Posts for ais523

Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
These sorts of points were why there were a row. I think everyone would be happy if you managed to beat the improvement, though. It's the same-route-same-time that lead to all the arguments. (I guess you could use different camera angles, too, within reason.)
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
There was a huge row when someone recreated another person's Pokémon Yellow TAS from an encode and submitted it. (I think it was nonetheless eventually published, though.)
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
Just encode the first hour, then have keyframes spaced several hours apart and encode a minute or so after each, with the rest of the video as whatever compresses best. Sure, it won't be a complete encode of the whole run, but nobody will ever find out.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
As this has been unrejected, I'd like to explain the category of this run. In Myst, the aim of the game is to complete and use linking books. There are several such books in the game, but three are particularly tied to the plot: Sirrus's, Achenar's, and Atrus's. Sirrus's and Achenar's books start missing five pages each, whereas Atrus's is missing only one. At the start of the game, the pages in Sirrus's and Achenar's books are hard to access (you get one of each in each of the four main levels), whereas the missing page for Atrus's book is in the hub world (and thus easy to access), but hidden behind a specific sequence of actions (that is intended to be determined via clues in two of the main levels, rather than known at the start of the game). These actions are always the same, and can be performed regardless of whether the character knows what they are or not. As such, this TAS skips all the levels of the game but the hub world, and starts off by just getting the missing page in Atrus's book. The game also requires you to find the linking books before you can add pages to them. Sirrus's and Achenar's books are trivial to find, but Atrus's (and one of the pages for Sirrus's and for Achenar's books) are hidden behind a combination lock (that's the lock that the TAS takes so long entering). When four of the five pages are added to Sirrus's or Achenar's books, they tell you how to determine the combination. However, the combination is also always the same, and can be entered even if the character doesn't know what it is, so the TAS, after obtaining Atrus's page, just skips directly to obtaining Atrus's book. Finally, you get the best ending for completing and using Atrus's book, and bad endings for merely completing Sirrus's or Achenar's books, or using Atrus's book before it's complete. (The bad endings aren't considered wins by most Myst players, but rather nonstandard game overs.) As such, the TAS only needs to fetch Atrus's book and the missing page for it, and it completes the game with the best ending. A 100% run would retrieve or locate all 11 missing pages, but would only use 9 of them (4 for Sirrus's book, 4 for Achenar's book, and 1 for Atrus's book that's used in a cutscene after losing control). (The last two pages cannot be used because that would lead to a bad ending.) This submitted run is any%, best ending (in fact, it's a low% for the best ending, with a somewhat major sequence break due to the use of codes that have not been revealed in-game). An any%, bad ending run would be somewhat faster if it aimed for Atrus's bad ending (as it would be the same without obtaining Atrus's page); it would be considerably slower if it aimed for Achenar's or Sirrus's bad endings, as that would require completing most of the game to gather the pages. A 100% would be longer still, due to needing to get both of those sets of pages; it would also be quite repetitive, because doing so requires moderately large amounts of backtracking / repeating areas. The most entertaining category for the game would definitely involve fetching four pages for at least one of Achenar's and Sirrus's books, which is meant to be required in order to locate Atrus's. You could argue for an "unspoiled" category, where the character does not use information that they haven't learned ingame, to enforce this, but that's a bit of a stretch and hard to judge objectively. Merely doing everything with Atrus's book, though, causes the TAS to not actually visit any of the levels, as everything required is in the hub world.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
I don't really like the idea of "vault" as being "below" regular movies. In some cases, it will be. In other cases, a run aiming to be a pure speedrun will be just as entertaining as entertainment goals in other runs. However, this can be fixed with a pretty small change, IMO. The idea is that runs can be accepted due to being entertaining, or due to being fast; and some runs will (obviously) count as both. Runs which are fast but not entertaining go in the "vault" tier, and runs which are both fast and entertaining, or merely entertaining, go in the "regular" tier, just with this suggestion; but the regular tier makes a distinction between movies that are records (any% or 100%), and movies that aren't aiming merely for the fastest possible completion in any% and 100%. That way we can have a list of entertaining movies to look through, but if someone's looking just for records for a game, they can look through the list of any% or 100% records, regardless of whether they're in the vault, or in the regular tier. Summary: just because a run's entertaining, shouldn't disqualify it from being a record.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
