There's a guide on gamefaqs that shows how to play through the game low enough level to get all the enemies for rages while keeping a very low level.
I actually followed that guide a long time ago so I could get everyone to lv 99 with esper boosted stats. From what I remember, you have to grind a lot of Locke steals to afford halfway decent equipment. The first half of the game was hell, but suddenly became a lot easier when you got equipment that would grant a unit 128 magic avoid, since they would become de-facto invincible.
Come to think of it, a low-level run TAS by itself would be interesting to watch if it can manipulate away most of the grinding done in RTA.
First off, you can rule out getting max stats. That would require doing a low-level run until you get the best stat-boosting espers, and then you still have to make an arbitrary decision about which stats you want to focus on (vigor, mag, spd, etc). Also, even fighting nothing but double tyrannosaurs, it takes HOURS to get everyone to lv 99.
As for the other things you listed, my thoughts:
At least 1 of every item- Very reasonable
Perfect Stats - Forget about it; already explained why
All Treasures - How is this different from getting 1 of every item?
All Characters - Definitely
All Levels - Hell no
All Spells - For every character? Absolutely not
All Rages - Are you insane? That must be an hour of veldt grinding alone
All Lores - Sure
All Dances - Sure
But in my own view I imagine there's some way to overflow the items back to 0 anyway, using some bug. (All hypothetical; I don't really know. But the mere thought of that probably existing makes me think that it sounds better to get at least 1 of each item, in order to fill up the entire item list obviously, instead of wasting away at 99 items, of which some can't even be obtained as such.)
You can't just hope that there's an overflow/underflow glitch that would allow that to happen. You either need to understand the inner workings of the game and discover it yourself, or assume it doesn't exist until someone else actually finds it.
Yeah, that was confirmed by me and Sam6555 when he streamed Path of Radiance prologue TASing.
Well, looks like I've hit another roadblock: When TASing on dolphin my frame advance... Takes very long just to advance 1 frame and savestates take forever to load. It's so painful to work through that I can't even make a decent TAS of prologue, so all signs point to having to get a more powerful machine. I'm going to look towards getting a motherboard that supports an i5 or i7 processor, but that'll take a while since I'm strapped for cash.
Finally managed to emulate this and Path of Radiance, close to 60 fps in most cases too.
Unfortunately, after a couple quick tests I'm rather convinced the RNG is not affected by path drawing in this game. Nor does switching weapons, talking, shoving, trading, dropping items, rescuing, ending a character's turn (ie graying them out), or moving other characters around.
Strangely enough, having Micaiah and Sothe converse did change the outcome. Perhaps because the game registered an increase in support value? Vigoring also changed the RNG.
I can't really do a serious run yet without knowing the inner working of the RNG, which I'll leave to more capable hands than I. In the meanwhile, I'll try and whip up a dtm of 1-P and see if I can make a halfway decent encode.
A couple of years ago I made a series of what I like to call “warpskip” videos- videos where I have maxed out mag units use the warp staff repeatedly in a chapter to achieve unbelievably low turn counts. In fact, with the exception of C21 (2 turns) and C24 (5 turns), I’ve managed to 1 turn every single chapter in the game where you have access to the warp staff. I’ve gone to painstaking efforts to not only restore all the missing warpskip videos, but also make them slightly better in various ways, such as acquiring treasures in C20 and better optimizing my trades.
I’ve split the videos into 3 parts: One for Ilia route, Sacae Route, and the final chapters.
Link to videoLink to videoLink to videoVbm links
To play the vbms (in fact any FE vbm I’ve ever made), you must use VBA Re-Recording V23.5.
Ilia Route Chapters (+C16 & 16x)
Chapter 16Chapter 16xChapter 17Chapter 18Chapter 19Chapter 20Chapter 20x
Sacae Route Chapters
Chapter 17 SacaeChapter 18 SacaeChapter 19 SacaeChapter 20 SacaeChapter 20x Sacae
Final Chapters (21-24)
Chapter 21Chapter 21xChapter 22Chapter 23Chapter 24
General Notes
The purpose of these videos is to show the absolute minimum turn count that can be achieved without regards to run continuity. That means I could spend 1000 turns arena abusing my units in previous chapters if I were so inclined, so long as it helps achieve my goals for the chapter in question. Instead of spending hours arena abusing, I use codebreaker codes to give my units realistic stats (usually average stats in everything except str, skl and spd, which get maxed out). I am very vigilant about ensuring that my hacks don’t go beyond what’s actually possible in normal play. That means not using items I shouldn’t have been able to acquire yet, not using characters I shouldn’t have been able to recruit yet, not giving units stats that are impossible for their level, and so on.
