I don't think this game lends well for a third type of run which would be interesting enough. It's not like SMW in which a extremely useful item warrants a small-run only , significantly different from the standard cape run.
Hooray for thread necromancy, I guess.
Regarding a no coin run, I tried that one time. I gave up on the first level. I'd love to see a no-coin run of as much of the game as possible, tool-assisted or otherwise.
Regarding a 100% run, we need one of those as soon as possible. The 100% run on SDA is pretty damn awesome; I can only imagine how much more awesome it would be with tool-assistance. Since you technically have to spend the Star Coins and not merely collect them, and you can't spend the last 80 of them until after you beat the game, does that mean the TAS should forgo using a time-saving game reset to show the credits? (Actually, now that I think about it, you might not be allowed to save until the credits finish playing, so this question is pointless, but it was on my mind for a LONG time.)
Personally (speaking as an outsider to this particular game's speedrunning community) I think that a 100% TAS shouldn't bother to actually spend all the coins. Spending coins is not interesting.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
Personally (speaking as an outsider to this particular game's speedrunning community) I think that a 100% TAS shouldn't bother to actually spend all the coins. Spending coins is not interesting.
The Speed Demos Archive run spends all of the Star Coins, because doing so is necessary to get all three stars on your save file, and also because spending Star Coins to remove a sign post gives you the opportunity to save (the run is segmented). I'd say a 100% run of the game should do everything necessary to get a three-star save file, as that is how the game indicates 100% completion, and that the Star Coins should therefore be completely spent.
I'd say the coins should be spent, as it gives the run opportunity to use extra power ups that otherwise would not be used, because the time saved from using the power up is not greater than the time spending the coins. A 'individual level time' vs 'full game time' argument I guess.
Er, just spending the coins doesn't get you the powerup -- you still have to go to a mushroom house and get the item there. Or am I misremembering?
The SDA run uses a few Mushroom House power-ups. Keep in mind that due to the run's objective, the Star Coin signs have to be removed regardless of whether the Mushroom Houses behind them are being used, so the amount of time used on these ventures is smaller, but still significant enough that the amount of time saved has to be worth it.
The most memorable such occurrence, for me, is that one level in World 7 where you ordinarily use Bob-Ombs to explode the blocks around the secret exit pipe; the SDA run uses a perfectly timed Mega Mushroom instead. Mini Mushrooms and Blue Shells are also obtained from Mushroom Houses for use in levels.
Hooray for thread necromancy, I guess.
Regarding a no coin run, I tried that one time. I gave up on the first level. I'd love to see a no-coin run of as much of the game as possible, tool-assisted or otherwise.
WRT no-coin, you can get to 8-Tower1 with no coins, even warpless; I've done it on console. (Mini Mario is your friend.) And yes, the first level is surprisingly difficult that way. (I'm still very interested in whether 6-1 is possible coinless; that level can be skipped in a warpless by doing world 7 instead, but you can't do warpless no-alt-exit without 6-1, the best possible there is one alt exit, whether coinless or not. 6-1 is not obviously impossible, but I can't do it; the walljump section is too hard for me to avoid the coins in.)
100% WIP
Here's another 100% WIP I started on Desmume 0.9.8 with Advanced Bus Timing off, but had to stop again because I got stuck at 2-A. I can't seem to jump to the second springboard with the fireball > jump trick. If you see at the end of the WIP, I would be in perfect shape to go straight to World 2 Castle and do the last fireball trick to go to World 4. It's also interesting that after powering up to Fire Luigi, I can't seem to super corner boost in 2-A. Though that's not a huge issue, since I only lose a few frames, the bigger issue is the jump at the springs because I can't seem to jump far enough. If anyone can help me figure this out, I would greatly appreciate it.
Oh my gosh, I'm really sorry I didn't get around to giving feedback on your last WIP - I saw it, but I was lazy to reply at the time and then completely forgot about it...
