TASVideos

Posted: 8/3/2011 6:36 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

I thought this was a bit slow to start with, but the end made up for it, and the fact that the sequence is being broken is obvious even to someone who knows nothing of the hack. Yes vote.

Posted: 7/20/2011 6:20 AM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

Another thing I'd suggest is to add options for the player which make it both harder and faster, potentially stacking indefinitely. (A simple example is the shoes that double your speed in Sonic 2; many casual players avoid picking them up because they make Sonic harder to control and require faster reflexes, although they're obviously useful for a speedrun. For a TAS, imagine if they carried across levels, with the effect stacking, but only if the player didn't take a hit; now, that would allow players to make the game basically impossible for themselves if they wanted, but avoidably, and a TAS could actually get away with it. There are more complicated variations on the same basic priniciple; low-grinding runs in RPGs, for instance.) I'd say, contrary to many of the other comments, having one autoscroller in a game is good for a TAS, as it gives the author a chance to play around without losing time, making the result more entertaining. But you have to give sufficient material in the level for them to play around with. Another good thing to add is to provide the player with things to do that are generally "not worth it", or only obviously useful in specific situations, but which have a lot of other uses to players with a lot of planning. (For instance, consider the weapons you gain in a Mega Man game. Casual players often only use them in boss fights, using the buster elsewhere, unless they're really overpowered; but speedrunners (both TAS and non-TAS) find all sorts of interesting uses for them, both intended and unintended.) Also, if the game has more than one way of moving around, and it should (running along the ground vs. running with repeated jumps is a common case in platformers), try to make them take identical time in ordinary cases; that gives TASers the scope to use one or the other to hit small timesavers or make the run more entertaining. If, say, running was always faster than jumping, then otherwise interesting techniques involving jumps would probably be suboptimal and have to be avoided. One other thing I'll note is that although hard to do, a TAS can be much more interesting if the game has scope for creativity that doesn't interfere with the run itself. ("Shooting in time to the music" is a simple example that's possible in some but not all games, especially as in a TAS you can often vary whether you kill an enemy early or late without losing time. Some of my favourite TASes are ones where you have similar control over the game, but in a much wider scope, allowing for more impressive versions of the same basic concept. Making a game heavy on sound effects can sometimes allow for this, but won't automatically manage it.)

Posted: 7/16/2011 3:37 AM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

Why are you saving at the end of each level? (And if you have to save for some reason, wouldn't it be faster to save in a different slot each time to avoid the prompt that asks if you want to overwrite your save?)

Posted: 7/15/2011 6:22 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

I'd like to see the category name for movies dominated by a single glitch specify the glitch in the category, e.g. DOS NetHack "memory corruption glitch" vs. DOS NetHack "minimum turns, no memory corruption". In practice, there tend to be only two (or three, depending on how you count) sorts of glitch that end up dominating a game enough to require a new category: memory/save corruption glitches, and out-of-bounds glitches that lead to skipping most of the sequence of the game. Instead of "forgoes time-saving glitches" (which ones?), I'd like to see "corrupts memory" and "forgoes memory corruption", "corrupts save data" and "forgoes save data corruption" (we already have these two, so no change is needed there), "goes out of bounds" and "forgoes going out of bounds". A good example of why this sort of thing is necessary is in the Generation I Pokémon games, to distinguish the categories in [1702] GBC Pokémon: Yellow Version "save glitch" by p4wn3r in 01:27.23 (save corruption) from [1700] SGB Pokémon: Red Version "warp glitch" by p4wn3r in 41:02.38 (arbitrary memory corruption) from [950] SGB Pokémon: Blue Version "trainer escape glitch" by primorial_soup in 1:18:58.78 (out-of-bounds). Note that all three of the movies are tagged "heavy glitch abuse"! At the moment, 1700M isn't tagged with a branch name, nor is 1702M, which might leave the viewers wondering why they take such different routes in such similar games. I'd like to see 1702 marked "save corruption glitch", and 1700 marked "memory corruption glitch", to distinguish the branches. (Out-of-bounds is more normal; I personally consider 950 the current "normal" category for Pokémon, and would watch a no-out-of-bounds run as well, although I doubt it would be different enough from the other runs to be interesting and publishable.) I agree with the OP that "glitched" by itself is pretty unclear, especially as it's used inconsistently. Instead of trying to imply "more glitched than" or "less glitched than" or whatever (which will probably depend on the order in which the movies were originally published), why not say exactly what the category actually is? (I also say this because I know there are people who enjoy watching entirely non-glitched TASes; I like both glitched and non-glitched versions, but people who want glitchless runs have a lot more than the runs in the "forgoes time-saving glitches" category to look at, because there are many games with no known major exploitable glitches. (Keeping with the Pokémon example, Pokémon Ruby/Sapphire, while there are a few known glitches, don't have any known ones that save time at the moment. No doubt TASers are working on correcting that, but for the time being, a third-generation Pokémon TAS is going to be glitchless just because that's fastest.)