A major obstacle in this challenge is a matter of spacing. Often I have to warp a unit onto an exact tile to make a strategy work; if any given map were one row or column bigger, some of these strategies would probably be 2 turns. Another spacing issue involves trades. One of the most important actions performed in these challenges is to have a magic user trade for the warp staff and warp another unit who hasn’t yet moved, all in the same action. Not only that, I need to chain together multiple trade-warp actions in a row, which creates a sequence of trades that is hell to optimize. Another exploit I use is to force a rescue staff user to place a unit on a certain tile by blocking other tiles adjacent to the rescue user with other units.
Warp staff range is (mag / 2) + 5. Up until C21x, Lilina is my best warper with a 30 mag cap (20 range), followed by Sophia, Ray, Lugh and Hugh (28-29 cap, 19 range), Clarine, Cecilia and Ellen (26-27 cap, 18 range), and finally Saul, Yodel and Niime (17 range or less). Based on this difference in warp range, I usually assign Lilina the most important warp tasks (the ones that require at least 20 range) whereas units like Saul get relegated to less important tasks, such as using the hammerne to repair the warp staff. After C21x, I get the Apocalypse tome, a dark tome that gives a +5 mag boost. This makes Sofiya my best warper with 34 mag (22 range), although the added range is only helpful on Roy’s warp in C23.
Speaking of chapters 22 to 24, you’ll notice all my units have 15 movement. This is due to the fact that in C21, there is a secret shop with buyable stat boosters, boots, purge and boltings. In theory I have access to an unlimited money supply through arena abuse, which means I should be able to buy as many stat boosters as I desire. This is also why Lalum barely gets hurt from a mamkute attack in C24; I maxed out her hp and def on the basis that I could give her as many seraph robes and dracoshields as I want.
An unfortunate aspect about TASing this is that the trades are so fast you can barely notice what’s going on. For that reason I will do my best to explain everything that’s going on in each chapter below:
Chapter 16
• Lilina warps Percival within striking range of the boss
• Dieck trades for the warp
• Clarine takes the warp, does an intermediate warp for Roy
• Lalum dances Clarine
• Clarine warps Ray beneath Roy
• Thany trades for the warp, re-moves besides Ray
• Ray trades for the warp, warps Roy to within seizing distance
• Percival critkills the boss
• Roy seizes
Chapter 16x
• Lilina warps Thany to the boss
• Sophia rescue staffs Roy
• Lalum dances Lilina
• Lilina warps Roy
• Thany critkills the boss
• Roy seizes
Chapter 17
• Lilina warps Clarine to within bolting range of the boss
• Clarine kills the boss (she doesn’t have a critrate)
• Saul trades for the warp, warps Roy a very short distance
• Lance trades for warp, re-moves adjacent to Dieck
• Sophia trades for warp, warps Cecilia (who has the rescue staff)
• Dieck trades for warp
• Lugh trades for warp, warps Roy (still an intermediate warp)
• Ray uses the hammerne to repair the warp
• Lalum dances Lugh (he currently holds the warp staff)
• Cecilia rescue staffs Lugh. Note that the cavalier left of Cecilia forces her to rescue Lugh to her right, which is crucial in terms of spacing.
• Lugh warps Roy to the throne
• Roy seizes
Chapter 18
• Sophia intermediate warps Lugh (who’s carrying bolting and will kill the boss)
• Tate trades for the warp, re-moves down-left of Lugh
• Ray trades for the warp, warps Lugh to within the boss’ range
• Miledy trades for the warp, re-moves 2 spaces above Saul
• Saul trades for the warp, warps Hugh (who has the rescue staff)
• Lance trades for the warp, re-moves adjacent to Lilina
• Lilina trades for the warp, intermediate warps Roy
• Ellen hammernes the warp staff
• Lalum dances Lilina (also uses the double occupancy glitch)
• Hugh rescues Lilina (who’s currently holding the warp staff)
• Lilina warps Roy to within seizing distance
• Lugh kills the boss
• Roy seizes
I won’t continue with these move-by-move chapter descriptions since they’re exhausting, but they should give you a pretty good idea how the strats generally work. For instance, when I have to warp someone so far they need 2 warps to reach their destination, what I usually do is the following: Use an intermediate warp on the desired unit, warp someone with the rescue staff, dance the warper, rescue staff the warper, then have the warper warp the desired unit again.
Also, bolting kills on bosses are very common since chapter spacing really is that strict. It’s fairly common that I use every single unit slot I possibly can as well. In fog of war maps, you have to actually see the boss to reveal him (or her), and you can only warp as far as you can see. For this reason, I actually have to warp thieves in C17S and C19I. In fact, the strategy for C17S is so farfetched I have to use a unit sacrifice at the end just to make it work.
