Slashdot Mirror
'Perfect' Zelda NES Speed Record Beaten
An anonymous reader writes "The last verified human Legend of Zelda (NES) speed record was 34 minutes. A few months ago, a re-recording emulator was used to make a 'perfect' video which was 31 and a half minutes. A team worked to optimize the path, and using an emulator created a new video which is 26:56, four and a half minutes faster. The video is 14% faster, and is the first Zelda run to be under a half hour. Furthermore, it achieved a sub-27 minute time, which was presumed impossible. Definitely worth checking out - you can grab the BitTorrent version of the AVI replay at Bisqwit's NES time-attack movie page." There's a thread on the NESvideos forum discussing the attempt, but can anyone succinctly explain the exact tricks the team used to speed up their time?
Thanks to Google, I've found other speed runs:
http://www.planetquake.com/sda/other/
blue ring? wuss. First of all if you get the "it's a secret to everybody" guys you can get the blue ring before you get anything else in the game. Secondly don't you think you would beat the game a lot faster if you didn't get useless items like the blue ring? When going for a speed record you need to get only what is absolutely necessary to get to the end. Leave all other items behind.
The GeekNights podcast is going strong. Listen!
you can download the fmv file, which basicly saves only what keys were pressed when. then, you can run the game on your own computer, and check the time. j just make sure you have the same rom they used
Before anyone else posts something about how they "faked" the god damned movie, please refer to this page on thier website: http://bisqwit.iki.fi/jutut/nesvideos/whyhow .html#why
They are very open about using various tricks of the emulators to achieve a quick run of the game. Even if you consider it "cheating" that is very much beside the point, because they clearly state that it is not so much a competition as much as for entertainment.
They also comment on the idiots on various forums around the web that bitch and moan about their assumptions that these videos are faked, who also don't read the FAQ. I would hereby like to welcome you all (who have bitched and moaned about just that) to that group. Thank you for your contribution to stupidity.
It IS faked. It says so right there in the text of the story, let alone the article. It WAS made using an emulator.
This practice is called a "Time Attack" and is completely different from speed completions, which are indeed done legitimately. Time Attacks are not world records and are never claimed as such - they're just cool, is all. If you understand Japanese it actually says clearly on the website of the guy who made the 11 minute SMB3 video that he faked it. It was an unfortunate misunderstanding that most people accepted it as legit.
That's not to say that Time Attacks are easy to do. It takes a lot of time and effort to put one together.
qntm.org
1) Using (abusing) the Select-Continue to go back to the start of a dungeon (after grabbing an item, before fighting the boss) and to get back to the starting overworld area right quick.
2) Going through the dungeons in a different order, which seems to keep the new guys almost a full dungeon ahead through most of it.
3) Skipping out on some things - new one didn't get the master key from dungeon 8, and is using the new dungeon order to ensure they can get enough keys. They're also skipping out on the extra hearts from bosses for a few extra seconds - as it looks so far (half way in) it doesn't look like they're going for the level 3 sword, just level 2.
4) Bombs. Lots of bombs. MFried used bombs only for blowing doors and some enemies. Sleepz is using them in regular combat to hit 4-6 enemies in one blow. Sometimes using multiple bombs for the silver knights and such.
5) Flute abuse - sleepz is hitting the flute twice in a row, before the first whirlwind gets to them. Looks like this is still getting them to the same spot. Nice trick though :)
All in all, I recommend people watch both at once :P It's entertaining to see them identical going after the sword, then one select-continuing, peeling left, and the other walk out, and go right. Neither have been in the same screen at the same time since :)
Nope, the fmv file prevents this, it only records the controller action so you simply download it and run it on your own computer and check the time.
It would reduce the time shown in the avi, but not the time shown when you check fmv file.
This has actually been discussed here before.
The only "official" world record organization for video gaming is Twin Galaxies. Twin Galaxies definitely wouldn't accept any of the time attacks on Bisqwit's site (including Zelda) for a number of reasons.
Does that answer your question?
Rob
Whoops. I put this in the wrong place. These videos were NOT sped up. They should run at the exact same speed. No frames were removed, no frames were edited, this is actual gameplay that you could get if you pressed the buttons at the exact same time. The only reason that this is 14% faster is because a better route was thought of. The only "cheating" that was used was slowing down the gameplay(the actual movie runs at normal speed though), and rerecording.
Using an emulator and save states to redo every single move over and over again and compiling them into a single movie, imo, isn't talent. ANYONE can get a fast time with emulators, because you can continuously correct your mistakes as you make them, essentially erasing your errors. All that's needed is time. True speed runs are runs that are done WITHOUT emulation, WITHOUT taking out mistakes, doing everything as quickly as you can do without being able to bail yourself out by rewinding the game back with a save state. THAT takes talent. It takes dedication, time, patience, and a true mastery of a game.
Using an emulator to compile moves into a "perfect" run can be amusing, but, by no means, should it be praised, or considered talent, nor should it be recognized or publicly released, because it downplays the mastery of the true speed runs.
As I understand it, playing the flute increments or decrements a counter depending upon which direction Link is facing on the screen. Then that counter checks some value (like which dungeon you finished last or some such), adds to it, skipping over any dungeons you haven't finished already, and that's your destination. So playing the flute multiple times means you get to basically decide where you want to go (within the set of dungeons you've finished).
we ('we' being the team of people who worked on the first run) are working on a 2nd quest run, and have mapped out several different dungeon paths, trying to find the optimal route.
i would expect sleepz or someone else will attempt a run within a week or two, although it might be a month until a run is good enough to publish.
as far as faked: it's not 'faked', it's 'save stated', which means it's very much legitimate... with dozens (or hundreds) of attempts for every few seconds of footage, all streamed together in a seamless movie. you could do the exact same thing on your own, possibly even without saving at all.
we debated about the use of up+a, but since it was documented in the manual, we felt it was 'ok' to use.
it's like the theoretical 5:08 (which is now 5:05) for super mario bros 1.. the human best still stands at 5:17.. which shows just how close to perfection the console run is.
likewise, we have a fast zelda run, and we've shown that the ideal time is about this close... which means the console times of 34 minutes and such actually leave room for a fair amount of 'improvement'.