika is guaranteed to learn Deban from Juza fight.

Also, in a funny turn of events, did you know it's possible to beat Gy-Laguiah with no TP for Rika, and no Crosscut for Chaz?  Because it is. I forgot to heal before going to Ladea Tower.

Adgey

2015-09-22 04:47:38 am

I just lost a run that was on insane pace to lassic opening with anothergate before I got barrier off. I think it might have been on 3:30-3:35 pace or something. I tried a new motovia strat and it worked out that time but it seems a bit risky for now, it's killing off hahn in the sandworm fight then you don't need to do ANY slug farming. I managed to get a 1:01 Zio with this and I forgot to buy shields for rika on the zio fight so I had to equip carbonshields. Sub 1 hour Zio is definitely doable. and sub 3:30 seems likely too.

philosoraptor42

2015-09-22 08:45:36 am

Dapper as fuck.

The work you guys are putting into this is amazing.

Jiseed

2015-09-22 03:52:30 pm

Very interesting stuff, if we no longer want Hahn to live we can effectively cut out the grinding in the Wreckage (still do a couple fights for Rika though), as well as the first trip to Aiedo. Not using Psyshields on Rika for Zio seems extremely frightening, I wonder what kind of impact it'd have on Raja's survivability on DF1 as well. Looking forward to doing some testing on this stuff.

Adgey

2015-09-22 04:14:02 pm

I think youd want to still go to Aeido and buy Hahn's stuff because you still need to wrekage grind I think becasue if Rika and Chaz are too low level finishing the sandworm fight with only them alive will be far too risky. Also I'm still getting Hahn to level 7 for the time being because you still need him for the majority of the fight. I think the new party order you'd want on sand worm is Gryz>Alys>Rika>Hahn>Chaz for the best chance to make it through with just Rika and Chaz. If Hahn is at the back it's a real pain to kill him because he needs 2 hits usually, but if hes too close to the front he can die too early. It's a hard thing to balance although I've only tried a couple of times.

I'm also still using psyshields on Rika its just that time I forgot and had to improvise heh.

squidfeatures

Edit history:

squidfeatures: 2015-09-22 10:18:44 pm

2015-09-22 10:18:31 pm

It won't be hard to have Hahn die in the fight, the tricky part will be getting him to die at the right time. The Vol grinding in Passageway might end up being a good backup strategy, but it certainly isn't optimal anymore.

squidfeatures

2015-09-23 02:18:55 am

I've played through to Juza fight and did some tests with him (after having Hahn die at Sand Worm). It is probably a good idea to make two of the Slug killing macros, replacing Vol with Eliminate (thanks for the tip Taloof!), and using Chaz to get rid of the spare slug, so if you happen upon 2-3 Zol Slugs you can get some extra XP. I wouldn't replace the Tri-Blaster and Blizzard macros though, since you can use Tri-Blaster on surprise encounters in Zio's Fort for some extra XP. (Taloof mentioned to me that in one of his runs Rika was able to get Rika to level 14 before Juza!)

You can beat Juza with only Rika and Chaz alive at the start, you just have Rika use Saner first turn. The fight is obviously smoother with at least Alys and Hahn alive as well.

Shifted Rika attack does about as much damage as Shifted Chaz Attack

Also at level 13, Chaz's TSU does around 80 damage to Juza, which is less that Crosscut but more than a regular attack, even with Shift.

The Tri-Blaster and Blizzard macro's can be useful for Juza, since as mentioned above TSU is more damage than attacking, so when Chaz is out of Crosscuts, you can use Macro A when you don't need much healing, and if Hahn or Alys are still alive (unlikely but possible) they'll sneak some damage in too. Same with Macro B after Chaz runs out of TP. The macros are slightly more useful in the case where a party member (other than Chaz or Rika) is dead before the fight.

The fight isn't hard in any case, but all the little things help shave time off here and there.

Anyway, I won't be working much on the run until Saturday, and I might explore what happens if Chaz or Rika die in the Juza fight. I may pop on here before then and see what's up though.

I've played through to Juza fight and did some tests with him (after having Hahn die at Sand Worm). It is probably a good idea to make two of the Slug killing macros, replacing Vol with Eliminate (thanks for the tip Taloof!), and using Chaz to get rid of the spare slug, so if you happen upon 2-3 Zol Slugs you can get some extra XP.  I wouldn't replace the Tri-Blaster and Blizzard macros though, since you can use Tri-Blaster on surprise encounters in Zio's Fort for some extra XP. (Taloof mentioned to me that in one of his runs Rika was able to get Rika to level 14 before Juza!)

You can beat Juza with only Rika and Chaz alive at the start, you just have Rika use Saner first turn. The fight is obviously smoother with at least Alys and Hahn alive as well.

Shifted Rika attack does about as much damage as Shifted Chaz Attack

Also at level 13, Chaz's TSU does around 80 damage to Juza, which is less that Crosscut but more than a regular attack, even with Shift.

The Tri-Blaster and Blizzard macro's can be useful for Juza, since as mentioned above TSU is more damage than attacking, so when Chaz is out of Crosscuts, you can use Macro A when you don't need much healing, and if Hahn or Alys are still alive (unlikely but possible) they'll sneak some damage in too.  Same with Macro B after Chaz runs out of TP.  The macros are slightly more useful in the case where a party member (other than Chaz or Rika) is dead before the fight.

The fight isn't hard in any case, but all the little things help shave time off here and there.

Anyway, I won't be working much on the run until Saturday, and I might explore what happens if Chaz or Rika die in the Juza fight. I may pop on here before then and see what's up though.

Adgey

Edit history:

Adgey: 2015-09-23 05:56:13 am

2015-09-23 05:53:26 am

thanks for the info squid. I finally completed a run. I managed to get a really good time of 3:37:50! The run had a 1:10 Motov (leaving for Zelan), 1:26 Dezolis and 1:01 for the last part of the game. Pretty decent run overall and the last part of the game had pretty good luck especially the final boss. Dezolis was pretty sloppy to be honest and can at least be improved by 5+ minutes. Motovia was ok but that can be improved too. I think the zio fight will eventually shift towards Rika using claws instead of shields because it's soo slow with the shields. I used shields on this run though. If anyone wants to see it it's in two parts on my twitch page. I'll upload it to youtube sometime before it expires on twitch.

Here's part 1: http://www.twitch.tv/adgey10/v/17347744

edit: for some reason my stream is pretty laggy, ill try to fix it next time.