Well, that just about covers everything. If you still find the video too fast to follow, I recommend playing through the vbm on frame advance, particularly in trade heavy sequences. Oh, I forgot to mention, a lot of the characters I use are interchangeable, especially people who are just trade bots like Dieck. Anyway, hope you enjoy :)
I TASed the remaining FOW chapters, including C19xx (Hector mode only).
Link to video
Vbm links:
Chapter 18Chapter 20Chapter 23Chapter 26Chapter 19xx
Notes:
Chapter 18 (3279 frames saved, ~54.65 secs) - I couldn’t find any improvements compared to what Roland did for this chapter.
Chapter 20 (1005 frames saved, ~16.75 secs) - The only advantage of using the torch here is to end turn 1 EP prematurely.
Chapter 23 (601 frames saved, ~10.02 secs) - Marcus kills the mercenary 2 spaces above him because he’s the first enemy to move on EP.
Chapter 26 (7125 frames saved, ~2 mins) - This is where things get interesting. Basically the strategy boils down to controlling the enemies each turn and ending the turn as soon as possible. It’s not quite that simple, however, since the enemies that move first are placed fairly deep into the map, hence why I have to send Marcus chasing after them. I also have to acquire 2 chests, the boots and the rescue staff. What’s interesting is that if I load up an enemy thief’s inventory and acquire a chest item, the thief will send an item to Merlinus. This very exploit is used to acquire the rescue staff. Also, the first enemy phase is so complicated I figured I’d break it down by individual moves:
• Move a mercenary out of Marcus’ way
• Nove another mercenary 2 spaces down (forces an armor to attack Marcus from the left)
• Have 2 mages run away from Jaffar
• Have a hero rescue an archer (the hero is placed such as to force Jaffar to attack the nearby mercenary from below)
• Move another 2 enemies out of Jaffar’s range
• Have a bishop give all his items to a thief
• Advance said thief towards the rescue staff chest
A lot of the enemy placement I do involves forcing an enemy to attack from the particular tile I torched, as this allows me to trigger the glitch. That’s why, for instance, I move an archer above Florina on turn 3 EP. By turn 5 I’m basically spamming end turn since the chests are acquired and I have trapped the first enemy that moves into attacking Florina from a torched spot. Interestingly enough, I have to re-torch that spot on turn 13 in order to keep using the glitch.
Chapter 19xx- On turn 2, I deliberately place Priscilla in between an archer and an armor in order to force another armor to attack her from above. Interestingly enough, I manage to beat the chapter without even injuring a single enemy.
---
The total time saved adds up to about 12010 frames or roughly 3 minutes, 20 seconds.
I also updated the Sacred Stones torch glitch video to include C19 (see here).
If you don't have the patience to find some of the resources yourself (just look on the emulators tab on the main site!) then I really doubt you'll beat Saturn's TAS.
Spent a good 2 hours looking for new PWASE, the only leads I could find are desert forts and desert bones in C15, as they generate the correct amount of lag but don't allow the control enemy glitch as far as I can tell. I did make some useful findings, though, mainly that there are usually 24-25 frames of lag between the exp counter vanishing and the enemy greying out, and PWASE adds 4 extra frames of lag. I put my raw data in quotes below:
Go by first frame exp bar vanishes until first frame of greying out
Torch
29 frames (+4) (confirmed)
Snag also causes 4 frames of lag
No Torch
25 frames (confirmed)
Torch PWASE causes 4 frames of lag. Reset window is definitely larger than that though (test that now)
(at least 10 frame window)
---
C1
27 frames (confirmed)
C3
24 frames (confirmed)
C5
24 frames (confirmed)
C5x
24 frames (100% consistent)
(25 frames in certain chapters, 29 for PWASE)
PWASE CHECKS
PWASE
Desert fort (+4, doesn’t work?)
Desert bone (+4, same as above)
Non-PWASE
(Throne) floor
(Throne) pillar
(Throne) door
Plain
Bridge
Ballista
Forest
House
Mountain
Peak
Sea
Vendor
Outdoor stairs
Road
Desert
(Desert) ruins
(Desert) lake
Chest (unopened)
Chest (opened)
Untested
River
Cliff
Gate
(Throne) stairs
Fence
Village
Barrel
Gunnels
Armory
Chest (opened and unopened)
Only got the snag PWASE to work in a FOW map (narube river creature campaign)
To make any further findings, I need someone to help me isolate the memory addresses related to this glitch so that I can control enemies on non-PWASE. It's stated in the gamefaqs guide that you can control enemies even on non-PWASE but that the window of opportunity is dramatically smaller, sometimes a fraction of a frame. Obviously trying to guess those frames would take forever and likely prove a waste of time, hence why those addresses are so important.