Definitely amazing work so far! Some comments on the levels:
1-2 (Secret): Loved how you used the animation stalling trick on the teeter-totter! I assume that was to stay at a higher speed without having to go into the shell.
1-T (Secret): Clever usage of that shell-speed gaining trick in the room with all the moving blocks! I have suspicions that getting above 466 may be possible (which goes for any% as well); I did a quick test with the inclusion of that trick and was something like 14 frames off.
1-3: Yikes, I bet it was a pain to optimize each mushrooms' movements for tiny speed gains.
1-T (Normal): Haha, this level just looked fast-paced.
1-A: Love the little shell trick on the ground!
Also of note (and I'm sure you already know this, but to be sure), if you want to avoid double jumps, it's better to stay on the ground for a frame without moving left/right and then jump, rather than holding back to slow down.
1-5: Maybe you could keep the shell for a little longer to do a few slope boosts on the mushroom edges (to stay at 3100 speed); just assuming from simply looking, so it may not even save frames...
1-Castle: Hahaha, love the backward movement at the beginning, complete with a backward super corner boost!
2-1: Finding ways to kill those Pokeys is a lot of fun--I like what you did here!
2-3 (Normal): Like with 1-T, I love how the movement looks here. Levels that make you go all over the place look refreshing.
As for 2-A, it seems that it's because you didn't actually perform your first jump and it's not allowing you to perform the triple jump because of that. I went back in your case and actually did the first jump, as well as testing if ducking and un-ducking at the ledge for the first jump also qualifies, and it worked. No worries there!
But as for the level not allowing you to do super corner boosts, I have no idea... Maybe it's because the Red ? Block boost repositioned you in a way that won't allow you to do them at all anymore. But you've done that in other levels as well, with super corner boosting working just fine, so I'm not sure...
On another note, it seems there's more room in your WIP to optimize subpixels (or whatever they're calculated as in this game) whenever wall jumping/alternating from one direction of movement to another, rather sorry to say... I redid some parts of the 1-1 you had (hope you don't mind that I've done so) and saved 8 frames, 3 underground and 5 after the underground portion (though ~4 there came from wall jumping).
http://dehacked.2y.net/microstorage.php/info/579330649/1-1.dsm
I understand if it's too much work, though! It's definitely time-consuming to have to optimize every turn for a really long run like this.
All in all, keep up the great progress! This looks really nice so far and I'll definitely try to follow your WIPs more closely. You've motivated me to do more stuff with this, so I'd be happy to help whenever need be!
1-2 (Secret): Loved how you used the animation stalling trick on the teeter-totter! I assume that was to stay at a higher speed without having to go into the shell.
Well, it was to prevent going into the shell so that I could keep holding the koopa shell. I think going into the shell and triple jumping off the teeter-totter is still faster like in the any %, but since you need to bring a koopa shell to save more time and the only other way is to use the koopa jump trick used by Greenalink in his test videos (which unfortunately is slower), I think this might be the best solution. Credits to WaddleDx for the idea (I got the idea from his video).
1-T (Secret): Clever usage of that shell-speed gaining trick in the room with all the moving blocks! I have suspicions that getting above 466 may be possible (which goes for any% as well); I did a quick test with the inclusion of that trick and was something like 14 frames off.
That was a nice discovery you made, it saved around 10~20 frames in that room! Sadly, it wasn't enough to beat 466. You were faster than me though, the fastest I reached the flag, I was 23 frames off. It would just be nice if you could just get the blue shell here...
1-5: Maybe you could keep the shell for a little longer to do a few slope boosts on the mushroom edges (to stay at 3100 speed); just assuming from simply looking, so it may not even save frames...
Oh, I didn't know that, I might try seeing if it does.
As for 2-A, it seems that it's because you didn't actually perform your first jump and it's not allowing you to perform the triple jump because of that. I went back in your case and actually did the first jump, as well as testing if ducking and un-ducking at the ledge for the first jump also qualifies, and it worked. No worries there!