Nuzlocke seems a little pointless for a TAS. Its main effects are forcing you to play more cautiously (not a problem for a TAS, which can force enemies to miss if necessary), and limiting which Pokémon you can use (which basically just means route planning is less interesting). You're not going to actually use more than one of the mons you catch, except as HM slaves, because putting all the experience onto one Pokémon is the most efficient way to do it in first generation.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
Just as entertaining as the old run (it looks much like I remembered), except now it's faster. And for people who don't know, that's pretty entertaining.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
I'm OK to expand the site to contain tool-assisted speedrun records that are separate from the movies that aim for entertainment, and this seems like a reasonable implementation. Perhaps it might be reasonable for runs to be in multiple tiers (say the "moon" and "vault" tiers mentioned), say if you're making a run that's as optimised as possible but also trying to make it entertaining. (And then it could be obsoleted in either of the categories, by entertainment and speed respectively.) There will be a lot of overlap, primarily because going fast tends to be intrinsically entertaining.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
adelikat wrote:
If you give a skilled human about 1-4 frames of variance for reflexes, then a good human should reasonably do this run within about .2 seconds from the TAS time.
From my experiments in hardware TASing (i.e. inputting an input file into a controller by pressing buttons on it): if you have a metronome or countdown or similar device giving you cues on exactly when to press the button, and several seconds to prepare before you have to press it (no 30Hz mashing), a skilled human can hit an individual frame perhaps 1 time in 3, after spending several hours practising.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
OK, turska, I finally watched the run because you spent so much time trying to persuade me to, and it really was just like I expected it would be, with no interest beyond the submission text. In other words: this run isn't worth watching, there's no interest in an encode beyond the submission itself.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
The timer thing also worries me. Does this game have a global frame rule that governs the very end of the game? If so, and it's obviously impossible to beat it, the run is obviously unimprovable barring starting/ending/lag, and that's a problem in its own right.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
I haven't voted No, but only because this game/run seems boring enough that I don't even want to watch it, and it's unfair to vote a game down without watching it.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
GoddessMaria15 wrote:
One alternate way is to study the RNG oneself. >_> In any case, Challenge Mode would be preferred. However, it must be unlocked first and can only be unlocked on Black 2 after beating the game for the first time. Should a run like that be made, a verified SRAM is needed.
From what I've heard, starting a new game undoes the unlock, and the only way to start a game in Challenge Mode is to communicate with a Black 2 cartridge to transfer the Challenge Mode key just after starting. This might be very awkward from the point of view of trying to record everything.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
Am I going to have to do a fastest crash in NetHack, now? Actually, I can tell you how to do it yourself. Start a Healer, and type d4294967296$d4294967296$. Unless you somehow manage to die on turn 1, you should get a crash immediately on turn 2, in very short realtime. This may be improvable, especially in gametime. (You can get a buffer overflow turn 1, but on DOS, it won't crash the game immediately, and it requires rather more typing than the turn 2 crash does. I don't believe it's possible to get a crash on turn 0 (i.e. before gaining control of the character) without a custom configuration file, but who knows, I might be wrong.)
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
That's actually a decent point. If it hexes in, we may as well change it for the added entertainment value. If it doesn't, it's not worth redoing everything since then for something so minor.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
For the benefit of anyone judging this: I haven't voted, but I enjoyed watching the movie. Whether it's published or not should probably depend on how improvable it is, which I don't know.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
feos wrote:
[22:28] <Ilari> Heh, if/when Nethack is submitted, the submission comments are going to be so long that those are split into couple wiki pages...