Jiseed

Edit history:

Jiseed: 2015-09-24 03:05:11 pm

2015-09-23 04:15:34 pm

Good stuff, I knew that sub 3:40 would be soon given a bit of luck. I've done a little research of my own and I definitely think that skipping the first trip to Aiedo before going into the Wreckage works out well. It also allows us to buy 2-3 less telepipes in Tonoe, so perhaps we can take a little time off of the Rune grind if need be. I like the idea of killing off Hahn, usually if you can get all of the Triblasters off first you're set. Rika using Illusion seems to help once its just her and Chaz, but I might just have gotten lucky. Additionally, after looking at doing a little grinding in Kuran I realized that Kyra would indeed be pretty far behind Rika (depending on encounter rate of course). However, I did some test runs of the bosses and at the end of the day you really don't need to have Warla for this section. Lashiec is a bit rough because you're pretty much guaranteed to use a Star Dew on the second turn so Rika can use Saner, but its still quite managable. On top of that I did a little testing on the Laconium Sword vs the Flame Sword and the difference is pretty negligible. You do roughly 40 more damage per Crosscut on Lashiec and maybe 20 less damage to DF2 and the Triplets. Being able to push Lashiec faster seems to be a huge resource saver, so I'm opting for teleporting to Zoza post-weapons factory and grabbing the flame sword and Dimates instead of going to Jut. The town is smaller so you save time on walking around as well. It severely slows down the grinds in Vahal Fort, should stick with the Laconia Sword. I hope that we don't get a 33 Kyra for Profound Darkness if we skip getting her Warla for the Air Castle, as that might cause some complications.

Good stuff, I knew that sub 3:40 would be soon given a bit of luck. I've done a little research of my own and I definitely think that skipping the first trip to Aiedo before going into the Wreckage works out well. It also allows us to buy 2-3 less telepipes in Tonoe, so perhaps we can take a little time off of the Rune grind if need be. I like the idea of killing off Hahn, usually if you can get all of the Triblasters off first you're set. Rika using Illusion seems to help once its just her and Chaz, but I might just have gotten lucky. Additionally, after looking at doing a little grinding in Kuran I realized that Kyra would indeed be pretty far behind Rika (depending on encounter rate of course). However, I did some test runs of the bosses and at the end of the day you really don't need to have Warla for this section. Lashiec is a bit rough because you're pretty much guaranteed to use a Star Dew on the second turn so Rika can use Saner, but its still quite managable. [s]On top of that I did a little testing on the Laconium Sword vs the Flame Sword and the difference is pretty negligible. You do roughly 40 more damage per Crosscut on Lashiec and maybe 20 less damage to DF2 and the Triplets. Being able to push Lashiec faster seems to be a huge resource saver, so I'm opting for teleporting to Zoza post-weapons factory and grabbing the flame sword and Dimates instead of going to Jut. The town is smaller so you save time on walking around as well.[/s] It severely slows down the grinds in Vahal Fort, should stick with the Laconia Sword.  I hope that we don't get a 33 Kyra for Profound Darkness if we skip getting her Warla for the Air Castle, as that might cause some complications.

Taloof

Edit history:

Taloof: 2015-09-24 12:11:02 am

2015-09-24 12:10:22 am

Hey SF, that one sentence sounded a bit confusing to me :

(Taloof mentioned to me that in one of his runs Rika was able to get Rika to level 14 before Juza!)

Well what I said was that I used all the surprise attacks to get in some extra xp on my way to Juza as Rika needs ~600 xp after Sandworm if she is still lvl 13. So if you get a surprise attack on 4 Spears it is around 200-300 XP I think and Shadowsabers and Rippers give even more over 300. (Basically 2 Triblasters).

Taloof

Edit history:

Taloof: 2015-09-24 03:03:41 pm

2015-09-24 03:03:00 pm

I think I got a nice idea for an optimizationstrat which could potentially save some minutes. At least I hope it works. Will test it out this weekend before i spout nonsense again. Cheesy

Also what about buffs/debuffs during boss fights? And I mean those besides Saner, Deban, Shift. Are they worth it? For example Hahn's Vision increases the chance to hit and critically hit which at least in the first case helps Chaz to deal some damage against the Sand Worm instead of missing 5 times in a row.

Taloof

Edit history:

Taloof: 2015-09-27 04:59:26 am

2015-09-27 04:53:21 am

So I finished a Lashiec% run yesterday and tried to include a couple optimizations in my run.
Here is what I changed:

Makro for Dark Force 1:

D = Rika - Sar / Wren - Recover / Chaz - Githu / Raja - Sar / Rune Githoi
E = Rune - Gifoi / Rika - Sar / Wren - Flare / Chaz - Githu / Raja - Atharaxia

Unless DF1 kills off one of your characters immediately you are all set to mainly use makros throughout the fight with Chaz using Githu all the time.

After DF1 I tested what would happen if you ignore Raja and go straight for Kyra. So I created a savestat at the Spaceport and finished timing after using up all Skills and TP in the weapon factory.

My best time with Raja-grinding session was around 30-31 minutes. Using Kyra instead and giving her all the XP from the Silvertuskcave/Gy-Lagiah & Factory -battles ended up with around 27-28 minutes and she is guaranteed to learn Warla. In Air Castle you don't need to fight that many battles in order for Rika to learn GISAR. My initial time for Raja was 38:30 and with Kyra a 36:16. I guess practice/improving menuing and getting the right battles will be the main time improvements there.

My Kyra-Route:
1) Starting from spaceport --> Zosa for RYUKApoint --> Meese and get rid of Raja --> get Kyra. That should take around 2min30sec. You could either RYUKA to Meese and then drive to the Espermansion or just drive straight there. It is only a matter of 1 or 2 seconds unless you get stuck in battle so use RYUKA. (I tested to use the blackhole on all ICE-DIGGER-battles with 4 flying robots which is an additional 500-600 XP+MST but overall not really worth it)
2) Espermansion --> grab 4 chests --> Rune RFLC-Robe & Unequip Claw and Psy-Mail from Rika & Unequip Psy-Mail from Chaz & Psy-Crown for Kyra + unequip Laco-Slasher. Order --> Kyra/Wren/Rune/Rika/Chaz
Makros:
A = Wren - Att / Rune -Gra / Kyra - Gra /Rika -Att / Chaz - Att
B = !ATTENTION!! Rune - HEWN / Kyra - Flaeli / Wren - Att / Rika - Att / Chaz - Att --> Important for Makro H! I don't want to use up all Hewns on Kyra.
C = Rune - Flaeli / Kyra - Hewn / Wren - Att / - Rika - Att / Chaz - Att

H = Blizzard! Rune - Giwat / Kyra - Hewn / Wren - Flare / Rika - 2x-Slash / Chaz - Githu --> I use that makro for the 2nd Gy-Lagiah fight which kills it in turn 3. You can also use it in Air Castle to kill those Rippercreatures which are more resistant to fire but take tons of dmg from waterbased attacks. 4 Rippers infront of you? BAM BLIZZARD IN YOUR FACE! So it is faster to get some fights in on your way to the triplets.