Oh, so you actually have to do a first jump. And I thought running off a ledge and double jumping would work. I wonder why I didn't think of actually trying that. Thanks though!
But as for the level not allowing you to do super corner boosts, I have no idea... Maybe it's because the Red ? Block boost repositioned you in a way that won't allow you to do them at all anymore. But you've done that in other levels as well, with super corner boosting working just fine, so I'm not sure...
Yeah, that is just weird, I suddenly can't super corner boost anymore. I think I might just have to move on, I only lose a frame or two from not boosting and I can't think of a better setup to go to World 4.
But as for the level not allowing you to do super corner boosts, I have no idea... Maybe it's because the Red ? Block boost repositioned you in a way that won't allow you to do them at all anymore. But you've done that in other levels as well, with super corner boosting working just fine, so I'm not sure...
Dang, 8 frames... That's quite an improvement though! Only a few frames and the level could be a 364!
I understand if it's too much work, though! It's definitely time-consuming to have to optimize every turn for a really long run like this.
I'm going to try to hex-edit the improvements in. The game is friendly with hex-editing and as long as luck manipulation is kept the same (number of double jumps/wall jumps), I don't think it should be a problem (If you don't mind me using your input from 1-1).
All in all, keep up the great progress! This looks really nice so far and I'll definitely try to follow your WIPs more closely. You've motivated me to do more stuff with this, so I'd be happy to help whenever need be!
Thanks! Appreciate the help.
I do have one question though: In the underground of 1-4, I can't seem to wall jump off the pipe with shell-Luigi like any other power-up (like Fire in 2-1). Could you confirm this if possible?
Well, I gotta say, this looks amazing so far. Along with everything LBY said, I noticed you ground pound on two columns of bricks at once instead of just one in 1-1 and 2-2 for the Star Coins. I'm sure destroying only one column would be faster than two. I understand for 1-1, but 2-2 seems like you could only destroy one column and move on from there. I have a feeling that you did that for a reason or you just didn't know :P
That was a nice discovery you made, it saved around 10~20 frames in that room!
Ahh, yeah, thanks. I honestly don't know how I discovered it, haha. We need to figure out what allows it to work... actually, there's a lot of things in this game that need further investigating. Stuff like pipe entry delay, wall jumping mechanics (namely, why it won't work for certain jumps)... even that case you've just come across! (I'd like to get a full understanding on what specifically allows double/triple jumps.)
No problems with using my 1-1 attempt! I don't mind at all if you use any of my inputs in general. I've actually lined up the manipulation to stay the same, so it should fit in just fine (with the exception that the score and coin count will be a bit different). I'd also be fine with checking up on the other levels you did, though I'm afraid I'd be too overbearing if I were to do so, haha...
Edit: Hold that, I made a silly mistake (failed to cancel the ground pound to the third Star Coin). I'll improve it soon...
About wall jumping off of a 2-tile-high structure with the shell, I tried before (and just tried now) and was unable to pull it off as well. It may have to do with how you can't move while ducking in the shell; not sure... I guess you shouldn't worry too much about it.
I forgot to mention before, but excellent job on map manipulation! As far as I can tell, World 1 and what you've done in World 2 looks perfect.
Also, mindnomad: It looks like breaking both columns in 2-2 is faster since it allows you to build momentum (you barely can in a 1-tile-wide gap).
___
Just to note this somewhere:
With the shell, slightly slanted slopes allow you to keep 3100 speed, and 45-degree slopes only let you keep 3087 speed.
I'd like to get a full understanding on what specifically allows double/triple jumps.
What exactly do you want to know about this? I'm sure just jumping on the first possible frame would allow for a double jump, unless you mean something else that is...
What exactly do you want to know about this? I'm sure just jumping on the first possible frame would allow for a double jump, unless you mean something else that is...
