Username:
B
I
U
S
"
url
img
#
code
sup
sub
font
size
color
smiley
embarassed
thumbsup
happy
Huh?
Angry
Roll Eyes
Undecided
Lips Sealed
Kiss
Cry
Grin
Wink
Tongue
Shocked
Cheesy
Smiley
Sad
page  <- 12345678910111213141516171819 -> <- 1 .. 14 .. 19 ->
--
--
List results:
Search options:
Use \ before commas in usernames
Yo guys, I have now uploaded a video of me grinding from Level 13 to Level 17!  Not only that, but I've even compiled another quick video, showing the first part of what a speed run of the remake of this game would look like, for your viewing pleasure!




So what do you think?
FQ, I've been reading this thread for quite some time. Have you ever read the rules of speedrunning on this site? You can't do it on an emulator.
So?  That's only if I want to upload it onto SpeedDemosArchive.  I don't personally care whether it ends up on here or not, I simply wish to host these videos both for the sake of entertainment and to give the speed runners here a helping hand.
Yes, a worthless avatar riding my posts.
Quote from Darkwing Duck:
Quote from Fionordeguester:
Actually, that's untrue.  The HP/MP growth type is worse than both STR/AGI and STR/HP.  I'm not sure how you came to that conclusion, but I'm guessing you ran through a couple of quick battle simulations only a few times?  Because the amount of times that the hero wins isn't always consistent, the best thing to do is to run something like 10 battles at once, several times. 

You're right, I meant Str/HP is better than Str/Agi for Dragonlord.


Better strictly versus the Dragonlord because he has better odds in that fight alone, but I think Str/Agi will save more time throughout the game.

Pre level-4 is rough because the high agility hasn't kicked in yet so fights are a bit harder to run away from. +1 HP to start with doesn't help.
You can fight Wyverns from 12-13 easily.
You can fight Green Dragons at level 18.
...I think there was something else that you get access to one level sooner as well, I need to check again.

I've been reading up on my DVD recorder's operation, however my small TV is really old and beaten down so it won't work for recording this run. When I get my other, bigger TV I think I'll be ready to attempt recording again. Just gotta relearn how to navigate the Mountain Cave in the dark, I plan to use RADIANT in Garin's Tomb and the Erdrick's Sword raid, and the Torch from the start of the game in the final encounter in Charlock. Bumping into walls will cost more time than opening the menu and turning the lights on, I imagine.
You have to admit it would be pretty beast to be able to run the thing in the dark though:  Especially at the end of such a long speedrun.
But you shouldn't need to get to Level 19 at all to defeat the DragonLord.  I remember doing some time calculations and coming to the conclusion that beating the DragonLord after a failed attempt at Level 18 is faster than beating him in one attempt at Level 19, and it IS possible with smart strategy and a little bit of luck.

Also, I've never had problems fighting Green Dragons at Level 17.
Hello everyone!  I have been working on a segmented run of Dragon Warrior for quite some time, but until today I didn't know whether I'd have the fortitude to finish.  However, all the most difficult, long-odds segments are finished, so I feel safe talking about my endeavour.  Seven out of ten (or eleven) segments are done:


I choose long-term Str/HP growth for easier going in the first and last segments.  You would think that long-term Str/Agi would give an advantage in grinding by moving enemies to the safe zone earlier.  However, by sheer coincidence, the equipment order and timing I use means this never happens in a useful way.  Levels two and three are reached in Zone 1 (slimes, red slimes, and drakees), in the hills southwest of Tantegel -- this is, I think, the fastest I've ever reached level two.  Level four is reached via ghosts in Zone 2, northwest of Tantegel.  Finally, level five requires lots of spell-casting against the critters in the Hills of Hubris.  I'm not using the gold glitch (it would be five to ten minutes shorter than my route), and I need the extra gold and durability from the extra level.