Posted: 7/10/2011 5:16 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

Hmm, this game doesn't look so interesting in a TAS as some of the other Mega Man games, but it's still OK. (Also, bunny hopping in a Mega Man game, how bizarre...)

Posted: 7/7/2011 3:57 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

System Error wrote:

Zowayix wrote:
Nope, the GS Ball only works if a certain flag is set. Obtaining the GS Ball through an event will set this flag, as will the proper cheat code. Simply hacking or glitching the GS Ball into your bag will not. All event items work this way.
It's been a while, but I'm pretty sure it works if it's just the item. I swear I was able to work it with just the item way back in the day. The flag stuff was only implemented into the third generation Pokemon games. I've found no information to the contrary that it only works if a flag is set (and in fact, many parts talk about potentially using it to get teh GS Ball), but the next day may need to come from giving it to Kurt. Which might exclude it anyway unless the run's long enough.

Just set the game's clock (which can be set upon new game) so that a day rollover happens during the run.

Posted: 6/30/2011 6:42 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

I suppose TASvideos would have to be moved to a server outside the US.

Posted: 6/26/2011 7:05 AM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

RPG which is mostly cutscenes, meet TAS which isn't mostly cutscene skips (but only because they are, by definition, quick). I approve of this solution. The actual gameplay is amazing too; some of those temple solutions are faster than anything I've seen even in other really broken runs.

Posted: 6/26/2011 5:30 AM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

That was pretty impressive. The only bit that made me snort with laughter, though, was Wallace sending out a Luvdisc, and that happens no matter what. There are places where it looks slightly sloppy, presumably because you're stallling to get the right frame on which to manipulate an event. One question: did you test whether it's faster to bike out of the Cave of Origin, or fly to the Poké Centre and surf from there? I haven't checked myself, but it most be quite close.

Posted: 6/9/2011 9:26 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

I got to about halfway through disc 2, and then got bored. The issue with this game is that it's some of the worst game choice for a TAS I've ever seen; the actual TASing serves mostly to manipulate away encounters (making the run more boring), skip dialogue (making it harder to tell what's going on), and win battles really quickly (which makes the bits that are actually luck-manipulated and tight really short). I don't think this is avoidable in this sort of game, but the end result is that the run is mostly cutscenes that would look much the same no matter what, and watching the TAS would likely be judged on the merits of the game rather than anything the runner could do. That said, I'm far from convinced that that's a reason to vote no. This could be quite interesting if the cutscenes were removed (or better still, a video was shown of the differences between the TAS and a non-TAS run, but that would be very hard to compute). I think better still would just be a long description of the strategy, and then the actual run would be mostly unnecessary. The other thing is that on a well-known game, I'm not convinced bad game choice is a reason to reject a well-done TAS like this one. There's not a whole lot of difference between published and submitted, except that publishing a run makes the best TAS of that run easier to find, for people who really want to see a TAS of the game. Voting Meh.

Posted: 6/2/2011 6:47 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

Miles wrote:

You'd have to wallcrawl up Sun Tower (or use Spider I guess?). Also, Flaahgra's flower isn't there below the artifact, so IIRC you have to wallcrawl and fall on top of it. I seriously doubt it'd be faster to skip Varia, just explaining how it'd be possible...

I assume it's impossible to do Magmoor on the starting E-tank without the Varia Suit, even with TAS luck? If it's possible manipulating health drops from every enemy, 20% might be possible via that method.

Posted: 5/25/2011 7:26 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

[20:24] <FerretFaucet> This is the first time I've drank in three weeks. [20:24] <FerretFaucet> I doubt it's a tolerance issue [20:24] <FerretFaucet> Besides, doctors recommend drinking 1-2 liters of vodka per day

Posted: 5/25/2011 4:16 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

EEssentia wrote:

If you could jump directly to the last boss--or heck--even directly to the ending by utilizing TASing, where would the fun be in watching the game? That would be utilizing TASing to its very extremes.