Oops! I meant specifically when ducking to cause any of the three jumps, as it doesn't work sometimes. Along with this, I'm not sure why ducking and quickly hitting a ceiling to cancel your duck allows you to go into your double/triple jump.
Alright, I fixed up the ground pound and was able to save another 8 frames--please use this one instead:
Sweet, 364!!!
mindnomad wrote:
What exactly do you want to know about this? I'm sure just jumping on the first possible frame would allow for a double jump, unless you mean something else that is...
Oops! I meant specifically when ducking to cause any of the three jumps, as it doesn't work sometimes. Along with this, I'm not sure why ducking and quickly hitting a ceiling to cancel your duck allows you to go into your double/triple jump.
Yes, that would be good to know because World 1 Castle might be able to be done a few frames faster if it was possible to duck > double jump and then triple jump to the pipe, but the duck > double jump doesn't seem to work there.
No problems with using my 1-1 attempt! I don't mind at all if you use any of my inputs in general. I've actually lined up the manipulation to stay the same, so it should fit in just fine (with the exception that the score and coin count will be a bit different). I'd also be fine with checking up on the other levels you did, though I'm afraid I'd be too overbearing if I were to do so, haha...
No, absolutely not, by all means, go ahead and look through the levels. It's much better if you discover improvements early on so that they can be hex-edited in more easily than if you find them later.
If you guys are interested in TAS'ing this game again and want to collaborate, you're welcome to join. There's also a few people on youtube (I don't know if they have accounts here) who are also planning on TAS'ing the 100%, maybe we can get a group going. If not, you're more than welcome to keep pointing out improvements, the sooner, the better.
I don't want to download an emulator. Give me a YouTube encode proving the 364 time, or it didn't happen. Kappa
Just an excuse to use another Twitch emoticon.
If you guys are interested in TAS'ing this game again and want to collaborate, you're welcome to join. There's also a few people on youtube (I don't know if they have accounts here) who are also planning on TAS'ing the 100%, maybe we can get a group going. If not, you're more than welcome to keep pointing out improvements, the sooner, the better.
I'll see what I can do. my issue is having a crap computer, so the emulator will be slow, and having a bunch of individual WIP dsms would be troublesome to me. I guess I can just try out some strats for other levels or something...
No, absolutely not, by all means, go ahead and look through the levels. It's much better if you discover improvements early on so that they can be hex-edited in more easily than if you find them later.
Ah, alright... I'll start trying to look for any possible improvements soon.
zk547 wrote:
If you guys are interested in TAS'ing this game again and want to collaborate, you're welcome to join.
I guess I just basically joined now, heh...
zk547 wrote:
There's also a few people on youtube (I don't know if they have accounts here) who are also planning on TAS'ing the 100%, maybe we can get a group going. If not, you're more than welcome to keep pointing out improvements, the sooner, the better.
The more help the merrier, I suppose! I'm personally not too in touch/familiar with them all, though...
mathgrant wrote:
I don't want to download an emulator. Give me a YouTube encode proving the 364 time, or it didn't happen. Kappa
I just looked at 1-1 again and the walljump off the pipe looks awesome! I've been trying to see if I could save time with that trick in other places (namely 1-2 (2nd) (right before the flag) and 1-5(in the sky area with the third star coin)), but without much success. Maybe I'm not doing it right, though I think in 1-2, it might still be faster to walljump off the wall on the left and if it does save time in 1-5, it would also have to be redone because then I would become Fire Luigi earlier to take advantage of the walljump > fireball trick. Though that would mean that 1-Castle also has to be redone...
1-5: Maybe you could keep the shell for a little longer to do a few slope boosts on the mushroom edges (to stay at 3100 speed)
I tried to do this, but I can't seem to get it to work. Could I see a demonstration of it?
Sorry for my inactivity. I've lost motivation to work on this faster than I thought... I'll get to looking through at least 1-2 eventually.