To grind reasonably quickly at this stage, I need a better weapon, and for that I need money.  So it's off to the Mountain Cave for some luck manipulation.  I use torches because I think it's slightly faster than bump-walking, since you lose a fraction of a second not only with each bump, but with the slight delay before movement from a stop.  This was actually my second Death Necklace -- the first try was sloppy and unlucky, so I elected to try again.  I kill one Drakeema in a lucky fight (downing it in two shots) because they offer unusually good XP for their HP.  This fight saves about thirty seconds, given the next segment's rate of progress.  Two trips to the chest are needed to gather enough gold.


I dash to Rimuldar to buy a shiny new Broad Sword.  I then grind around Kol to level seven, remembering to pick up the Fairy Flute in the meantime.  The Rimuldar monsters then bring me to level eight.  Someone using the gold glitch would gain these levels much more quickly, but then they'd also have to sit through twenty minutes of chest looting.  Anyhow, level eight is all I need, while the speed gains of the gold glitch route would have overtaken this route around level nine.  In the kept segment I had extraordinary luck towards the end, sleeplocking five enemies in a row.  I finish by buying six keys and using Wings to return to the king, rather than death-warping, as gold will continue to be tight all the way into segment six.


I need a Large Shield for segment six, so it's off to Garinham.  Since I don't want to visit this town more than once, I then head into the Grave, which turned out to be relatively easy even at such a low level.  I need the Cursed Belt and 240 gold to fill up on Herbs and Fairy Waters, and I reach my target on the very last step thanks to a fortuitous excellent hit.


Axe Knight at level eight.  This fight is ridiculous, give it a look.


Level eight, Erdrick's Armor, a Large Shield and the Dragon's Scale are exactly enough to push Wolflords into the safe zone, so I use them to grind to level ten.  The huge agility boost at this level pushes the next segment into the realm of the plausible.  I also take the opportunity to pick up the Stones of Sunlight and Erdrick's Token.


The various Axe Knights of Charlock Castle sure did exact revenge for their fallen comrade.  Anyhow, on this hundred-and-fifth attempt J manages to retrieve Erdrick's Sword, having first finished the remaining quests.  Elapsed time so far is 1:50:11.

That's where I am now.  Next up are two-and-a-half hours of level grinding, which will be time-consuming but not too difficult otherwise.  Advice and opinions would be welcome for this portion, and if anyone has a plausible idea for beating the Dragonlord at level seventeen, I'd be happy to hear.  I'm currently weighing two different plans:

Segment Eight: Grind in Zone 9 (Wyverns and such) until level thirteen, resting in Zone 8 (Wolflords etc.).  Finish by picking the Silver Shield, as I'll have enough gold shortly thereafter.
Segment Nine: Grind in Zone 10 (Rogue Scorpions, Knights, etc.) until level eighteen, resting in 9.  Oddly enough, during testing I found that Zone 13 (inside Hauksness) was slightly slower on average, even at level seventeen.  This was using suboptimal strats against the Starwyverns, though -- I was using Stopspell at the start of the fight, rather than waiting to see how much damage my first hit did.
Segment Ten: Kill the Dragonlord.

Alternatively:

Segment Eight: same as before.
Segment Nine: grind to level sixteen in Zone 10, resting in 9.
Segment Ten: grind to level eighteen inside Hauksness.  While slower on average, it would have a higher variance, so after optimization (hope for lots of wizards and werewolves) it could end up faster.  The extra segment would cost about a minute, though.
Segment Eleven: Kill the Dragonlord.