Then meet Lutz and yada yada yada
3) Drive to Gambious first to get the 4 chests. --> Then go shopping in Jut and sell all the stuff to buy 2x RFLC-Mail and 1x Lacosword. Depending on money you need to sell a Star-Dew.
4) Ryuka to Zosa --> Get Silvertusk but also get a couple of fights there (=Minigrind like use up all Makro Bs) --> HINAS and RYUKA to Zosa and drive to Weapon Plant --> 3x Makro H --> Get weapon for Wren + maybe a couple of fights along the way.
5) HINAS + RYUKA to Jut and drive to Weapon Factory, business as usual.

So timewise it takes me roughly 14 minutes to get all the steps done up until point 5. the rest is grinding which depends heavily on which fights you get and their frequency. My fastest time was a late 25minutes 5X seconds to use up all TP/Skills but that was a really bad run as Wren died so much to the Lightingbeams which ended in Rune using 2 Gras in one fight.

Conclusion:

I think it can be faster to get Kyra first if it is not the same pace as with Raja but Kyra getting Warla before Air Castle is guaranteed. I don't necessarily like giving Raja too much XP if we don't use him. In my Lashiec% run it took me 2 whole minutes during the sandworm fight to actually get Hahn killed. Same happened during a couple of my test runs on Dezolis where Raja just didn't die and soaked up XP from fights which resulted in my team being more underleveled after Weapon Factory then just with Kyra in it and having Warla.

I also want to point out that I don't have that much experience with the 2nd part and especially the 3rd part of a full run so I am not quite sure wether or not you should go for Laco-Sword immediately or afterwards and maybe buy a RFLC-Shield as well. But fact is that Chaz and Rika can sustain dmg a lot easier in the Silvertuskcave/Factory if you get their RFLC-Mails immediately.

It also depends heavily on how your run prior worked out. For example after finally killing off Hahn in the Sandworm fight I got no nice Slugbattles to grab some extra xp or MST (which I missed later on to buy a 2nd Psy-crown...) So depending on how much MST/XP you could gather the results will obviously influence shopping/grinding habits on dezolis.

I will start focusing on practicing Dezolis (Garuberk Tower I am looking at you <_< ) and start with the 3rd part soon so that I can get a full run in hopefully. Smiley

So I finished a Lashiec% run yesterday and tried to include a couple optimizations in my run.
Here is what I changed:

Makro for Dark Force 1:

D = Rika - Sar / Wren - Recover / Chaz - Githu / Raja - Sar / Rune Githoi
E = Rune - Gifoi / Rika - Sar / Wren - Flare / Chaz - Githu / Raja - Atharaxia

Unless DF1 kills off one of your characters immediately you are all set to mainly use makros throughout the fight with Chaz using Githu all the time.

After DF1 I tested what would happen if you ignore Raja and go straight for Kyra. So I created a savestat at the Spaceport and finished timing after using up all Skills and TP in the weapon factory.

My best time with Raja-grinding session was around 30-31 minutes.  Using Kyra instead and giving her all the XP from the Silvertuskcave/Gy-Lagiah & Factory -battles ended up with around 27-28 minutes and she is guaranteed to learn Warla. In Air Castle you don't need to fight that many battles in order for Rika to learn GISAR. My initial time for Raja was 38:30 and with Kyra a 36:16. I guess practice/improving menuing and getting the right battles will be the main time improvements there.

My Kyra-Route:
1) Starting from spaceport --> Zosa for RYUKApoint --> Meese and get rid of Raja --> get Kyra.  That should take around 2min30sec. You could either RYUKA to Meese and then drive to the Espermansion or just drive straight there. It is only a matter of 1 or 2 seconds unless you get stuck in battle so use RYUKA. (I tested to use the blackhole on all ICE-DIGGER-battles with 4 flying robots which is an additional 500-600 XP+MST but overall not really worth it)
2) Espermansion --> grab 4 chests --> Rune RFLC-Robe & Unequip Claw and Psy-Mail from Rika & Unequip Psy-Mail from Chaz & Psy-Crown for Kyra + unequip Laco-Slasher.  Order --> Kyra/Wren/Rune/Rika/Chaz
     Makros: 
     A = Wren - Att / Rune -Gra / Kyra - Gra /Rika -Att / Chaz - Att
     B = !ATTENTION!!  Rune - HEWN / Kyra - Flaeli / Wren - Att / Rika - Att / Chaz - Att             --> Important for Makro H! I don't want to use up all Hewns on Kyra.
     C = Rune - Flaeli / Kyra - Hewn / Wren - Att / - Rika - Att / Chaz - Att

H = Blizzard!    Rune - Giwat / Kyra - Hewn / Wren - Flare / Rika - 2x-Slash / Chaz - Githu     -->   I use that makro for the 2nd Gy-Lagiah fight which kills it in turn 3. You can also use it in Air Castle to kill those Rippercreatures which are more  resistant to fire but take tons of dmg from waterbased attacks.   4 Rippers infront of you? BAM BLIZZARD IN YOUR FACE!  So it is faster to get some fights in on your way to the triplets.

Then meet Lutz and yada yada yada
3) Drive to Gambious first to get the 4 chests. --> Then go shopping in Jut and sell all the stuff to buy 2x RFLC-Mail and 1x Lacosword. Depending on money you need to sell a Star-Dew.
4) Ryuka to Zosa --> Get Silvertusk but also get a couple of fights there (=Minigrind like use up all Makro Bs) --> HINAS and RYUKA to Zosa and drive to Weapon Plant --> 3x Makro H --> Get weapon for Wren + maybe a couple of fights along the way.
5) HINAS + RYUKA to Jut and drive to Weapon Factory, business as usual.

So timewise it takes me roughly 14 minutes to get all the steps done up until point 5. the rest is grinding which depends heavily on which fights you get and their frequency. My fastest time was a late 25minutes 5X seconds to use up all TP/Skills but that was a really bad run as Wren died so much to the Lightingbeams which ended in Rune using 2 Gras in one fight.