I'd only update the run if the control enemy glitch could be used in every chapter of the game and make sub 1 hour realistic. With the new FOW strats, there's still roughly 7 minutes that need to be saved, although I have a nagging feeling that it is possible to control enemies in every chapter, it's just a matter of finding some new PWASE or looking into the extra "gray out" lag critcaluser was talking about.
Just theorycrafting, here's how the control enemy glitch could save time. Bolded what I consider noteworthy chapters:
Prologue: Not really worthwhile.
C1: Take the boss off the throne, Eirika doesn't get attacked.
C2: Wouldn’t save any time.
C3: Boss off throne, nothing significant.
C4: Would have to be used on turn 1, no point since the map is cleanly 2 turned.
C5: Could avoid a lot of combat on turns 1 and 2, although the turn count is unaffected.
C5x: A lot of time could be saved moving the enemies in the 1 space corridor out of the way, also avoid a lot of combat on the last turn.
C6: Used already.
C7: Would save a turn, ease up the manipulation (archers don’t have to attack) and would even allow Vanessa to gain more exp since she could crit the boss on turn 2 player phase.
C8: Enemy phase heavy chapter, would be worthwhile on almost every turn.
C9: Would save an enormous amount of time since I could direct the pirates towards my units instead of away and at least manage a 4 turn like in the normal mode run. At least 7493 frames (~2 mins) could be saved.
C10: About 36 secs can be saved just by ending turn 1 enemy phase prematurely. Then another ~58 secs can be saved by doing the same thing on turn 2.
C11: Used already.
C12: Doesn’t seem like much time could be saved since the spiders can only move 1 space per turn in the mountains. Lowering the turn count at all would cut down on a ton of reinforcements. Hopefully Tana can ORKO the spiders with a double crit, if not she could always be power levelled somewhere else (probably C9).
C13: Can’t save time on the 1-turn.
C14: Disarming enemies and moving them out of the way would save so much time. Could even recruit Rennac if you wanted to.
C15: A turn can definitely be saved (6 less reinforcements), possibly even 2 (another 2 less reinforcements). I would probably have to make Tana help Saleh with the troubadour kills.
C16: Same idea as C14. About 40 secs could be saved just by ending the second to last turn prematurely.
C17: Could cut down on a lot of enemy movement on turn 1.
C18: There would be surprisingly little benefit to using the control enemy glitch here.
C19: 1 turned already.
C20: Cutting out enemy phase would save ~90 secs.
Final: Cutting out enemy phase would save ~20 secs.
Tallying that up gives about 6 mins saved plus probably an extra 6+ mins for intangibles (the stuff I didn’t time estimate for). Thus, if widespread application of the control enemy glitch is possible, sub 1 hour could be achieved in theory. An insane amount of work would be needed to optimize the run properly, at least 100k re-records easily.
I’m going to spend a few hours trying to find new ways to control enemies and will stream my attempts on twitch.
Link to video
Vbm links:
Chapter 6 (1498 frames saved, ~25 seconds)
Chapter 11 (3382 frames saved, ~56 seconds)
Chapter 19 (2251 frames saved, ~38 seconds)
The time saved was better than I expected, about 2 minutes minus the time taken to purchase the torch staff. I’m also happy that I saved a turn in C6 and 3 turns in C11, which would cut my total turncount down to 89.
There are some slight drawbacks to these new chapter approaches. While Seth & Vanessa conserve weapon uses from increased critting, they also kill fewer enemies, especially in C11 where Dozla eats up so many kills. This affects their exp gain and weapon rank (eg/ it takes longer for Vanessa to hit S-rank and get that nice 5% crit boost). Then again, I have more control over Vanessa’s level-ups in C11, which would increase her stats a bit.
Edit: I added C19 since I found the 2 turn could be improved using the torch glitch. I also had a look at the phantom ship (Eph route C11), but there's really nothing you can do to prevent the bazillion enemy reinforcements that show up on turn 2 player phase. I didn't bother to make a video for that since the chapter would be incredibly boring to watch; basically an incredibly long turn 2 enemy phase of enemies suiciding into Seth and Duessel.
Good news is I finally got a gfx card and installed it. Bad news is my power supply can't seem to handle it and I might even have to upgrade other components as well, if not build a new PC from scratch.
I'd be interested in knowing what specs people managed to run this game and Path of Radiance on since I recall on the Dolphin boards that people could run those games with relatively low-end PCs.
Regardless, it might be another month or so before I can even emulate this game properly :/
Yeah, I noticed that right away from Roland's video and feel silly for not having tried to do that myself :S
By the way, the PWASE remains active as long as the torch effect is active (I mean, you don't need to use a torch every time you want to use the glitch).
Good to know, although it will be tricky to get a new enemy to land on the same square if my units are advancing.
I don't know if it's faster but you can buy a torch on serafew (I think you can't do it in the chapter but you can do it in the map after beating C5). Maybe you can kill Novala faster.
