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
1 page
--
--
List results:
Search options:
Use \ before commas in usernames
Edit history:
ktwo: 2013-06-16 01:35:13 pm
ktwo: 2013-06-16 12:39:05 pm
I won't say that I'm proud of creating a thread for this game, but I feel it has to be done at some point, so might just as well be over with it... I'm not too far off on getting a recorded run, so I wanted to share some information in the unlikely event that someone else had spent enough time with the game to give input. Or maybe sparking the interest of someone wanting to give it a try (as awful as the game is, it's really easy to get into).

Anyways, I thought this would be a suiting project to prove that I can handle the full 60 Hz unleashed from ntsc, after previously having been stuck with only pal... More seriously, I wanted to try out my new recording setup more extensively on something easy and not risk to achieve something that was not recorded properly. I remember hotl from emulator and that it was basically only to remember the route and then run through the whole game while avoiding all enemies in 5ish minutes. Unfortunately, I had forgotten about how unsuited the nes's d-pad is for diagonal presses, as a speedrun is basically 5 minutes of pressing diagonal up (to run). This actually makes it a whole lot more challenging (on top of being a very luck-dependent game).

There is a TAS of about 0:05:17 (SDA-timing). The tas is no-death, but I'm planning to abuse deaths in my run. Having no deaths basically just means restoring health as necessary near the end of the game (and hoping that the damage inflicted by enemies is minimal). Abusing deaths requires at least a tiny bit of planning.

Current route (I use the same route as the tas, so below are just how I deal with various critical points in the game):
- Move Goldmoon to the backrow asap.
- Run through everything until the first dragon in the second section of the game. As far as I can see, the only reasonable way to pass dragons without fighting them (not an option as that would take forever) is to use "protection against evil" (PAE). PAE can only be used by Goldmoon and she has to be in the front row to cast magic. The idea is therefore to kill off one guy just before the dragon. Because of how much luck there is in the game, this is not evident, but works most of the time.
- Use PAE again in the section with the flames (still in section 2). By running past the first flame before casting PAE, it's possible to pass the dragon in the next room.
- Use PAE twice more against dragons in section three.
- Possibly use the menu trick to pass an enemy or two (see below). Each time costs ~0.5-1 second, but sometimes there is no other option if a lot of damage has been damage inflicted. If everything else goes well in a run, this could maybe be accepted once.
- Along the way, kill off the three guys in front of Goldmoon to make sure she is in front when facing the dragon. This is a major problem. Goldmoon usually dies in an instant from the three trolls before the dragon. It requires a lot of luck to have one of the guys survive long enough to just pass the last troll and then die off to leave Goldmoon in front (and not die together with the last guy, which is also very common). If Goldmoon is not in front, an additional menu change is needed, which costs ~3 seconds.

Useful tricks/features:
- The character gets a small boost when quitting the menu. This means you can pass through some enemies without stopping if you are close enough and then open and close the menu. Overall it costs time though (0.5-1 second, depending on how annoying enemy it is), so it's only useful when energy needs to be conserved or when you anyways have business in the menu.
- Protection against evil (PAE) protects against damage (but not entirely) and makes it possible to pass through dragons (and other enemies are also easier to pass through). The only spell I use in my route.
- In some of the previous route iterations, I used healing spells.
- There are several spells that do more or less the same thing (web, sleep etc). You shoot out a fireball and when an enemy is touched by it, it freezes in place. Some of them might be a little bit faster to cast than PAE. Unfortunately, they have a very high likelyhood of passing above dragons. There is a small possibility for the fireball to go low, which then hits dragons, but I haven't figured out a way to manipulate that.

My best time so far is 0:05:23. I'm still very new with this game, so I wasn't fully prepared for how to act on the last troll. I should have turned around and let it hit me once more. That would have killed off the guy in front and avoided having to switch Goldmoon to the front in the dragon room. 2-3 seconds lost, depending on how fast the troll would have hit again. There was also a small hesitation on one door (0.5 s), and one door I overshot a bit (0.5). Technically, I'm probably losing a little bit of time on every single door transition and by getting stuck on enemies, but I think this is about as good as it will get. Maybe a second or two can be shaved with luck, but I'm just speculating. Right now, I'd like to get a run without having to switch to Goldmoon in the dragon's room. I think a 5:20 would look pretty clean.
Thread title:  
Might be magic...
I finished the PC version of this game (which was pretty awful compared to the NES version) a very long time ago. Glad you're taking it up Smiley
It sounds incredible that the game I'm playing should be an enhanced version of something else, but I believe you if you say it...