Conclusion:

I think it can be faster to get Kyra first if it is not the same pace as with Raja but Kyra getting Warla before Air Castle is guaranteed. I don't necessarily like giving Raja too much XP if we don't use him. In my Lashiec% run it took me 2 whole minutes during the sandworm fight to actually get Hahn killed. Same happened during a couple of my test runs on Dezolis where Raja just didn't die and soaked up XP from fights which resulted in my team being more underleveled after Weapon Factory then just with Kyra in it and having Warla.

I also want to point out that I don't have that much experience with the 2nd part and especially the 3rd part of a full run so I am not quite sure wether or not you should go for Laco-Sword immediately or afterwards and maybe buy a RFLC-Shield as well. But fact is that Chaz and Rika can sustain dmg a lot easier in the Silvertuskcave/Factory if you get their RFLC-Mails immediately.

It also depends heavily on how your run prior worked out. For example after finally killing off Hahn in the Sandworm fight I got no nice Slugbattles to grab some extra xp or MST (which I missed later on to buy a 2nd Psy-crown...) So depending on how much MST/XP you could gather the results will obviously influence shopping/grinding habits on dezolis.

I will start focusing on practicing Dezolis (Garuberk Tower I am looking at you <_< ) and start with the 3rd part soon so that I can get a full run in hopefully. :)

Jiseed

2015-09-27 12:42:32 pm

I remember testing the Kyra first route a while back, it almost always guaranteed a second trip down to the weapons factory... seems you planned it out better than I did. I would say that while none of the exp is wasted doing this, and you're probably going to get a 35 Kyra for PD doing it, I don't think that it'll beat a four person party pre-Kyra simply because leveling up Rika is the priority. This is just my opinion of course, I could be totally wrong. I actually managed to catch a bit of your stream and it got me thinking about some stuff that I saw. First, 2 claw Rika at Zio seems slower simply due to the fact you have to micromanage pretty much every turn. With double shields you can go on autopilot with macros for the most part, which I think is still faster. Perhaps a set up of a shield and a claw would work out better if you wanted the damage; it'd make Hewn casts essentially worthless for Zio, with the added bonus of some attacks from Rika on occasion. In addition, I think it might be worth it to fight some of the 'hard to run' fights in Garuberk tower. The exp from them is usually on the top end of Weapon Factory fights, and since you're going to be wasting some turns trying to run, it might be more prudent to kill them and perhaps cut down on the grind in Vahal.

Taloof

2015-09-27 03:59:01 pm

Yeah about my run in general: I had a couple of equipmenthiccups. The Zio fight was a silly mistake as I forgot to equip Rika with the shields. Maybe a 1x Claw + 1x Shield is a nice balance to finish the fight faster. The moment I noticed I forgot to requip Rika I facepalmed.
The Garuberktower was actually the first time in like years that I have been in there as with the rest of the game. Had no idea at all where to go etc. but it was really nice to finally get a taste of the third part of the speedrun.

I will do some more testing with the whole dezolis part and not just right after DF1 in order to gather more data. Need the practice anyways Smiley

Adgey

2015-09-28 09:02:48 pm

grats on your 3:32 run Jiseed. A couple of things that make the sandworm more consistent I think - buy a graphite shied for Rika so she has two shields in the fight, also, more importantly make a Rika defend Chaz attack macro. Even if Hahn dies very early I found the fight pretty consistant with these changes. The two shields on Rika also up your survivability slightly on the way to Juza. I think it also speeds up the fight slightly because the defend animation is way faster than wood caning. If you get down to only rika and chaz early I use the wood cane/chaz attack macro if chaz is not at full hp and rika is below 35-40 hp(not sure whats optimal yet), otherwise you can just use the rika defend chaz attack macro.

Jiseed

2015-09-29 04:37:40 pm

Thanks Adgey, interesting idea with the defend macro, I'll definitely have to give it a shot. I did a little testing on some things that might be worth doing. I tested Rika using 1 claw and 1 psy-shield on Zio, and the results were pretty promising. Generally you're safe to do a double slash whenever he uses a hewn/low variance corrosion when Rika is at full health. It consistently allowed me to not use the Hewn macro, so its definitely faster. I also did some testing on killing off characters for the Dark Force 1 fight, and this might be a good way to reduce the grinding on Dezolis. You can have only Rika survive, which makes the level 28 grind extremely short - but the chances of beating Dark Force 2 are virtually non-existent without some additional grinding for Wren. I personally prefer having Rika and Wren being the sole survivors, as this usually guarantees a 25 Wren for DF2, which is quite necessary. Having Rune survive isn't bad either, might have to do a couple extra battles to get Wren up to speed though. Generally a good way to set this up is to do the first turn with the protections macro, then 6 turns of the other macros with a cast of St.Fire. This will usually set it up for DF1 to have 200-300hp left, which Rika or Wren/Rune can easily take care of themselves. Sometimes people are a bit slow to die, but I think that in the end it'll save some time. Chaz only loses around 5k exp, so its not a huge loss... I actually managed to kill PD with a 31 Chaz in my PB, but that could have just been decent luck.

Adgey

Edit history:

Adgey: 2015-09-30 06:42:04 am

Adgey: 2015-09-30 06:01:43 am

Adgey: 2015-09-29 08:51:05 pm

Adgey: 2015-09-29 08:49:42 pm

2015-09-29 05:40:38 pm

That makes a lot of sense with killing off characters for dark force 1. I've always at least tried to kill off raja in that fight, but killing off other people didn't really occur to me, and that also opens up if you should kill off characters for the triplets fight and every other fight after that. It all comes down to what do we NEED to complete the game. We need Rika and wren at at a certain level obviously, we need Chaz to be able to survive hits from megid (think he only needs level 29 or something?) we at the very least need a 33 Kyra and I THINK we need rune to at least have nafoi (is the air castle possible without nafoi?) Also, getting the pulsevucan cant be optimal I don't think if the run is heading where I think it is (hardly any grinding). Very interesting.. it just shows how far off optimal the run is still.

I actually had an insane run with the current strats yesterday that was 3:21:17, the ending was perfect. The only difference was it was slightly more streamlined grinding wise. I leave the weapons plant a couple of battles after 27 for rika and grind out the rest on the way to the triplets, and I also fight a few battles in the g tower (4x zombie guys are really worth it, along with a few other battles) on the way to dark force 2. I also had a really good vahal fort section that I ended early and grinded a little bit on the way to dark force 3. The ending third of the game was only about 55 minutes. I think that time will be hard to beat with these strats but thankfully it seems they're far from optimal..

edit: havent done any testing yet but just thinking about things for the air castle I think we need chaz to be able to survive anothergate (lvl 22 I think) rune to have nafoi (25 is it? )and we might be able to get away with wren level 23ish. If this is right I'm confident the weapons plant can be cut out entirely.