If you can get a torch for C6, you can kill Novala on turn 2 player phase and it should save considerable time.
I noticed an increase of lag frames when a unit grays out in a PWASE. I don't know if you noticed that too.
Perhaps the added lag is what allows the glitch to occur?
So for C11, I'm thinking on turn 1 you execute the glitch, move all inconveniently located enemies (eg/ bottom-right corner, top-right corner) towards Seth/Vanessa, using as many rescues as possible so that the glitch can be used again on turn 2 with as many drops. This would cut the amount of enemies moved on turn 2 as much as half. Alternatively, you could move as many 1-range units as possible towards Dozla. There's a lot of possibility. I'll give it a serious look when I'm up to it.
What about lua scripts? I didn't find a lua menu in the latest official build (v3.5) and it will almost certainly be needed for an rngdisplay script similar to the following:
Language: rngdisplay
function nextrng(r1, r2, r3)
return AND(XOR(SHIFT(r3, 5), SHIFT(r2, -11), SHIFT(r1, -1), SHIFT(r2,15)),0xFFFF)
end
function rngsim(n)
local rngbase=0x03000000
local result = { memory.readword(rngbase+4), memory.readword(rngbase+2), memory.readword(rngbase+0) }
for i = 4, n do
result[i] = nextrng(result[i-3],result[i-2],result[i-1])
end
return result
end
local phit = 0x0203A550
local pdmg = 0x0203E1BE
local pcrt = 0x0203A556
local ehit = 0x0203A5D0
local edmg = 0x0203E1BC
local ecrt = 0x0203A5D6
while true do
local nsim = 20
rngs = rngsim(nsim)
for i = 1, nsim do
gui.text(228, 8*(i-1), string.format("%3d", rngs[i]/655))
end
gui.text(210,0,"RNG1:")
gui.text(210,8,"RNG2:")
gui.text(210,16,"RNG3:")
gui.text(194,24,"Next RNs:")
gui.text(0,0,"Player")
gui.text(0,8,"Hit: " .. memory.readbyte(phit))
gui.text(0,16,"Damage: " .. memory.readbyte(pdmg))
gui.text(0,24,"Crit: " .. memory.readbyte(pcrt))
gui.text(0,40,"Enemy")
gui.text(0,48,"Hit: " .. memory.readbyte(ehit))
gui.text(0,56,"Damage: " .. memory.readbyte(edmg))
gui.text(0,64,"Crit: " .. memory.readbyte(ecrt))
emu.frameadvance()
end
According to this thread, you can run memory viewer in dolphin's debug mode by inputting Dolphin.exe /d in the command prompt (run -> cmd). I just tried it and got it to work.
In all likelihood holding B will not make the cursor any faster (should always move at most once every 2 frames) but it will cause the camera to keep up with the cursor movement, which is generally more pleasing to watch I find. Though, in some cases it's actually faster for the camera to follow the movement as little as possible as to minimize camera transition... Kind of hard to explain, I'd have to show you where it's used in Toothache's FE6 TAS.
By C-stick I assume you meant the analog/control stick since the c-stick only affects camera angles. They will probably come out to the same, but definitely test both because you never know.
Now once you get memory viewer working, it's very important you locate the address for the RNG. Isolating it will be rather tricky because you need to find an action that burns only 1 RN at a time.
First, see if you find an address that changes every time you move the cursor diagonally 1 square. I'd do this in prologue. Select Ike, move the cursor up+right in 1 frame, press b to cancel, check memory viewer, repeat. If path drawing is what affects the RNG like in the GBA games, you should be able to isolate the address(es) fairly quickly.
If that doesn't work for some reason, try landing a low% crit (maybe 3-5%) and search for a value less than the displayed crit rate. I wouldn't bother with hitrates since those use 2 RNs and level-ups will use at least 8. If you're lucky it's possible acquiring desert items only use 1 RN, though they might use a different RNG altogether like in FE6.
For the FE8 run to sync, you need to use VBA Re-Recording V23.5. I recommend you read my submission comments since a lot of the same tricks will apply to this game.
Luck manipulation will most likely be based on cursor movement, with the most RNs burned when you have a flier (or an unobstructed paladin) move the cursor diagonally as far as you can and alternate between 2 squares. To track changes in the RNG you'll have to find the RNG's address in whatever memory watch feature dolphin uses.
But instead of alternating A every other frame to start menus and stuff, how should I know when the first frame is, that I can press it. Sure you can say just try it every frame until you get it, but there's no cues to when the first frame is as the menu isn't on-screen when it's clickable, least I don't think it was.
There's frame rules behind many actions in the game that you can either write down explicitly or learn intuitively from TASing the same actions over and over again. Think selecting one menu that makes you wait x frames to access another one. Except for occasional lag, the wait will always be those x frames.