Yup, it seems that the wall jump you already have for 1-2 is faster. For 1-5, though, it did appear to save time (tests below).
To do the mushroom boosts, you land on a mushroom edge, run down for a frame, go into the shell without pressing left/right for the next frame, and then jump on the third frame (while still staying in the shell). Though, I think I recall running into a situation where it just wouldn't seem to work on some of the mushrooms. Maybe it has to do with "decimal pixel" positioning...
I went ahead and made two tests of 1-5, for you to look at the inputs for the mushrooms and to compare losing the shell early to losing the shell later:
Lose shell late - http://dehacked.2y.net/microstorage.php/info/1509080633/1-5%20Test.dsm
Lose shell early - http://dehacked.2y.net/microstorage.php/info/308023824/1-5%20Test%202.dsm
Notes on the tests: Interestingly, the two strategies are only (about) two frames off! Losing the shell early wins out because having the fire powerup in the bonus room makes wall jumping off the pipe more worthwhile, since you can use a fireball to gain enough speed from the wall jump to land on the moving mushroom and gain even more pixels as it moves.
These are both unoptimized, though, but even so, I'm doubtful that there's much else to be saved in the "lose shell late" strategy to beat what could potentially be saved in the other strategy. Hopefully I'm not overlooking anything, though!
(I guess 1-Castle may have to be redone, then, unfortunately...)
Wow, that's a lot closer than I thought!! I think losing the shell early is the way to go. I think I can still hex edit some parts of 1-Castle, really only the middle part, from where I originally get the fire-flower to triple jumping to the pipe, has to be redone.
Those things aside, I've been talking with LFNMai (Firestar7891 on youtube), who is also interested in TAS'ing 100% and collaborating, though wants to use desmume 0.9.7 because it has less lag. He pointed out something interested about Bowser Jr. Apparently, his health increases from World 4 to World 7 when he becomes masked, so you don't have enough time to shoot enough fireballs before stomping on him for the kill. Interestingly, his health is reduced in World 8, which is why the any% has no trouble. I haven't been able to try it out for myself yet, but it would be nice know if it is possible to still beat Bowser Jr. with the fireball > stomp (and there was something that was overlooked) or if new strategies have to be created for those Tower levels.
So after re-doing 1-5, I managed to save 2 more frames. I tried to see if I could get a reserve flower in 1-Castle to save time in 2-2 and if you look at the dsm, I get really close to making it, but fall short. I don't think there's anything I overlooked, but it would be really nice if this strategy worked.
1-5 redone and 1-Castle test
Excellent job on the frame saving! Wow, I thought running down mushrooms whenever possible in general was saving pixels. My bad!
I just tried to get under the moving wall in your WIP as well and wasn't able to, sadly. (For future reference (to anyone else), screen scrolling won't affect the timing of that specific large wall portion.)
To LFNMai, sorry to say, it is not less lag that decides what version is generally preferred, but the version that emulates lag most accurately. I'm willing to believe the most recent versions are the ones to emulate lag most accurately, I think.
Speaking of which, 0.9.9 was apparently released a bit ago. As with the other versions, the RNG is different, so other runs won't sync on it. I think it'd be okay if you stuck with 0.9.8, though, given how much progress you've already made and how much trouble it'd be to resync it.
Interesting point with Bowser Jr.'s health; I personally never knew that! Odd that they'd randomly decide to reduce the amount of health for him in the last world. (Thinking about it, that was actually time-saving for when we worked on the any%; simply copy-pasting input from W1's fight to W8's was effective!)
Also... I'm really, really sorry. I actually held off making another post here until I at least had some form of progress or update to show, but I just couldn't find any motivation to work on this anymore already... I don't have the same patience for optimization as I used to have during the making of the any% run. Extremely sorry for the false joining. :<
Again, though, I'll gladly continue to provide feedback or ideas as they come along. Please keep up the great work with what you already have, and apologies again.