Please give a look, and let me know what you think.  Also, thanks go to the various contributors to this thread (Ryan8bit, ShinerCCC, and Darkwing Duck especially), whose discussions have helped me in planning.  And enjoy SDGQ, I sure am!
Heh, this is the thread that made me join the board, so it's nice to see it still continueing, grindfest or not...
Highly Evolved
Why aren't you using the gold glitch?
Edit history:
Lhexa: 2013-07-26 04:53:07 pm
Glad to hear it Melodia!  And to Darkwing Duck:

Using Fionordequester's video for reference, s/he reaches level 8 at around 1:23, whereas I reach it at about 1:02.  Even subtracting six minutes for a Death Necklace worth of chest looting, you would need fifteen minutes of optimization over his video to match my time.  Some of that would come easily (using Ghosts instead of Drakees for level four, not taking the long way to the Mountain Cave, plus more), but the rest would have to come from luck, and I don't see that happening.  I found a standard deviation of only about fifty seconds (on a mean of ~20:30) on my first segment, and a standard deviation of only 1:52 (on a mean of 39:00) on the third, so fifteen minutes aren't easy to come by.  Of course, for a single-segment game the gold glitch is the clear winner -- no Death Necklace most likely, and you'd be going up to level thirteen or so before getting Erdrick's Armor.

For what it's worth, I'm tracking the segment statistics because I want an objective criterion for when a segment is good enough, not just my own opinion.  My own perfectionism would probably drive me to distraction, then.  For a high-mortality segment, I set my goal at 1.5 deviations below the mean, and for a low-mortality segment it's 2 deviations below the mean.  For the three long-odds segments I took whatever I could get, of course.
Highly Evolved
I'll look at this a bit more when I get back home from SGDQ.  I have some experience with segmented runs, so maybe I can see some things there.
Thanks!  I saw your E.V.O. run, by the way, and it was hilarious.
Highly Evolved
Thanks!
Quote from ShinerCCC:
For those who didn't know, Ryan8bit finally released his simulator in webpage form:
http://www.woodus.com/den/games/web/dwsim/battle.htm


I think I saw this before... and I did my own calculations and thought a Level 18 win was unreasonable... But Str/HP growth, Level 18 with best equipment wins about 1/3 of the time according to this simulator... Now I really want to see a win at Level 18...
I have to say, that must be one of the most ridiculous Axe Knight fights I've ever seen.
Yeah, "ridiculous" is a great way to put it.  In emulator testing I managed to beat the Axe Knight a dozen or so times, but that was always with a succession of high damage rolls.  Then, in the actual run, this happens...

Level 18 is the current goal, and it isn't that hard... whenever the Dragonlord attacks twice in a row after you heal, a Str/HP character will get in two attacks before healing again.  If this happens four or so times, you win.  For a Str/Agi character, you also have to have low damage rolls on the boss' attacks, making success much less likely.  The real prize of a Dragon Warrior speedrun would be a level 17 win, since that's when you get HEALMORE.  But then you only do half as much damage, so a conventional strategy is plain impossible, even if you get double strikes every time, with maximum damage.  Meanwhile, a SLEEP-based strategy has odds on the order of 10^(-5).  I'll try to estimate the odds more precisely later on.
Edit history:
Lhexa: 2013-08-03 11:32:02 pm
Progress!


This segment is for grinding three more levels and obtaining the best shield.  It turned out that Zone 9 (Wyverns etc.) is the best zone all the way up to level thirteen.  I need thirteen Goldman kills to raise enough money for the Silver Shield, but sometimes twelve will do, so I ended up putting off the thirteenth kill until I was sure I needed it.  I was amused by how close the margin for affording the shield ended up being.  The segment ends with a quite lucky run to Cantlin.

After testing, I've found that Zone 10 (Wyverns, Rogue Scorpions, Wraith Knights, Knights, Demon Knights) is best up to level 17, at which point Zone 12, of all things, becomes the best one.  That's the Charlock Castle zone, which has Knights, Magiwyverns, Demon Knights, Werewolves, and Starwyverns.  There's a small patch of hills south by southeast of Hauksness within that zone.  Zone 13 (inside Hauksness), despite its tasy Wizards and Green Dragons, ends up being slightly slower, with Zone 10 a close third.  Zone 12 has the added advantage of not requiring an MP refill, so I can save a minute and a half by having one fewer segment.