There are also visual cues you can use. So, for instance, you might be able to navigate a menu the frame before the menu appears, or on the same frame.
I suppose that's a bit of an abstract description. I recommend you watch parts of the published GBA TASes frame by frame and watch carefully for patterns, and what I wrote above will start to make more sense.
What an amazing find!
Unfortunately, having looked into the trick in depth, I don’t think it would save time in any of the currently published TASes. The problem is that you can only control enemies that are already revealed and you can’t affect the fog whatsoever on enemy phase. Combine that with the extremely limited torch range on your magic users (excluding Pent), the time taken to actually acquire torches, the length of the torch animation and various situational factors, and the trick is basically rendered useless for TASing. A real pity too, given how much more versatile torch PWASE are compared to other PWASE.
A quick chapter breakdown:
FE7
C9: Beaten before FOW appears
C18: No torch staff yet
C20: The 2 turn can’t be sped up by controlling the boss
C23: Same logic as above
C26: Would actually be useful since Pent has an impressive torch radius of 9-10 and occasionally enemies densely populate certain portions of the map.
FE8
C6: No torch staff yet
C11: Seems to be slower based on the sheer number of torch uses needed to save turns. (Vbm)
C19: Pointless since the chapter is 1 turned already.
Now as far as I know, in FE7 the torch staff can only be acquired in the C21 secret shop, which will more than waste whatever time was gained in C26 procuring more crits.
Nonetheless, the torch led me to discover a few more interesting properties of the control enemy glitch. For instance, the Control Enemy Glitch faq on gamefaqs is slightly wrong in describing how the glitch is performed. It states:
To perform the trick, you need two things. An enemy that can attack your unit, and a unit who cannot counterattack. To activate the glitch, you must end your turn, allow the enemy unit to attack your ally, then AFTER the HP bars vanish, but BEFORE the foe greys out, hard or soft reset your game.
The bit about counterattacking is inaccurate. All that matters is that the targeted enemy gets greyed out on the PWASE, which means you can hit the enemy without killing it. The stated timing is also inaccurate since you can get the glitch to work even slightly after the enemy’s greyed out.
It was also cool to find out NPCs (ie “Other Phase” units) can be controlled using the same PWASE reset exploit.
Edit: I crossed out a bunch of misinformation since it turns out you can control enemies within the fog.
Hooray for bumping a 7 month old topic.
I've spent the past month or so reading through K&R's "The C Programming Language". I was actually managing pretty well until I got to the chapter on pointers. Once I got to a section that mixed pointers and arrays, my progress through the book ground to a halt and I was spending so much time debugging my programs that I decided to ragequit for the time being. After a little research, it seems C is really meant as a low-level language, and was even being compared to assembly. Considering that, I'm impressed I got as far as I did given my virtually nonexistent knowledge on concepts such as memory management.
It's clear that, at least for the time being, I want to be learning a high-level language, preferably dynamically typed and object oriented, something I can focus more on creating practical programs and less on debugging. Python seems like the perfect candidate, so I'll be focusing on that for the next month.
Here’s the overall run comments I promised.
---
Although bossfights are the most entertaining part of the run, optimizing them would actually save the least amount of time compared to other mistakes in my test run. This is due to the fact that bossfights generally last between 15 seconds to a minute and most of my fights are reasonably optimized from about Demitel or Jahmir onwards.
So what would save the most time? It all boils down to holy bottle management. When I run out of holy bottles and am forced to run from random encounters, huge swaths of time are lost. Similarly, when I walk through parts of a dungeon and end up with a surplus of holy bottles, I could save a ton of time by running said sections. Aside from that, I made a couple errors navigating menus before I understood their frame rules properly (eg/ scrolling basically always takes 4 frames) and I never checked if clipping corners saved time, so that could add up significantly over time.
Funds are a complete non-issue. The only time you even need to make a detour for money is getting the elixir at the harbor northeast of the Mystical Forest, which sells for 10,000 g. After that point very early on in the game, bosses just drown you in money and the only things you generally need to buy are holy bottles, flare bottles and new weapons.
In fact, in a broader sense, item management is generally a non-issue. For instance, I beat the game without ever changing Cless’ armor. Giving him a perfect set of defensive gear would only halve the damage he takes in the endgame and Cless only takes damage when it involves cancelling an attack recovery, which is reasonably rare. Even orange gels are barely used and the only accessories that matter are the jet boots, sorceror’s ring (for puzzles), battle rune and mystical rune. As for weapons, most of what you need can be bought in stores or are acquired as part of the plot (eg/ eternal sword). There’s not a single case where getting a ?weapon drop from a random encounter proves worthwhile. What else? The only spell worth buying is thunder blade for the aerial battle. None of the techs get mastered so combos like mecha blade are pointless. The only character that ever gets used is Cless, with the exception of Chester early on and Arche in the aerial fight, against Dhaos and 2SR. There’s never a case where I have to grind for levels because boss fights are so fast that the time saved from learning a technique early would probably be lost on the first random encounter.