We already discussed it with Nach. It seems that it's possible that they'll be accepted by the site's parser because they're so long that they overflow the page length variable past the negatives and back into the positives again…
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
I guess I have to mention #3080: ais523's DOS NetHack "fastest death" in 00:01.15, where the input length is shorter than the game+OS's loading time. (However, the input is done right at the start and stored in the BIOS keyboard buffer, with a result that the input end is before the game actually loads.) This was an April Fool's submission, though, so probably shouldn't be taken seriously.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
I'd agree that if you don't care about learning low-level languages at all, there's no point in learning asm or C. Just if you do want to learn them, do it in the right order :P And about the "is C a good language" argument, my point of view is that it's clearly improvable but there have been no serious attempts to improve on it until recently. (Mozilla are doing interesting work with Rust, but it's still in early stages.) As such, it's still very widely used when a portable low-level language is wanted. Re adelikat: I'd say that if you're learning Java already, learning C# too won't gain you much. The two languages are very similar, differing mostly in that C# has some extra complexity and more modern libraries. (And C# has much of the same problems as Java as a result.) It's best to have more variety in your first few languages. (Especially as if you know Java, it won't be hard to pick up C# if you need it for a job.)
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
(Good variants of) asm is actually a pretty easy language. It's tedious and slow to write anything but very small programs in it, which is why it isn't widely used. But it isn't hard to learn, and it's a good way to get introduced to concepts like pointers and call stacks (and perhaps stack frames, too). I'd argue against JavaScript (basically because the library support sucks; the language itself has a few flaws but is interesting), and against PHP (badly designed). This thread is going to turn into a language flamewar, isn't it?
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
I happen to teach undergraduate programmers, so I have some idea about this. I'd start off by recommending against C for a first or second language; it's hard to understand and easy to get into bad habits about. It's definitely worth learning, but I'd suggest learning assembly language first (your choice of platform; the 6502 may be an interesting one if you're into TASing, because it's what the NES, and several other consoles of that era, use). Python, although I personally dislike it, is a reasonable choice; it's similar enough to Java that you can transfer your skills, but different enough that it'll give you a better-rounded view on things. Don't attempt to learn C++ without professional help. (And even then, you need good professional help.) Few things are more scary than someone who's self-taught in C++ attempting to program in it. (I should know; that was me once. I know better nowadays.) If all you care about is employment, Java is the language to learn. It's awful and has several issues, but it's the most in demand right now, and will probably be highly in demand for the next couple of decades (for much the same reason that there's still demand for COBOL programmers). You might also want to look at some language which is very different from the other languages you've looked at so far. Typical off-the-wall choices include Prolog, Haskell, and Lisp. You're unlikely to get a job in these, but having them on your resumé will cause people to give you a second look (assuming you can back it up). Finally, I'd suggest learning by trying to write programs. Aim slightly beyond what you think you can manage, and learn on the way there. (Aiming a long way beyond, you'll just fail and get discouraged; aiming within what you can manage, you'll get bored and won't learn anything.)
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
You can afford to miss out on 90 notes altogether, right? So RBA already has enough to skip the maximum number of jiggies. Of course, there's the question of whether it's worth skipping the puzzles for every level you can, or just only the more expensive ones.
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
What does RBA stand for, in this context?
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
Even if the 810 note isn't skippable, it may be worth checking if levels can usefully be done out of order via note door skips. (My guess is no, but that doesn't mean it isn't worth checking.) This might be especially useful if it turns out that level unlocks in the middle of the game are unavailable because you want to skip a lot of early jiggies in favour of a lot of late ones (but that also seems a bit surprising).
Editor, Experienced Forum User, Published Author, Player (44)
Joined: 7/11/2010
Posts: 1022
FWIW, I fully support Toothache's decision here. I'm one of those people who's interested in how low a game can be taken and what it takes to do that; I don't really like the concept of TASes that are merely "good enough". ("Best currently known but probably not perfect" is a different matter; loads of TASes are like that, and there's not much you can do about it.)