My best time is now (a very high) 5:19. I had some outstanding luck (I know it's a strong statement after still being fairly new to this game, but things really fell into place in that attempt). There were also a few not so good looking mistakes, but it would have been a 5:19 even without them. It's certainly possible to still cut a second or two off, but I'm not sure I'm up for the task of going for that kind of optimization on a game like this. I'll think about it for a few days, but I think this is as good as it gets from me. I have bigger fishes to catch. :-)
Time to revive this thread. A new TAS has been released for this game and it includes a few time savers that are easy to implement in a real-time speedrun:
- Walk left in the second room of the game. This is around 3 seconds faster than going right (which every TAS/speedrun until now has done).
- If Goldmoon is the first character, she can throw magic by attacking without going into the menu each time. This changes things up towards the end of floor 2 and all of floor 3.

I've come up with three basic route choices for a real-time speedrun:

a. Follow the TAS route with keeping Goldmoon as the first character throughout. Enter the menu slightly before the fires near the end of floor 2 and then activate "Protection against evil" and "Hold person" (and most likely a health refill). This is possible, but the odds seem very low to survive long enough. The best way to avoid health losses is by attacking enemies and then try to run through them. Unfortunately, this is in no way a fail-safe method, so the enemy is just getting chipped away as you go along no matter what. Even a run that's save-stated in each room isn't trivial to complete. A clean and successful attempt like this could get very close to the 5 min mark (possibly below). I've tried attempts with this route, but haven't really gotten anything that was close to survive deep enough into floor 2.

b. Start out by a character swap by putting Goldmoon as the second character. This additional menu trip costs ~3 secs. This makes it quite a bit more manageable than 'a' and also possible to play slightly more aggressively, opening up the possibillity to gain back a little of the time lost from the extra menu. It's still a bit tricky to get into floor 3 though.

c. Similar to 'b', but instead swapping Goldmoon to position 5. The idea is to avoid "Protection against evil" (1.5-2s saved) and just run through the fires in floor 2, hoping that the first four characters die off and leave Goldmoon in the first position on the other side of the fires. Killing off the characters loses time and it's going to be slower running through the fires since you'll get hit quite a lot. So overall, this is going to be close time-wise to 'b'. It's much easier to get runs going though.

For now, I'm sticking with 'c' and earlier today got my first completed attempt after doing attempts again (https://vimeo.com/403814132). It was a pb of (a very high) 5:04. There were no particular mistakes, but the menuing was bad, I mistimed a few of the doors and the RNG was average (got stuck on a few more enemies than I had hoped for and I would also like to avoid Raistlin ever being the first character). Hopefully, I can bring the time down some more, but I don't see sub-5 as possible unless attempting 'a'. While 'c' seems to be by far the easiest of the three, I'm probably going to try some 'b' attempts as well. It's not evident to calculate which of 'b' and 'c' is the fastest, so it would be useful to get some attempts of both going into floor 3 and then compare the times.
I improved my time today by 0.15s. I had bad RNG in the last room, which cost around a second. Hopefully, I can get another attempt on pace within not too long and close it out without any major time losses. It would be nice to shave off that additional second. My best pace has been up to two seconds faster, so a 5:02 is possible if absolutely everything goes my way.

I've done some more testing and it seems like 'c' is a little over a second faster than 'b'. 'c' is also much easier to get runs going with. 'a' is still the fastest and the only one that would be possible to achieve a sub-5 min run with. The RNG requirements are too demanding though, so I stick with 'c'.
Thanks for the new efforts! I'd been lacking in ideas on how to reduce my times on this except I did adjust the route as you described above already - going Left at 2nd screen instead of right. Does save a few seconds.

Would not have thought of hold person with Goldmoon on my own. Planning to get some practice and runs in trying that!

Just wanted to say "Thank you" mostly again. Dunno if I can figure out improvements or get god-RNG to get 5 or sub 5, but will join you on those attempts Smiley
Nice to hear you'll try to push the time lower! Good luck with the attempts and let us know if you find any improvements.

As for me, not long after my previous posts, I got a 5:03. Since then I've been on a break, waiting for an improved TAS to be published. I know 'arc' is working on it. I'm thinking that with someone looking at the game with a magnifying glass, chances are something new will be found. So I might do more attempts, but for now I've put them on hold.

I actually thought the TAS would be ready back in April, so I didn't bother posting about the 5:03 or publishing it (it was streamed, but the vod has disappeared). But I've uploaded it now, in case you're interested, https://vimeo.com/417686667. The strats are very close to my previous pbs, but I found that it's slightly faster to take the last death in the last room of level 2 (the one with the pits) instead of in the room before it. Making it that far is RNG though, so it's of course only to adjust to whatever the game gives you.