It should come as no surprise that most of the treasures I acquired proved to be unnecessary. Charm bottles, iron boots, items with high re-sale value (eg/ talisman)
The combat in the game isn’t as easily manipulated as I’d have liked. Most of the manipulation goes towards having bosses not block your attacks, and making the bosses as idle as possible since their attacks generally kill Cless instantly. That leaves little room to manipulate crits and dizzies. Moreover, from brute force testing, the RNG just doesn’t advance quickly enough for it to be possible to string 10+ crits in a row. I’m also disappointed by how overpowered fury slash and magma rift are, as I feel it kind of ruined the entertainment value of the last third of the game. Even tiger blade was abused in the midgame and there’s a sense that the fights are too repetitive. I was especially sad that I would never get to use soul wave or teleport because fury slash was just so dominant in terms of damage output.
A lot of my intuitions about tech speed were wrong. For instance, tiger teeth is significantly slower than I expected while lightning bolt is significantly faster. For a real run it’s very important to test the damage per second of each tech.
There’s a lot of subtleties to the combat system I learned as I went along. For instance, a tech can be used a few frames before a regular slash on any attack recovery. Another trick is repeatedly slashing when Cless’ position largely overlaps an enemy’s, which commonly happens when spamming a tech in a corner trap. What will happen is that Cless slashes and hits the enemy, but gets stunned and recovers more quickly than the normal recovery on a slash attack. Another little trick is that using a flare bottle can alter the enemy’s pattern slightly. I haven’t bothered to look into that more closely but I took advantage of it in a few places.
---
All things considered, I don’t think I’ll make a submittable TAS for this game. The main reason is that 90% of the run consists of optimizing running and holy bottle management, which is tedious and not something viewers would really notice and appreciate unless they actively looked for it. Skipping text is also pretty tiring, even with a simple turbo script that mashes A and B on alternate frames. The other main consideration is that the testrun is actually very optimized, especially towards the end, but not quite good enough for submission mainly because of my earlygame sloppiness, holy bottle mismanagement and predictable planning mistakes (mainly acquiring a lot of unnecessary treasure). A fully optimized run would look almost exactly the same, so it just doesn’t seem worth it. TASing the game has also made me painfully aware of the game’s combat limitations, particularly being limited to 2 short-range techs and long-range techs. You can’t change techs mid-battle, the targeting system is awful, spells interrupt the flow of the game completely, random encounters are far too frequent, and so on. Not that I regret making the run; some of the fights are just awesome. But honestly I don’t want to touch this game for the rest of my life.
Someone actually recommended that I look at the PSX re-make of the game, as well as Tales of Destiny. I haven’t even played either game, but from what I’ve seen PSX ToP has revamped graphics and, more importantly, fixes a lot of the combat issues I discussed above.
P.S. Just for the hell of it, should I submit the testrun for vault publication? Also, if anyone wants to make a youtube encode of the run, by all means go for it.
Test run complete!
Link to videoLink to videoTestrun smv
----------------------------------------------------------------------
3.26- The Future of Alvanista and the Grand World Tour
----------------------------------------------------------------------
It might be faster to re-stock on holy bottles in Freezekill (as opposed to Alvanista), but I’m too lazy to test something so insignificant right now.
----------------------------------------
3.27- Freeze! Fenrir's Ice Cavern
----------------------------------------
Wolfang fight: frame 702996
I wasn’t sure what skills to use so I ran a quick test.
Magma rift: 82 frames, ~3500-3600 damage
Fury slash: 98 frames, ~4000-4300 damage
Tiger teeth: 108-112 frames, ~1900 damage
Magma rift is ~19.5% faster than fury slash but fury slash does ~14-19% more damage. I’d say magma rift is superior since it is divided into two hits and the first hit impacts relatively quickly. However, fury slash has more range and is more likely to swipe multiple enemies.
I’m really surprised about tiger teeth since I always had the impression that the attack was lightning fast and easily quicker than magma rift. While tiger teeth definitely impacts early, I suppose the recovery time is quite large.
As for the fight itself- Normally this boss is an absolute nightmare, but I manage to manipulate him into doing nothing the entire fight and make him look like a complete joke.
Well, once again I run out of holy bottles. But this time I found something very interesting by accident. When I use the sorcerer’s ring on torches, the reRNG increments by 1. In fact, if you increase the counter past 255, it rolls to 0 and you can’t face any random encounters until you leave the room! Amusing, but the applications are very limited since the fastest you can fire the ring is once every 4 frames.