And yet there are a few runs that do that on this site. Notably, though, they tend to be accompanied with other runs that don't; default for TASes seems to be three categories, "glitched any%" that hits the boss or ending via any means possible (and typically skipping most of the game); and any% and 100% as per SDA. Many games either don't allow glitched any%, or people haven't found a way to make it happen yet.

Posted: 5/24/2011 9:20 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

I've been following this submission in the forums, and was very impressed about it, even if I didn't post. Not voting for now, because I'm at work and haven't watched it. I expect I'll most likely vote Yes once I have, though.

Posted: 5/21/2011 6:19 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

Potato Stomper wrote:

There's a new trick: http://www.youtube.com/watch?v=ntStTtt27L8 I doubt that it saves time compared to the current way, but I'll let you know anyway.

Was that first wild encounter, in view of a trainer, a manipulated random one, or some glitch involving holding a direction when saving? Without seeing the input, it's hard to tell from the video. (I can sort of see everything else that's going on there, though.)

Posted: 5/20/2011 9:25 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

SXL wrote:

Can't watch this, keeps blinking. What codec do I need ? Care to youtube it ?

What do you mean by "blinking"? It might be that what you're seeing is exactly what the WIP is; a lot of things happening very fast. For instance, we cast a spell over and over again in Vlad's Tower, which would necessarily change the screen between two views very quickly. (Ilari says it's a typical x264 encode with no unusual options, so it's unlikely to be a codec problem if you can watch other TASes on the sites, especially other encodes by Ilari.) Also, are you trying to watch the 30fps version, or the actual-time one?

Posted: 5/20/2011 9:08 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

New WIP encode, people! WIP by me and dwangoAC, encode by Ilari: http://sonic.net/~ac/tas/nethack-tas-dwangoac,ais523-wip-20110519.mkv (full speed, you can't see much but luck manipulation delays here), http://sonic.net/~ac/tas/nethack-tas-dwangoac,ais523-wip-20110519-special.mkv (one frame every time anything happens, with timings arranged to make it a constant 30fps; this version is probably much easier to follow, especially with slowdown). This encode charts much of the slaughter of the Castle, which is how we decided to fix the alignment issue. (We've since slaughtered the other barracks too, but that isn't in the encode.) As usual, I'm not entirely sure what I'll do next once alignment and luck are back to normal. We have several more glitches that need setting up, although we have everything we need for actual ascension now but the candles. (Everything beyond here is setting up things to make the ascension run go faster.) Much of what we'll be doing is filling levels with monsters, one way or another; there are a few creative ways to do that. We also need a huge number of cursed potions of gain level, and I'm still not sure about how to go about getting them (some other NetHack players I've been talking to pointed out that chain-wishing is the fastest method, but I don't want to do too much of that). One other thing we need in order to get really large numbers of monsters around quickly (which we'll need for Air, as we want to fill about 2/3 of the plane with monsters in a single turn) is the ability to polypile; we have a spellbook of polymorph but can't currently cast it. (I was wondering about using the cursed armour destruction bug to max out Intelligence, but don't have the items required to do it yet.)

Posted: 5/3/2011 12:27 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

There's something bizarrely wrong with the titles (the HTML titles, normally shown in browser address bars) of the pages such as http://tasvideos.org/movies.cgi?id=1234,1235,1236 where multiple movies are specified in the same URL. Spaces seem to be inserted in the wrong locations in the title, making the numbers hard to read.

Posted: 5/2/2011 5:40 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

Surely the biggest reason of all would be a game that doesn't emulate properly? More seriously, for a game to be harder to make an interesting TAS than to make an interesting speedrun, would be a game where the two routes were very different. (Imagine a series of jumps that nobody can make in realtime, that are easy to do in frame advance, and lead to a faster but massively more difficult route through the level. It's hard to see how that could happen in the absence of some really weird glitches, though.)

Posted: 4/30/2011 2:04 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

Yes vote. I found this a bit hard to follow, but likely because of the game; I could tell that a lot of effort went into the routeplanning, though. And you really show off how weak the hero is with the death abuse.

Posted: 4/30/2011 1:52 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

errror1 wrote:

that's what I figured, for each turn you would need to manipulate about 5 different things for each foocubus, but I think it's interesting in theory. I'm guessing you can't get your final alignment point during the monsters turn rather then having to kill something.