That makes a lot of sense with killing off characters for dark force 1. I've always at least tried to kill off raja in that fight, but killing off other people didn't really occur to me, and that also opens up if you should kill off characters for the triplets fight and every other fight after that. It all comes down to what do we NEED to complete the game. We need Rika and wren at at a certain level obviously, we need Chaz to be able to survive hits from megid (think he only needs level 29 or something?) we at the very least need a 33 Kyra and I THINK we need rune to at least have nafoi (is the air castle possible without nafoi?) Also, getting the pulsevucan cant be optimal I don't think if the run is heading where I think it is (hardly any grinding). Very interesting.. it just shows how far off optimal the run is still.

I actually had an insane run with the current strats yesterday that was 3:21:17, the ending was perfect. The only difference was it was slightly more streamlined grinding wise. I leave the weapons plant a couple of battles after 27 for rika and grind out the rest on the way to the triplets, and I also fight a few battles in the g tower (4x zombie guys are really worth it, along with a few other battles) on the way to dark force 2. I also had a really good vahal fort section that I ended early and grinded a little bit on the way to dark force 3. The ending third of the game was only about 55 minutes. I think that time will be hard to beat with these strats but thankfully it seems they're far from optimal..

edit: havent done any testing yet but just thinking about things for the air castle I think we need chaz to be able to survive anothergate (lvl 22 I think) rune to have nafoi (25 is it? )and we might be able to get away with wren level 23ish. If this is right I'm confident the weapons plant can be cut out entirely.

squidfeatures

Edit history:

squidfeatures: 2015-09-30 02:02:34 am

2015-09-30 02:01:49 am

Great run Adgey, I'm glad I was able to catch the end of it live. It was also good to see you in the channel when I was reviewing the run today.

Anyway, as some of you know I've been working on a guide document to serve as a repository of our combined knowledge of the game. I was going to share a link today, but I changed my mind at the last minute since when I uploaded the offline version I had the table formatting got really wonky and in any case it's in a weird hybrid state of part original work, part Jiseed's run and part Adgey's run. I hope to have everything sorted out with the formatting by next weekend and I'll post a link here.

My end goal is to have a whole folder worth of stuff, including actual spreadsheets of the data from http://www.wrftpb.com/psiv/ so you could filter enemies by resistances or whatever, as well as dungeon maps with routes.

When I do end up sharing the docs I'll enable comments so you can add ideas directly into the document for me or others to try out.

Ok, I changed my mind again while typing this. I am going to share the doc in all it's horrible formatting and frankensteiny quality.

https://goo.gl/sIU5Uo

The guide is based on Jiseed's 1.0 document (you can find the link up thread) Although the formatting is horrible in the Motavia section, the guide is basically ok. There is a bit of a hiccup in quality around the end of Motavia after Zio because I got tired of Motavia and skipped Adgey's Motavia section so the macro names may be a little confusing in the transition from Zio to the Fissure of fear. From Dezolis up until you get the Silver Tusk the formatting and info are good and follow Adgey's most recent run pretty closely. After the tusk however, the guide is identical to Jiseed's 1.0 doc.

Any and all feedback is appreciated.

Great run Adgey, I'm glad I was able to catch the end of it live.  It was also good to see you in the channel when I was reviewing the run today.

Anyway, as some of you know I've been working on a guide document to serve as a repository of our combined knowledge of the game.  I was going to share a link today, but I changed my mind at the last minute since when I uploaded the offline version I had the table formatting got really wonky and in any case it's in a weird hybrid state of part original work, part Jiseed's run and part Adgey's run.  I hope to have everything sorted out with the formatting by next weekend and I'll post a link here.

My end goal is to have a whole folder worth of stuff, including actual spreadsheets of the data from http://www.wrftpb.com/psiv/ so you could filter enemies by resistances or whatever, as well as dungeon maps with routes.

When I do end up sharing the docs I'll enable comments so you can add ideas directly into the document for me or others to try out.

Ok, I changed my mind again while typing this.  I am going to share the doc in all it's horrible formatting and frankensteiny quality.

https://goo.gl/sIU5Uo

The guide is based on Jiseed's 1.0 document (you can find the link up thread) Although the formatting is horrible in the Motavia section, the guide is basically ok.  There is a bit of a hiccup in quality around the end of Motavia after Zio because I got tired of Motavia and skipped Adgey's Motavia section so the macro names may be a little confusing in the transition from Zio to the Fissure of fear. From Dezolis up until you get the Silver Tusk the formatting and info are good and follow Adgey's most recent run pretty closely.  After the tusk however, the guide is identical to Jiseed's 1.0 doc.

Any and all feedback is appreciated.

Taloof

2015-09-30 12:38:50 pm

Excellent job Squid!

squidfeatures

2015-10-05 03:17:37 am

I did a little bit of work on the doc today, got from the start all the way to Termi, so the only "updated but ugly" section is between Termi and Aiedo after the second Zio fight.

I also made the spreadsheet I mentioned in my last post: https://goo.gl/WlPrbt

This sheet has the ROM dump data from http://www.wrftpb.com/psiv/ and put it in a google spreadsheet so you can sort and filter, makes things a little easier for looking up enemy stats and abilities.

Jiseed

2015-10-06 05:45:44 pm

Very impressive time Adgey, I didnt think the time would get anywhere close to that with the old strats - the RNG factor seemed like too much to overcome. That time is 3 minutes behind my (old strats) sum of best, which considering the route changes and what not that had taken place seems pretty close to a RNG god run. Hope you bought some lottery tickets after!

The document looks awesome Squid. I was going to update my old route with some of the new stuff, but I think the documentation of the route is in much better hands with you at this point.

Now onto some of my findings. If theres a desperate need for time to be saved, we can skip the entirety of the Wreckage and simply head to Aiedo, grab the quest and head into the Zol Slug cave to grind. We miss out on the Crmcshield and Armor like this, but all the fights are doable if we buy a graphite shield for Rika. This is extremely risky, I think I succeeded in defeating Sandworm one time in twenty, so a 5% success rate. I estimate it saves around 2-3 minutes with some good luck though. Might be worth it if you have an abysmal Bio-plant and the run is tanking hard though.