--------------------------------------
3.28- Burn! Odin's Fire Tower
--------------------------------------
Random battle (Kary x2 & Jinn x2): frame 717939
I have to get into a random battle that includes a Kary in order to get a salamander ring drop. The ring gets transformed into a nymph ring using a rune bottle, and this allows me to navigate the lava areas of the dungeon without getting hurt.
After that I walk through the dungeon since I am tired of running out of holy bottles.
Forced battles (Kary x2, Druid & Djinn): frames 724763, 726553, 728204, 729885
My main strategy here is to land as many multi-hitting fury hits as possible, which is a challenge to set up.
Flambelk fight: frame 731828
As it turns out the best strategy is to… cheese fury blade. Not terribly exciting, but at least now Cless knows soul wave, my favorite tech in the game.
Well it seems walking the entire dungeon was foolish because I only used 4 holy bottles. Clearly I should’ve walked Fenrir Cave and ran Odin’s Tower.
--------------------------------------
3.29- Spirit of Creation, Origin
--------------------------------------
Figuring out this bossfight was a little trickier than Wolfang and Flambelk because neither fury slash nor magma rift even hit Origin. I ran some more damage tests since I wasn’t sure which tech would prove the optimal choice. Note that DPS = damage per second and that a flare bottle was used for each tech.
Tiger teeth: 109 frames, 2316 damage - 1275 DPS
Lightning bolt: 213 frames, 4229 damage - 1191 DPS
Gale shield: 356 frames, 8368 damage - 1410 DPS
I’m surprised lightning bolt comes out that fast, even though it has the lowest DPS. Clearly tiger teeth + gale shield is the way to go.
Origin fight: frame 745744
Another one bites the… gale shield. But seriously, the rewards after this fight are insane: You get the eternal sword, the soul stealer bow, the origin summon, and the teleport, soul strike and soul forge techs. To top it all off, Cless finally learns focus more after hitting lv 50.
--------------------------------------
3.30- Before Taking on Dhaos
--------------------------------------
Can anyone say cutscenes?
-------------------------------------
3.31- The Final Confrontation
-------------------------------------
Dark Eye fight: frame 784383
Dark eye is immune to the first hit of magma rift, as well as soul wave, so I only had to compare fury slash and teleport.
Fury slash: 98 frames, 3825 damage - 2342 DPS
Teleport: 205 frames, 4578 damage - 1340 DPS
Well that confirms it: fury slash is overpowered.
So I found an excellent strategy for consistently killing off my characters. Re-position everyone in front of Cless at the beginning of the fight, and have Cless use fury slashes from the back row until he triggers a boss attack that kills everyone else. Sometimes I have to use magma rift because fury slash will be blocked no matter how many frames I wait for the RNG to advance.
Hanzo fights: frames 789657, 791860, 793904, 796056
I timed Focus More since these fights are so short I was questioning whether it was even worth using compared to a flare bottle. It comes out to 127 frames, which is about the length of an extra fury slash/magma rift, so it’s probably worthwhile. Sometimes I have to use the flare bottle anyway to alter the RNG.
Seal Eye fight: frame 804926
The polearm and eternal sword do about the same damage to him but the poleax has a much higher crit rate, hence the weapon switch.
Hardest fight of the run so far. Hitting both Seal Eyes with Fury Hit simultaneously is a real pain since they tend to block so often, and most of the time the Seal Eye closest to Cless will kill him instantly after attacking. That’s why almost every attack resulted in a dizzy; it was necessary.
I finished the dungeon with 4 holy bottles remaining, so some running could’ve been done.
Dhaos 3 fight: frame 822806
Time for some more damage testing. Note that Focus More was used before each tech.
Fury slash: 6700-6903
Magma rift: 4158
Soul wave: 0
Gale shield: ~450 / hit
Teleport: ~5000-6000
Soul edge: ~4800
I hate to say it, but fury slash always comes out on top.
Dhaos starts the fight in a state where he sits still until you attack him but will always block your first attack. Thus, Cless’ opening jump attack only serves to advance him. After that the fight is disappointingly easy.
Neo-Dhaos fight: frame 826281
As for Neo-Dhaos, he can optimally be beaten using 4 Chrono Glasses, but I only had 3: I missed one in the cave dungeon very early in the game and used another in the 2SR fight. Nonetheless, I still managed to defeat him in 7 gale shields without Cless facing a single attack!
And that concludes the test run. I’ll add some overall run comments shortly when I feel up to it. For now I’ll say I’m amazed that I achieved a time of 3 hours and 50 minutes, never in a million years would I have believed the game could be beaten this fast. Nor did I expect Cless to surpass level 50 on forced battles alone.