Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
The halting problem is indeed decidable for systems with finite memory, on systems with much much much greater memory; imagine running a NES emulator and savestating every step, then checking to see if two of the savestates were the same (if they were, you have an infinite loop). This is different from solving the halting problem for algorithms in general, and no existing computer is powerful enough to do it even for a NES.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
How easy would this route be to replicate without TAS tools? In particular, are the sequence-breaking glitches frame-perfect, or are they possible by hand with a bit of retrying and experimentation?
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Voted Yes. IMO, it should obsolete the warped walkathon, but mention it and link to it in the description, so that people who want to see a walkathon with warps can find it easily.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Yep, you need to unlock characters to be able to use them (via taking them into the level or via cap) in the majority of levels, although some have caps even for characters that are still locked. (And there are some stars that obviously require Wario to do glitchlessly, such as the stars in what used to be the switch palace under the moat; there are no caps in the castle grounds no matter what, and there's a block only Wario can break blocking the entrance.)
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
I've been thinking about it myself. In order for standard AI techniques to work well in a platformer, it should a) have an obvious way to measure progress within the level (e.g. a run-right-for-great-justice game, where you can just take the X coordinate+subpixel), and within the game itself; and b) have a relatively small maximum speed for the character (this allows you to come up with a good A* optimiser so that the AI can focus on later levels rather than continuously trying to improve the early levels). SMB1 might be a good game to try out that sort of bot on.
Developing a bot that worked for even one game would be very difficult, though, and to play it optimally rather harder. (There are already bots using this sort of technique that play randomly-generated Mario clones, incidentally; I don't have a link handy, but it shouldn't be too hard to find one.)
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Err, sorry, thinko. I did know there were 150, I just got my categories muddled. (I've completed the game 150-star myself.)
I agree that Luigi is almost certainly useful enough that unlocking him is correct in a glitchless any% (which I suppose would make glitchless low% a separate, but probably uninteresting, category). I doubt that Wario saves enough time to be worth unlocking in anything but a 150-star.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
There are two kinds of TASes submitted by new TASers. The first sort is generally badly optimised, no fun to watch, and either fails to beat existing records (often even realtime records) or is of a bad game choice. The second sort comes out of nowhere and blows everyone away. I haven't watched the run yet, but it seems reasonably likely that this run is of the second sort.
Also, I suggest "no forward warps" as a category name. (Or perhaps "all levels" if it's accurate; I don't know the game well enough to tell that.)
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Note that I think it's possible to complete the game 80-star glitchless without unlocking Luigi or Wario; IIRC they're both optional characters. (120-star is meant to be impossible without unlocking every character.) You can do some fun things with them, though; for instance, I somehow completely missed that the ! switch in BitDW allowed you to get the first red coin, but managed to get the coin anyway, without dying, using Luigi's special slow-fall backflip move.
I don't think a 120-star run would be that awkward, although obviously it'd take much longer than an 8-star run simply due to having much more stars. Luck-manipulating the silver rabbits shouldn't take too long at all, especially if you're grabbing rabbits anyway to glitch through walls.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
applepieman wrote:
Is there a different ending for getting all the birds and taking no damage?
I thought it was a enjoyable run, voting yes.
No, but if you complete each level as a no-damage run at least once (and probably also get all the birds, I don't think anyone ever doesn't do that) it unlocks three bonus levels at the end, which are long, empty-ish levels with no healing points.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Watched the whole thing, voted Yes. (I've never played the game.)
What were the 100 jumps in the second Belome battle for? Based on what happened subsequently, it was clearly to get some sort of reward for the 100-streak, but I didn't see what it was and don't know why it was important.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
From my point of view, I can generally only watch runs from the encode, but I don't mind waiting until there's an encode to watch the run (and I can still comment on the submissions thread if I know the game). I see no issue with letting people like judges get a head start on watching a run, when I know I'll eventually be able to watch it myself anyway.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
sgrunt wrote:
Hi all:
I have produced an additional experimental downloadable encode. This encode was created with 10-bit x264 in the YUV444 colour space - only very recent builds of media players or codecs are going to be able to play such an encode back, but the resulting visual quality should be significantly higher for a very similar file size.
I would appreciate feedback on who is able to play back this encode successfully and what people think of it relative to the encode I previously posted.