One small timesaver that is easy to implement is not getting a Reflect Shield for Rika in Jut, and instead using the left over Psy-Shield until we get the Reflect from Seth. This allows us to do all of our armor/weapon shopping in Jut the first time and skip the revisit. Trimates aren't really an issue as long as you're picking them up along the way, so no need to go to the item shop either. The defense difference is pretty marginal between Reflect and Psy, so its not a big issue in the end. Just remember to keep the Psy-Shield for Kyra at endgame. Additionally, I had forgotten about the extra Star Dew on the DF2 level of Garuberk tower, so if you want to speed up the Lashiec fight by using some Star Dews, theres an extra one thats fairly close to the boss.

Some notes on levels: 24 Wren on DF2 seems to almost certainly guarantee death to both Wren and Kyra from Shadowbreath, even with the best armor available equipped. I guess the boost he gets to Barrier at 25 pushes it over the edge to make it manageable. I may just have been getting extremely unlucky though, its been known to happen. Leaving Vahal Fort after getting all the items at 27/28 with Wren is pretty good too, I'd veer more on the side of 28 due to less micromanagement in the DF3 fight, but its usually not hard to hit it with the small grind leading up to the temple. Lastly, 29 Chaz at PD3 is certainly viable, but overall its much, much riskier than 30. I think I had around a 20-30% Megid mortality rate at 29, and Rune wasn't doing too great either, usually living with single digit hp. 30 seemed to lighten the load quite a lot, and generally you'll level up to that naturally through the endgame boss fights/small grinding in the towers so maybe its not much of an issue. 33 Kyra is still solid, she has all the casts of Warla as she does at 34, and her speed isnt too terrible... a bad shadowbreath might kill her in the first phase though.

Lastly some thoughts: Not going to the weapon plant to at least get Wren's Armor is almost certainly going to make him die every time to Thundershower, not to mention the Triplet's combo move (You lose around 14 magic defense over default, and 9 over the equipment gotten in Kuran). So I think its probably for the best to leave it in for now, if just for the DF2 fight. Skipping the Pulse Vulcan is pretty doable, it increases the risk of death in Vahal Fort noticably though. Killing off Rune in the DF1 fight and having Rika+Wren only be alive seems to be the way to go. You can balance this out slightly by having Wren die in the Triplets fight (no recover in the macro), which actually saves a marginal amount of time... you do have to make two macros though (We still need Wren for Lashiec after all). Last but not least, be aware of the dreaded Rykros softlock, which happened to me in my last run. I didn't think it actually existed, but it does. I hope we can figure out what causes it so we don't have to deal with some random BS at the end ruining the run.

Very impressive time Adgey, I didnt think the time would get anywhere close to that with the old strats - the RNG factor seemed like too much to overcome. That time is 3 minutes behind my (old strats) sum of best, which considering the route changes and what not that had taken place seems pretty close to a RNG god run. Hope you bought some lottery tickets after!

The document looks awesome Squid. I was going to update my old route with some of the new stuff, but I think the documentation of the route is in much better hands with you at this point.

Now onto some of my findings. If theres a desperate need for time to be saved, we can skip the entirety of the Wreckage and simply head to Aiedo, grab the quest and head into the Zol Slug cave to grind. We miss out on the Crmcshield and Armor like this, but all the fights are doable if we buy a graphite shield for Rika. This is extremely risky, I think I succeeded in defeating Sandworm one time in twenty, so a 5% success rate. I estimate it saves around 2-3 minutes with some good luck though. Might be worth it if you have an abysmal Bio-plant and the run is tanking hard though.

One small timesaver that is easy to implement is not getting a Reflect Shield for Rika in Jut, and instead using the left over Psy-Shield until we get the Reflect from Seth. This allows us to do all of our armor/weapon shopping in Jut the first time and skip the revisit. Trimates aren't really an issue as long as you're picking them up along the way, so no need to go to the item shop either. The defense difference is pretty marginal between Reflect and Psy, so its not a big issue in the end. Just remember to keep the Psy-Shield for Kyra at endgame. Additionally, I had forgotten about the extra Star Dew on the DF2 level of Garuberk tower, so if you want to speed up the Lashiec fight by using some Star Dews, theres an extra one thats fairly close to the boss.

Some notes on levels: 24 Wren on DF2 seems to almost certainly guarantee death to both Wren and Kyra from Shadowbreath, even with the best armor available equipped. I guess the boost he gets to Barrier at 25 pushes it over the edge to make it manageable. I may just have been getting extremely unlucky though, its been known to happen. Leaving Vahal Fort after getting all the items at 27/28 with Wren is pretty good too, I'd veer more on the side of 28 due to less micromanagement in the DF3 fight, but its usually not hard to hit it with the small grind leading up to the temple. Lastly, 29 Chaz at PD3 is certainly viable, but overall its much, much riskier than 30. I think I had around a 20-30% Megid mortality rate at 29, and Rune wasn't doing too great either, usually living with single digit hp. 30 seemed to lighten the load quite a lot, and generally you'll level up to that naturally through the endgame boss fights/small grinding in the towers so maybe its not much of an issue. 33 Kyra is still solid, she has all the casts of Warla as she does at 34, and her speed isnt too terrible... a bad shadowbreath might kill her in the first phase though.

Lastly some thoughts: Not going to the weapon plant to at least get Wren's Armor is almost certainly going to make him die every time to Thundershower, not to mention the Triplet's combo move (You lose around 14 magic defense over default, and 9 over the equipment gotten in Kuran). So I think its probably for the best to leave it in for now, if just for the DF2 fight. Skipping the Pulse Vulcan is pretty doable, it increases the risk of death in Vahal Fort noticably though. Killing off Rune in the DF1 fight and having Rika+Wren only be alive seems to be the way to go. You can balance this out slightly by having Wren die in the Triplets fight (no recover in the macro), which actually saves a marginal amount of time... you do have to make two macros though (We still need Wren for Lashiec after all). Last but not least, be aware of the dreaded Rykros softlock, which happened to me in my last run. I didn't think it actually existed, but it does. I hope we can figure out what causes it so we don't have to deal with some random BS at the end ruining the run.

squidfeatures

Edit history:

squidfeatures: 2015-10-07 04:58:36 am

squidfeatures: 2015-10-07 04:57:56 am

2015-10-07 04:56:01 am

Thanks Jiseed, I'm glad you found the guide interesting. I updated it some today as well, the formatting is fixed up to Myst Vale. The run is still changing quite a bit in the details, so I'll keep the guide aligned mostly with your and Adgey's most recent runs, since we actually have confirmed times for those. I'll try to leave as much flexibility in as possible, briefly explaining alternate routes, or equip choices (like you mention above, skipping Wreckage, dual shield Rika in sand worm fight, etc.)