Even if we could, it wouldn't matter. We can't get permission to enter the quest until the end of turn 2000 action 1 (on Neferet's turn), so we may as well use that action itself to gain alignment. Arranging a monster to gain us alignment, acting before Neferet, would simply leave us with a blank action on 2000:1 which we couldn't use for anything.

Posted: 4/29/2011 9:31 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

errror1 wrote:

the d100 only comes up if you get a good result, if you get a level drain with Drain resistance nothing happens and you can be seduced the next turn. You still have a ton of stuff that you would need to manipulate like the 1 in 25 cancel chance, and getting them to seduce you as often as possible. But the big advantage I would think is that you could do other stuff as you waited for the monsters, like wish, practice mapping, or manipulate inventory. not that there is anything wrong with killing monsters, but it's going to take you awhile to get the alignment you need and even mass slaughter might get boring ( unlikely I know ).

Looking into it, multiple foocubus encounters a turn will be completely impractical, simply because we'd have to manipulate all of them giving the right random numbers at once. The chances of that would be incredibly remote if we were trying to do that at any sort of sensible speed. (Generally speaking, we can discard random results to change future results only on our own actions.)

Posted: 4/29/2011 4:02 AM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

errror1 wrote:

foocubus can begin an encounter of its own volition right, if it does that does it still use up your turn? If it doesn't and you can do multiple encounters per turn then with an unreasonable amount of luck manipulation and a room full of foocubus it seems like you could gain alignment quickly.

I hadn't thought of that, so I looked into it: there's a d100 turn cooldown on that foocubus ability, and due to terrain layout you'd thus, even using a huge number of foocubi, be limited to gaining 8 points a turn, as you have to wait for the monster turns to come around. (Probably slightly less, as luck-manipulating constant 1% chances while surrounded would be very tricky due to the lack of a wall.) Compare that to the two kills a turn that can typically be gained with a relatively fast polymorph form, and you get 10 points a turn, minus timings you need to move around rather than attack. (Three is also possible, but would take amounts of manipulation that I don't want to do in non-time-critical parts of the run as it'd just force us to waste the turns elsewhere instead; four is possible in air elemental form, but we'll be using that so much lategame that I want to avoid using it until I have to, to make things less repetitive.) Thus, I think the two approaches would be similar in gametime speed, but I suspect killing things is both faster in realtime, and less repetitive to watch. (Besides, what's an RPG run without the occasional mass slaughter of monsters?)

Posted: 4/29/2011 3:04 AM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

errror1 wrote:

would it be quicker to get a Helm of opposite alignment and steal from/damage some shops?

That's possible, but at least five times slower, probably much more because it's harder to repeat. (Shop-related alignment changes go one point at a time; kills typically go five points at a time, except in a few circumstances that aren't relevant here, and can't easily be made relevant.) Foocubi likewise only go one point at a time.

Posted: 4/28/2011 10:37 PM

ais523

Editor, Experienced Forum User, Published Author, Player (44)

Joined: 7/11/2010
Posts: 1034

eternaljwh wrote:

The thought occurred to me that, if you're dedicated to only having seeds possible normally (after the current version was released), you could still use the entire seedspace thanks to the eventual overflow.

I have no idea what happens to DOS if it overflows past year 2038. That's unlikely to be a particularly sensible mode of operation, though... (And we found a pretty good seed anyway, and are unlikely to redo the initial seed now.) Meanwhile, dwangoAC and I have been making some progress on the run. We've left Gehennom and wrote a bunch of spellbooks back at our low-RNG-jitter stash on level 16, and then went up to the Oracle level, to generate the last two spellbooks we needed (both of which are still currently inside statues that we're carrying around with us, as we don't need the corresponding spells yet, and we need a source of rocks for a glitch later on), and did an infinite carry capacity glitch (by repeatedly stoning, stone-to-fleshing, and taming a water nymph obtained from one of the Oracle fountains; details are on the game resources page). The main issue at the moment is that a side-effect of the glitch left us at almost -300 alignment, which clearly needs to be fixed (in the long term because of the Quest, in the short term because it prevents us using artifacts). As buying our way to a god's favour doesn't work within the first 5000 times, we'll have to go kill a bunch of monsters instead; I'll probably do it at the Castle, so people can have the excitement of watching us clear a really crowded level, unless anyone has a better idea. (And I'll abuse the carry cap glitch at the same time; we only really need it to carry cursed potions of gain level later, but that's no reason not to abuse it for other things too.)

Posts for ais523