I tried playing it back in Totem, with the default set of codecs. It thought it could play it back, but it made a complete mess of it, so I guess it was lying.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Voting yes, on condition that the movie's category is changed to explain that it starts from the sixth world (otherwise, someone looking at the time and not the description might come to an invalid conclusion about how much faster the TAS is than a speedrun). It's nothing special, but it's passable enough.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Serious question arising from this frivolous submission: with compression techniques available nowadays, I don't see why you couldn't work out a repeating sequence of input sufficient to max out the high score counter, and compress it to within TASvideos' size limits. Even if the uncompressed file wouldn't fit on a hard drive, it'd be simple enough to patch the emulator (or on UNIX/Linux, use mkfifo/zcat, assuming that the emulator didn't seek in the input file) to be able to play it back without decompressing it.
The next stage would be to implement some sort of fractal playback support in the emulator, HashLife-style, so that it could play back an incredibly long compressed movie file by only emulating a small fraction of the steps. For a game like this, that might actually be viable.
Somehow, I doubt anyone would put in the effort. But it'd be a work of art if they did on a technical level, and then who'd care about the game?
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Just finished watching this, and it really isn't as bad as everyone seems to be making out. Imagine what a TAS of an RPG typically looks like: all battles run from except a few plot battles that are beaten with cheap tactics or luck manipulation, leaving basically the entire run as cutscenes, fetchquests, and minigames. And isn't that just what this game is? As a benefit, the plot makes rather more sense than the average RPG's.
Voting meh, as per my policy on voting on RPG runs that skip or power through all the interesting bits (in this case, there are no interesting bits to skip, but I don't see why that should make a difference). The bits of the game that are done (all of it), whilst a little repetitive and uninspired, are done pretty efficiently and precisely, and I don't see why the run shouldn't be published so that fans of the game can watch it; it's not like we're likely to get a better run of the same game any time soon (unless it gets frame-warred). I really don't think popularity of a game should factor into a decision on whether to publish a TAS, nor how fun the game is to play in unassisted conditions.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
That information about the controller line is pretty interesting, as it means you could measure lag with a multimeter or modified controller (NESbot-style). Thus, you might not even have to get the emulator to calculate lag for you, if you can measure it from the console itself.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
I found this site via YouTube. The links to TASvideos on every publication really do help! (Strangely, it was a while before I found the site after learning of the existence of TASes, because most non-tasvideos TASes don't link here.)
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
I've found a fun glitch that probably needs TAS tools to test out (and I don't use Windows, so can't run Hourglass). Try quicksaving and reloading inside the Gravitron; the timing I used was just after the countdown started, but other timings might work. As you reload, hold down right. You should escape the Gravitron out of the top with the timer still going. Weird things seem to happen as a result; the timer running out does strange and unpredictable things (in a way that reminds me of memory corruption, and it's caused double-frees on the Linux version of VVVVVV, and so is probably well worth examining with a debugger), and even with the timer going there's strange behaviour (Gravitron enemies spawning outside the Gravitron and mostly getting stuck on walls, and Viridian's hitbox seems to be a different shape too, allowing him to escape off the top of the map and end up out of bounds).
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
mklip2001 wrote:
You did a nice job with this, Randil, but honestly as someone who's never played the game, I'm not sure I really get the appeal.
It looked to me, at first, like you could pretty much just spam the drill button to get down to the bottom fast (sure, it's more efficient to go through bigger blocks of the same color, but it didn't look necessary if you wanted to merely beat the game). Looking at GameFAQs told me that there's also the challenge of having enough air and not letting falling blocks land on you. With that in mind, the run looks more impressive, but I'm still not sure I see the main challenge.
I'll hold off from voting for now. I'm inclined to say Yes since the play is fast and colorful.
Spamming the drill button doesn't work as soon as X blocks are involved; if you go sideways to try to avoid them and aren't careful, you typically end up getting crushed, and drilling straight through them gives crippling penalties. (There are unbreakable blocks in some modes, too.) It's also far from the fastest strategy, as it's faster to run into the area vacated by drilling a large block than it is to drill a sequence of small blocks. (Some characters are faster at drilling, some at running, but running tends to be faster even for characters fast at drilling.)
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
It's only worth going to the trouble of unlocking a character if you can get the unlock in and still finish the game before you would if you didn't play through the unlock, for a run like this. And I rather doubt that's the case.
IIRC the Time Trial runs in Mr. Driller adapt their time limits for each character, so it's not even as if you're using a suboptimal character anyway. You're just playing a particular set of levels.