And just a reminder to everyone, anyone with the link has comment privileges on the doc, so if you have an idea or want me to test something. or even a question as to why something is in the guide, make a comment and I can incorporate it.

Also, I've been making some progress plumbing the depths of the RAM. I've discovered the RAM locations for enemy stats, which isn't exactly the most helpful since we have all of that info in the ROM dump, but I plan to use it to see what happens with Lashiec Reinforce and if PSYCO-WAND has other effects we may not know.

Some of my other goals are find how encounters are determined and if there is any way to manipulate them, and also figure out what combination of stats (if any) determine running from battle success.

Thanks Jiseed, I'm glad you found the guide interesting.  I updated it some today as well, the formatting is fixed up to Myst Vale.  The run is still changing quite a bit in the details, so I'll keep the guide aligned mostly with your and Adgey's most recent runs, since we actually have confirmed times for those.  I'll try to leave as much flexibility in as possible, briefly explaining alternate routes, or equip choices (like you mention above, skipping Wreckage, dual shield Rika in sand worm fight, etc.)

And just a reminder to everyone, anyone with the link has comment privileges on the doc, so if you have an idea or want me to test something. or even a question as to why something is in the guide, make a comment and I can incorporate it.

Also, I've been making some progress plumbing the depths of the RAM.  I've discovered the RAM locations for enemy stats, which isn't exactly the most helpful since we have all of that info in the ROM dump, but I plan to use it to see what happens with Lashiec Reinforce and if PSYCO-WAND has other effects we may not know.

Some of my other goals are find how encounters are determined and if there is any way to manipulate them, and also figure out what combination of stats (if any) determine running from battle success.

squidfeatures

Edit history:

squidfeatures: 2015-10-12 03:20:44 am

2015-10-12 03:20:28 am

I made some interesting, albeit mostly useless pieces of info about encounters. It's kind of hard to explain so I'll just simplify it a bit, but basically if you are in a dungeon, as long as you are moving, there is something like an invisible counter that is counting down the steps until your next encounter. It doesn't matter where you move, as long as you are moving, the counter counts down. As soon as you stop moving, even if only for a single frame, the counter changes it's value to some other random number, and starts counting down as soon as you start moving again.

Another interesting thing is that as long as you are moving and the counter is counting down, what buttons you press affect the group you fight. So if hold down "right" and move right until you're encounter you might get group A, but if you hold "left" and move left, you'd get group B, or if you hold down "left" for a while, then "right" you might get group "C", but if you were use the exact same inputs you get same fight.

I've made a video that has some examples.

So far it's pretty much useless for a speed run, except for as I mention in the video, it might be possible to increase the chances of encountering a fight, depending on the distribution of "steps to fight". I haven't found the actual trigger conditions for encounters yet so I can't know that yet. It might even be possible to affect groups you get in battle, since holding the "B" button does change encounters without stopping the game, which could affect the distribution of fights in a useful way.

I'll post if I find anything new.

I made some interesting, albeit mostly useless pieces of info about encounters.  It's kind of hard to explain so I'll just simplify it a bit, but basically if you are in a dungeon, as long as you are moving, there is something like an invisible counter that is counting down the steps until your next encounter. It doesn't matter [i]where[/i] you move, as long as you are moving, the counter counts down.  As soon as you stop moving, even if only for a single frame, the counter changes it's value to some other random number, and starts counting down as soon as you start moving again.

Another interesting thing is that as long as you are moving and the counter is counting down, what buttons you press affect the group you fight. So if hold down "right" and move right until you're encounter you might get group A, but if you hold "left" and move left, you'd get group B, or if you hold down "left" for a while, then "right" you might get group "C", but if you were use the exact same inputs you get same fight.

I've made a video that has some examples. 
[url]https://youtu.be/rUwaBsBSSUs[/url]

So far it's pretty much useless for a speed run, except for as I mention in the video, it might be possible to increase the chances of encountering a fight, depending on the distribution of "steps to fight". I haven't found the actual trigger conditions for encounters yet so I can't know that yet.  It might even be possible to affect groups you get in battle, since holding the "B" button does change encounters without stopping the game, which could affect the distribution of fights in a useful way.

I'll post if I find anything new.

MrStarbird

2015-10-22 02:53:31 pm

Leader of Starbirdia!

I stop looking for a month and all this has happened?! Wow! Nice work everybody. I wish I had something to contribute to this but you guys seem to have things going smoothly.

squidfeatures

2015-10-23 09:42:43 pm

I don't know if you noticed, but Jiseed recently got a run in 3:17:XX, so the route is coming along quite nicely. A tight run all around.

I know that I don't need to make excuses for myself, but I recently got a new job, and it's completely mixed up my schedule, so it's going to be a while before I'll be able to stream or do any other reverse engineering stuff.

On the plus side however, I've been in contact with Lorenzo who posted the following disassembly of the PSIV rom http://www.romhacking.net/documents/697/ (you can follow another thread about this at http://www.pscave.com/forums/viewtopic.php?f=6&t=3415) In the near future I plan on working with him on the disassembly, but for now he helped me out quite a bit in figuring out exactly what determines random encounters.

The bad news is that there isn't much we can do at the moment to exploit it. But here is basically what happens when the game tries to do a random battle:

(if any of these checks are false, the battle doesn't happen and we try the check later)
1. Check if characters are not in a battle flagged zone (such as a town)
2. Check if characters are in a scripted sequence (such as when the party re-arranges itself after meeting Rune in Ladea tower)
3. Check if we are stepping on a special square (such as stairs, or a healing tile)
4. Check if we are next to a tile that can cause a map change (such as a town or stairs, but not doorways)
5. Check if we are idle (i.e. not in the act of moving to a new tile, more on this later)
6. Clear the "moving" bit, and check if we weren't already idle the previous check. (no fights if you are just standing still)
7. Decrease the timer, and check if it is equal to 0 (timer is set to 10 after battles and changing maps, so you effectively have at least 10 "steps" between battles) then set timer to 1

If we've made it this far (every check has been true) the game generates a new random number between 0-31 (walking) or 0-127 (vehicle). If the number is 0, a fight happens.