However, this means that the next segment will be a mind-numbing two hours long.  It'll take a while to finish.
Yes, a worthless avatar riding my posts.
Look on the bright side, my run is one mind-numbing 6 hour segment.
Good luck!  I don't have the stamina (or the equipment) to try to finish a single-segment run.

Happily, it's now clear that my run will come in under five hours, which was my goal starting out.
Edit history:
bj_waters: 2013-08-23 11:26:42 am
Quote from Lhexa:
Happily, it's now clear that my run will come in under five hours, which was my goal starting out.


Wow, under five hours!  That's amazing!  Can't wait to watch the whole thing when you finish.

To be honest, I gave half a serious thought about running this about a few months ago, before realizing just how tedious of a grind it would be.  Still, I thought that if someone else was running it, I could support them, so I searched through these forums, found this thread, and read through all of it, only to realize that everyone else's route was going to be far more ambitious than anything I had come up with (I don't know if I can be a speedrunner embarassed).  However, after reading through everything, I did have two thoughts.

1) I don't know if everyone else has figured this out, but I don't remember it being mentioned in this thread.  I noticed, while playing the game, that you can kind of "autofight" by simply holding down the A button.  You can even start holding it as the random encounter swirl starts.  That way, if you run into an enemy that you just want to bash it out with, you don't have to rapidly mash A to get through it.  I figured it was worth bringing up, if only to prevent you guys from mashing your thumbs to death for 5-6 hours.  It also works in shops and long text boxes (such as when you first start the game and the King gives you the infodump).

2) Someone earlier in the thread lamented that there wasn't any kind of sequence break.  I haven't come up with anything, but I put some thought into that, trying to come up with ideas of how one could possibly "glitch" this game, and the only thing I could think of was to somehow underflow the player's stats.  Sort of like in Pokemon Red/Blue or Final Fantasy 4, we could have some amazing results if we come simply reduce a stat to a negative number, getting a ridiculously high number instead.  Unfortunately, as far as I know, there is nothing that reduces the player's stats in the game at all.  Not even the Cursed Belt or Necklace (you'd think they'd do something, but no, they only prevent you from selling them or entering the castle).  I don't even know if the game's programming could handle underflow or how it would work.  Still, I thought if I brought up the idea, someone else might be inspired to find something useful.

In any case, good luck (literally!) to all the runners for this game, and don't let it drive you too crazy. Grin
Thanks!  Yeah, I know about holding down the "A" button, and it works in other places, like long dialogues or buying herbs.  It's good to have on the record, though, for any future runners who want to read this thread.  Another cute trick is that you can press "B" to say "No" when a yes/no prompt appears.

There's some slight memory corruption that can arise when you use the infinite gold glitch, so that you can pull strange things from the Stones of Sunlight chest.  I don't know of anything useful that can be done with it, but then, the gold glitch itself was not discovered until four or five years ago.  This game might have a few secrets to be discovered.

My progress: I've done enough complete runs to establish that my ninth segment's goal is sub-2:04:00.  My best so far is a 2:06:24.  This segment consists of nothing but levelling to eighteen.  Meeting my goal would put me on pace for a total time around 4:54.
Here's a cute little trick I found while working on the long grinding segment of my speedrun.

I mostly grind by walking back and forth in the Zone 10 desert south of Hauksness. Every now and then a long interval occurs between battles, lasting into or beyond the sixth bar of the overworld theme. Eventually I noticed that, every single time this happens, I get the same monster, in this case a Demon Knight. Wondering whether this would happen elsewhere, I tried out grinding in other desert zones, and indeed found that I got consistent monsters after such long intervals. I started to chart out which monsters occur where, but it soon became clear that I always got the fifth monster from each zone, using the order found in Ryan8bit's FAQ.