My sum of best with the current route is 5:02, but I'm pretty sure that could go down to 5:01 with a few more completed level 3. I can definitely see a 5:02 run happening, but that would require pretty damn good RNG and, needless to say, no mistakes. Even though better execution through e.g. improved menuing and optimizing the door transitions could on average squeeze out a little more time, I have trouble seeing the current route go below 5. For that to happen, it's as far as I can see still required to attempt route 'a' in my post above.
Nicely done! Just watching there's about 2-2.5 secs of saves with perfect RNG, yeah.

Might save some time if you do have Goldmoon front position after fire room and pause/menu jump the dragon to select the Hold Person spell. Could save 1-3 secs there possibly.
Casting a spell costs ~40 frames, so that's the absolute maximum you could save from "menu jumping" through the dragon. In practice, you're probably still going to get stuck a bit on the dragon, so it's more like half a second faster than attacking it. But I like the idea. It's riskier than casting a spell, but if Goldmoon is in front of the party at that point, the time save is imo big enough that it's worth doing.
The new TAS was completed a few weeks ago, so I have started doing attempts again. I don't think there was anything not already known that was used in the new TAS.

I have still made one significant change in the route. I now postpone the character swap by two rooms. This gives me good (but not perfect) and predictable RNG against the first warrior and I can then menu clip through the dwarf (an enemy which otherwise has proven particularly annoying). There is virtually no difference in time by changing the place for the character swap. However, the first floor becomes much easier to complete now.

I lowered the time today to 5:02.8 (https://vimeo.com/426772682). I got the double character kill in the flame room (~0.4s saved over my previous pb), managed to jump through the dragon at the end of floor 2 with Goldmoon (~0.4-0.5s saved) and finally a floor 3 without any missed attacks (0.5s saved). Getting those three events in the same run was very fortunate. I did however lose 0.3-0.4s early in floor 2 from a bad enemy position, so the RNG wasn't entirely perfect. Still, very good RNG overall. Unfortunatly, I lost 0.3-0.4s to an execution mistake in floor 3. I got too aggressive and missed a door. Had it not been for that mistake, I would have been very happy with the run. I don't think it's reasonable to expect the three big time savers above in another run, so it's going to be a challenge to lower the time. But I'll continue the attempts. I'd really like to get a pb without big execution mistakes before I'm done.
Now I'm done. New pb is 5:02.2. It's the last attempt in today's stream. I got a really unusual runthrough of floor 2, which somehow ended up saving time. I know the situation was quite exceptional, but I have to look at it in more detail to understand where the time save came from. I'm glad to say that this run is clean without any execution mistakes (with the usual disclaimer that most things could of course be done a little faster).

I'll document the run in more detail and that will be it from my side in this game until something else is found.
Edit history:
ktwo: 2020-06-10 02:03:24 pm
Forget what I said above... I just found THE trick that will easily push the time below 5 minutes!

I mentioned there was something odd about my previous pb and that I had to look at it in more detail to understand. The thing that bothered me was that I had to play as Raistlin for a few seconds after the fire gauntlet. Raistlin's movement is slower than the others except when jumping. So to compensate for playing as Raistlin I did of course jump, with the result that I completely overshot one of the doors. Logically, that should have cost me quite a bit of time. However, it turns out that holding up while in the air counts for triggering a door entering! This has several important benefits and no disadvantages:
* Very easy to get pixel-perfect door transitions
* No lost time from moving the thumb from e.g. "up+right" to "down" on the d-pad (since this can now be done while in the air without any loss of momentum). Door transitions should therefore also be frame-perfect going forward.
* There should be almost no more lost attempts from being too aggressive and stopping before the character is in front of a door hitbox
* No need anymore to play conservatively in some rooms because the door hitbox can be shifted in either direction (to be considered random with the current knowledge)

While I'm sure this will cut seconds off the time, without any downsides I can see so far, it's worth noting that it doesn't quite work if you start jumping too close to the door. But that really shouldn't be a problem, so it's just something to be aware of.

Edit (2020-06-10): 4:57 in today's stream. Not a great run though. I should be able to lower it some more.
Edit history:
Vriaeliss: 2020-06-16 02:35:16 pm
Vriaeliss: 2020-06-16 11:49:12 am
Vriaeliss: 2020-06-11 09:00:23 pm
Amazing work!

Might be healthy enough myself to try this out soon. Thank you again for your ingenuity and efforts!