Just a quick explanation of "moving to a tile." If you've played the game, you've noticed that you only have to tap a direction, and your crew will continue happily on their way to the next tile in that direction. There is a little timer that starts when you press a direction and counts down to zero, and when it reaches zero, you are "at" the next tile. Even if you are holding a direction, as soon as the counter reaches zero, for a single frame you are "not moving to a new tile" Fun fact: this "vulnerable" frame is 2 frames before your characters sprites stop moving.

I've looked at the assembly for the random number generator, and it basically does some magic with the contents of two registers, which means if you could control the contents of those registers you could control the RNG. Unfortunately the calls to the battle checker routine happen between frames, so it would take some further digging in the assembly before we could find a way to use it. I plan on abandoning this line of exploration for now. I do plan on looking a little into how the enemies are chosen for a battle, but since the encounter tables are encoded, I may skip this for now too.

Anyway, it's easy to see why the behavior in the video I posted happens: if you wait a single frame, the number you get when you reach the original "battle" tile isn't the same, so the battle is avoided.

The only useful thing from this is if you are grinding for battles, stay away from stairs and towns, don't change maps a lot. Stutter stepping might help, but the RNG seems random enough that it likely doesn't matter. Also if you want to avoid a battle, drive, or change maps. I don't know how much time it would save to go up and down stairs fast to avoid a "potential" battle. The only place I can think of that is so dangerous that avoiding fights is worth that time is the Edge (maybe Garuberk Tower?), and it isn't even possible in the Edge since there are no mid-path map change potential. I wonder if that was intentional or a byproduct of their design choices.

Since I'm satisfied that the encounter stuff is a bit of a dead end for now, I'm going to focus on battle mechanics stuff. I was watching one of Adgey's more recent attempts and in the PD fight he got some pretty weird damage numbers on characters who had been dead then revived and re-buffed. Adgey also mentioned he saw something like that happen to Jiseed as well, so I think I may focus on seeing what goes on when characters die. I can tell this already, dead characters keep their buffed stats while they are dead. I don't know when the stats are lowered after revive, but it won't be hard to check.

Also, now that we have at least two good runs (that I know of) I'm going to try to get the route document up to par. That'll be my priority before the mechanics stuff. Let me know if you find someone who runs the game and has a decent time or has a novel strategy or style choice.

Thanks to the anonymous editor who made the suggestions to the document earlier this week, they were most helpful.

Anyway, great fun everyone, and I hope I can get my schedule sorted out enough so I can start streaming again!

I don't know if you noticed, but Jiseed recently got a run in 3:17:XX, so the route is coming along quite nicely. A tight run all around.

I know that I don't need to make excuses for myself, but I recently got a new job, and it's completely mixed up my schedule, so it's going to be a while before I'll be able to stream or do any other reverse engineering stuff.

On the plus side however, I've been in contact with Lorenzo who posted the following disassembly of the PSIV rom [url]http://www.romhacking.net/documents/697/[/url] (you can follow another thread about this at [url]http://www.pscave.com/forums/viewtopic.php?f=6&t=3415[/url]) In the near future I plan on working with him on the disassembly, but for now he helped me out quite a bit in figuring out exactly what determines random encounters.

The bad news is that there isn't much we can do at the moment to exploit it.  But here is basically what happens when the game tries to do a random battle:

(if any of these checks are false, the battle doesn't happen and we try the check later)
1. Check if characters are not in a battle flagged zone (such as a town)
2. Check if characters are in a scripted sequence (such as when the party re-arranges itself after meeting Rune in Ladea tower)
3. Check if we are stepping on a special square (such as stairs, or a healing tile)
4. Check if we are next to a tile that can cause a map change (such as a town or stairs, but not doorways)
5. Check if we are idle (i.e. not in the act of moving to a new tile, more on this later)
6. Clear the "moving" bit, and check if we weren't already idle the previous check. (no fights if you are just standing still)
7. Decrease the timer, and check if it is equal to 0 (timer is set to 10 after battles and changing maps, so you effectively have at least 10 "steps" between battles) then set timer to 1

If we've made it this far (every check has been true) the game generates a new random number between 0-31 (walking) or 0-127 (vehicle).  If the number is 0, a fight happens.

Just a quick explanation of "moving to a tile." If you've played the game, you've noticed that you only have to tap a direction, and your crew will continue happily on their way to the next tile in that direction.  There is a little timer that starts when you press a direction and counts down to zero, and when it reaches zero, you are "at" the next tile.  Even if you are holding a direction, as soon as the counter reaches zero, for a single frame you are "not moving to a new tile"  Fun fact: this "vulnerable" frame is 2 frames before your characters sprites stop moving.

I've looked at the assembly for the random number generator, and it basically does some magic with the contents of two registers, which means if you could control the contents of those registers you could control the RNG. Unfortunately the calls to the battle checker routine happen between frames, so it would take some further digging in the assembly before we could find a way to use it.  I plan on abandoning this line of exploration for now. I do plan on looking a little into how the enemies are chosen for a battle, but since the encounter tables are encoded, I may skip this for now too.

Anyway, it's easy to see why the behavior in the video I posted happens: if you wait a single frame, the number you get when you reach the original "battle" tile isn't the same, so the battle is avoided.

The only useful thing from this is if you are grinding for battles, stay away from stairs and towns, don't change maps a lot.  Stutter stepping might help, but the RNG seems random enough that it likely doesn't matter.  Also if you want to avoid a battle, drive, or change maps.  I don't know how much time it would save to go up and down stairs fast to avoid a "potential" battle. The only place I can think of that is so dangerous that avoiding fights is worth that time is the Edge (maybe Garuberk Tower?), and it isn't even possible in the Edge since there are no mid-path map change potential.  I wonder if that was intentional or a byproduct of their design choices.

Since I'm satisfied that the encounter stuff is a bit of a dead end for now, I'm going to focus on battle mechanics stuff.  I was watching one of Adgey's more recent attempts and in the PD fight he got some pretty weird damage numbers on characters who had been dead then revived and re-buffed.  Adgey also mentioned he saw something like that happen to Jiseed as well, so I think I may focus on seeing what goes on when characters die.  I can tell this already, dead characters keep their buffed stats while they are dead.  I don't know when the stats are lowered after revive, but it won't be hard to check.

Also, now that we have at least two good runs (that I know of) I'm going to try to get the route document up to par.  That'll be my priority before the mechanics stuff. Let me know if you find someone who runs the game and has a decent time or has a novel strategy or style choice.

Thanks to the anonymous editor who made the suggestions to the document earlier this week, they were most helpful.

Anyway, great fun everyone, and I hope I can get my schedule sorted out enough so I can start streaming again!