Well, as it happens, the fifth monster in the Zone 11 (hills far south of Hauksness) is a Metal Slime. So, when I realize I've made it to the sixth bar and have a Demon Knight coming, I could easily scoot over to the next zone and fight a Metal Slime instead. It also happens that these encounters after a long wait occur at definite points after a reset (for a given save file), and play out identically every time. So, some of these Metal Slime fights will be guaranteed wins, allowing some modest Metal Slime manipulation in my speedrun.

One downside of this trick is that I have to map out all of the candidate Demon Knight and Metal Slime fights, about thirty of them in a two hour segment. It's good to know both, since the damage numbers of the Demon Knight fights can identify exactly where on the list I am. The other downside is that I only seem to have a one in two or one in three chance to actually hit one of these fights, presumably because the RNG is called two or three times every step, and I have to be on the right subsequence of numbers. (I checked out the code in Ryan8bit's FAQ and only found one RNG call per step, but more could have been omitted.) Thankfully, two of the good Metal Slime victories are right at the start of the segment, so I can keep resetting until I hit them.

If there's a short repeating pattern in the Metal Slime fights, the trick could be useful in a single-segment run, too. The tricky part would be figuring out where in the sequence you are. Oddly, there only seem to be four distinct Metal Slime fights: an excellent hit on the first turn, an excellent hit after a preemptive strike, a kill in four hits, and a run after seven turns, only getting in one hit in the third turn. This last fight is as common as the other three combined, though, and I haven't found a pattern in their occurrence yet.

This trick is all but useless for most circumstances, but it should save me four or five minutes when I'm done.
I was able to record a segment which included nine metal slime kills, but bad luck elsewhere meant I only got a 2:00:19.  I'm going to try to shave one or two more minutes off.

The phenomenon is even more regular than I could have hoped for.  The times when the long-interval battles can occur take place roughly four minutes apart, with variation resulting from how long it takes me to hit buttons.  So, even when I miss some, I can tell when I should be expecting the next, and what entry in the list it is.  Moreover, the Metal Slime battles do have a pattern of period eight:

1a, 1b, 7, 7, 7, 4, 7, 4

where the battles go as follows, if you just hold "A":

1a: excellent hit on first attack
1b: preemptive HURT by slime, followed by excellent hit
4: kill in four rounds, no misses
7: slime runs after seven rounds, after taking only one point of damage on the third round

This is very strange, but I can't complain.  It should even be usable in a single-segment run, since half of them are victories (much better odds than usual), and if you get one of the first two you can single out where in the list you are.
^That^ is some crazy stuff. I would think that being able to manipulate monster encounters would be useful throughout the run (not that I think you should start from scratch, Lhexa; you're already this close, so just finish it!).  It makes me wonder, now that you're seeing this consistent pattern, what else can we pick out from ROM addresses and such?  You know, like how they figured out item drops in Legend of Zelda.  I would think with patterns that regular, someone could reverse engineer some things about this game.

Keep up the great work, Lhexa!  We are (or at least I am) rooting for you.
Wow, this thread is amazing. I'm just about to finish my first every playthrough of Dragon Warrior and I'm excited to learn more about the game.

BJ, your suggestion to track some memory addresses is a good one and I'd be willing to poke around a bit. I haven't investigated the ROM/RAM yet, but my guess would be that the regularly repeating, "long interval" has to be linked to a specific counter. Whether it's step-based or time-based is unclear since the game has a couple timers for the menu popping up and spells wearing off and so on. Lhexa, do the same number of battles take place between long intervals or is it only the first battle of each cycle that is predictable?

Beyond manipulating when metal slime's appear, the fact that Lhexa has found a repeating pattern within the combat is a good sign. The fact that it repeats in sets of 8 is a really good sign. That means the logic of each metal slime fight is tied to a byte-sized counter that should be easy to track. I'll see if I can find anything.

As it stands, it's pretty awesome that you can not only manipulate a metal slime fight but can run on the second turn if you don't get a hit. I wonder if all the fights could be routed for a single segment attempt? Feasel's been doing something similar with a cold booted Final Fantasy 1. Shiner have you noticed any patterns in your long plays?