Edit 6-16: Finally got things up and running again! Keep losing Goldmoon in Fire room or to the dragon. Need to get to the swap spot before she gets too hurt I guess. Keep getting wrecked by 1st man enemy Sad
Edit again: got a 4:59 on the books Cheesy lost 3 secs took safety waterfall. Got trolled once on floor 3. Goldmoon was beat up on floor 1 too.
I have to start by correcting something about the door transition trick. It's more powerful than I initially thought. Optimally, you should jump from as far away as you can, while still landing within the door's hitbox. This will save 35 frames per door and not only be as fast as an optimal "running door transition" that I wrote above. So this will save a lot of time in the TAS as well (I hear a new one is already in the works).

Good job on the improvement! Yep, I can confirm that the fire gauntlet and the dragon are big RNG reset points for me as well.

I've never played this game casually, so I had to look up what the waterfall was. :-)
After seeing what it was, I actually think once again something unintentional might push us one step forward. I thought that instead of considering the waterfall a 3s loss (or more like 2.5s, according to my testing), why not take the opportunity to explore the potential route options this opens up? Here is what I'm thinking. Keep Goldmoon in front the whole game and use the waterfall to refill her energy. At the Skeleton before the fire gauntlet, activate "Protection from Evil" (PFE) and "Hold Person". As usual with this game, it's a bit difficult to accurately time alternative routes because you never know how cooperative the enemies are going to be, but here is roughly what I'm getting:

Current route:
* Swapping characters - Takes roughly 3.5s, but the dwarf is rarely cooperating, so Goldmoon would probably lose something in the vicinity of 1s by trying to force her way through. So swapping characters costs ~2.5s?
* Killing off characters - We kill off 4 characters. Each death costs around 50 frames. However, with good luck Raistlin can be killed off at the same time as the blue guy, saving 50 frames. So killing off characters costs around 3.5s or, if lucky, 2.5s.

Waterfall route
* Visiting the waterfall - costs ~2.5s, according to my tests
* Activating PFE - just over 2s
* Presumably less hits taken in the fire gauntlet and in the next room with the green guy and the dragon - time save, but by how much?

Based on the above, I'd say the waterfall route should be 1-2s faster. I haven't done any attempts the last days, so I haven't tried this out in practice. But some limited emu testing point towards this route being considerably more RNG-dependent than the current route. I think a lot of attempts with the waterfall route will end around halfway through floor 2. And getting through the fire gauntlet and the followiing room is nowhere near guaranteed either as hits will be taken even with PFE activated.

I'll probably do a "waterfall route" session of attempts to let the statistics speak. My gut feeling is that this might have to wait until someone will attempt to really push the limits of this game though. But let's see how things turn out. It's regardless pretty cool that there is a faster route (at least in theory) that can be attempted.

My pb is down to 4:51 (from a few days ago). There wasn't much I was happy with in that run, so I'm not going to save it, but it's in my latest stream. It shoud be available for a few more days. Fun fact is that it was sub-TAS. Obviously only thanks to the door transition trick, but still. :-)
Edit history:
Vriaeliss: 2020-06-20 11:41:28 pm
Vriaeliss: 2020-06-20 11:25:58 pm
yeah, had to learn the nuance of the door jumping transitions. it does make it some much smoother, but the unspoken "You need to actually land within  the door's window" part kinda looped me for a bit.

I really like the thought of full time Goldmoon too. Would make the fire room and dragon SO much more consistent and if it's faster too, all the better!

It seems to be faster and about as reliable as just pausing through the drawf on floor 1 to just do a low attack and hit him then jump to the doorway. Wondering if a Cure Critical would ultimately be faster than the waterfall - as you can do it later if Goldmoon's HP is good to help get her closer to the fire room to begin with... unless there's a trick you have for getting through the ghost swordsmen and avoiding the magic missiles? I haven't found a consistent spot to jump to cancel their spells, but it works sometimes, otherwise I tend to get to the room before fire room with almost no health on Goldmoon.
I haven't gotten around to much practice or attempts since my last post, but I have at least played around a bit offline with the waterfall route.

I'd like to start by saying that my pb had a good first split, so I think it can be used as a good benchmark.

By jumping through the dwarf with Goldmoon, I usually got a -2.3 split (surprisingly consistent, assuming I didn't get completely stuck). Attacking the dwarf and jumping through without getting hit gave me up to a -2.6 split. The fastest first split I managed to get was howeveer by quickly running back and forth, a -2.7.

Now about the second floor... I made it a few times to the fire gauntlet, but despite having PFE activated, I got pounded very quickly each time. I know it's possible to get through that section, but the odds are bad. And then it's the green guy in the next room that can be tricky to get through as well. So theoretically possible, but would require a serious grind to get.

How much time is lost in floor 2 with the waterfall route? Due to the difficulty, I decided to give up trying to get a floor 2 split from a proper attempt and tested it on emulator instead. With what I consider roughly equivalent attempts, 260 frames were lost with the waterfall route. 130 of those frames (=2.2s) were due to having to use cure during the menu trip. So assuming fantastic RNG (= no need to use cure), the waterfall route would save around 2.7-2.2 = 0.5s. And that's actually assuming average RNG of the current route. In the comparison run with the current route, I killed off the last character in the pit room, which is slower than getting past the dragon as Goldmoon.

In summary, I think I will go back to the current route. The pay-off with the waterfall route doesn't seem to be worth it in my eyes.

About your question about curing. Menuing is extremely slow when scrolling upwards, so cure critical is very time-consuming. Light cure gives a random health and can sometimes give a huge chunk of health back. So that's the cure option that should be considered. It's still an around 2s cost though. Anyways, what you're describing is option 'a' in one of my posts above. I have tried it out, but I abandoned it after not getting any attempts far enough into floor 2. And that was despite including additional stops for dodging projectiles etc. It's theoretically faster though.

With the current knowledge, there is no method for getting safely through the swordsmen and those type of enemies. I have the impression (but no proof) that jumping through enemies on average works better than running through them. Otherwise, you can try to hit them and run through. That way you can avoid getting hit at all, but it's far from guaranteed that you get through, so it's pretty risky imo to rely too much on that strategy.
Edit history:
ktwo: 2020-11-08 08:21:19 am
I just got a 4:46 (or maybe high 4:45 - will have to retime) in my last stream. A few things could still have gone better, but I'm happy with the time. I've put enough attempts into the game to reconize that the RNG I got was well above average and it would require quite a bit of effort from my side to push the time even lower. I'll document the run and submit it. And then wish good luck to the next person trying to push the time lower.

Edit (081120): For documentation purposes, below is a copy-paste of the updates about this game from my twitch channel:
(2020-07-12): 4:46 (or high 4:45). There are still things that could have gone a bit better, but I'm done with this game until something new is discovered.

(2020-07-05): Mid-4:47 in today's stream. It was honestly a pretty good attempt (by my standards). The biggest time loss was around 1s near the end of floor 2. It was only because of average RNG though. It would be wrong to say it was bad, but it could have been better.

(2020-06-15): 4:51 in today's stream. Several mistakes and quite a bit of bad RNG, so it's difficult to know what a goal time would be. Maybe 4:47?

(2020-06-10): New major discovery. By jumping to doors, the room transitions are sped up. This will likely save 10+ seconds! So scratch what I said below. I'll continue the pb attempts.

(2020-06-09): And now it looks like I'm done. I achieved a 5:02.2 in today's stream. There is still a little time left to gain from better RNG, but overall I'm happy with the time.

(2020-06-07): New pb of 5:02.8 (https://vimeo.com/426772682). The lower time is thanks to better RNG. An unfortunate execution mistake in floor 3 cost ~0.4s though.

(2020-05-23): A week after my previous update, I got a new pb of 5:03 (https://vimeo.com/417686667). Since then I've held off on attempts, while waiting for a new TAS to be released (in the event any new time-savers were to be found). The new TAS is out now, so I'll give this a few more attempts. There is not really anything new in the route. It's basically just to see if I can get better RNG/execution.

(2020-04-04): How close to sub-5 is possible? A new TAS was recently released, featuring a few new ideas that can easily be implemented in real-time. This will be enough to take us one big step closer to sub-5, but actually getting below the 5min mark currently appears to be theoretical due to the RNG involved. So far, it has resulted in lowering the time from my previous pb of 5:11 to a 5:04. There is still room for improvement through tighter execution and better RNG. A 5:02 would probably feel farily solid, but taking it lower than that might require route decisions that rely too much on RNG.

Edit (2020-04-10): New pb by 0.15s during today's second stream. Around one second was lost due to bad RNG of one of the trolls in the last room. Otherwise, it went as planned.

Edit (2020-04-11): I lowered my pb to a 5:03 during today's stream.
Edit history:
Vriaeliss: 2020-07-16 12:54:21 pm
Vriaeliss: 2020-07-16 12:37:45 pm
Awesome! Very well done Cheesy

Finally got to watch it - WOW that finish. Super lucky Riverwind didn't jump in front of Goldmoon on that troll. Are you uploading to Vimeo or SRL?

Frame count I got 8573 Frames = 4:45.767 (at 30 fps)
Feel free to just verify the run in the verification thread if you weren't going to! Always better when it's other runners.