It's hard to believe, but even to this day, speedrunners are still competing over the crown for the fastest Super Mario Bros. run yet. The record has been going down, millisecond by millisecond, for years now, and a recent twitch streamer just set it a bit lower.
The record was previously set at 4:57.194 by Kosmicd12 on 3rd October, but it just took two days for Darbian to retake the crown with a cool time of 4:56.878. Of particular note is that this is the first time that a human has managed to get the time below 4:57; Darbian did a run that was believed to be only achievable by usage of a bot. Here it is, though do be warned of some language in the live comments:
What do you think? When do you think we'll finally hit the lowest possible run? What's your fastest time? Drop us a comment in the section below.
[source youtube.com]
Comments 35
Cue Kosmic revealing his 4:55 wife to the public in... 3... 2... 1...
even knowing he will break the record some of those close calls are nerve racking.
Oh man, that F bomb in the thumbnail of the video. That's kinda funny.
"Darbian did a run that was believed to be only achievable by usage of a bot"
Not really. Most TAS videos you see are done by someone who plays the game at low speeds on emulators which allow the visualization of RAM and manipulate the RNG, with save states/restore points, basically allowing to play a 'perfect' run that was planned beforehand.
@Gridatttack But this wasn't a TAS...
He's a X-men, it's sure!!!!
Another sub-5 minute Mario run, with TAS techniques such as walljumping, that cannot be performed without subframe precision. The level of B.S. is high. Also took two wrong pipes to achieve his lofty goal...
Wasn't he using a TAS? Although incredibly talented, I can see a lot of times when he made contact with enemies and didn't die. How is that possible?
I'm a bit confused, his clock says 4:57:02?
Wow !
But playing game in speedrun may destroy the fun aspect. I mean, we play the game too fast and then hello boredom (Because have finished the game quickly).
@Anti-Matter speed run
@Anti-Matter When you play a game to death, it becomes harder to play the game without doing something different, i.e. Ocarina of Time. I played it so many times, the cutscenes are mind-numbingly slow, and puzzles are now obnoxious obstacles. Because of this, I spice things up by playing the game differently, like a 3-heart run. Sometimes I even use glitches to skip entire dungeons, not to beat it as fast as I can, but to try doing things differently from what was intended.
People speedrun games because they love these games enough to keep playing them, no matter how well they know the game. And getting a world record, or even a personal best, gives a feeling of satisfaction, because you keep getting better at it, and continue to surpass yourself, or even the last person to hold a WR.
Tool-assisted. I understand that the enemy boundaries aren't 1:1, but there were several times when he was more than over the boundaries with the piranha plants. The frame rate drops and changes in jumping trajectory seal the deal.
@allav866 👍🏻
@Moroboshi876 Hitbox detection was wonky in this game. I've done runs of 4-1 myself where I'd basically jump through the piranha plants.
@mitchvogel Really? Then I must try it, but I've been avoiding enemies (or trying to kill them) my entire life :S
Yes there were some very close calls, but tool assisted? I don't think so. What evidence is there? He recorded himself playing in real time, and he didn't save or load any states. There were a few video stutters but to me that's the fault of the recording tool.
http://i.imgur.com/FEOJktH.jpg
I took a screen shot from his video. He's dead there. The goomba hit mario's head. Bogus run or tool assisted / cheated.
@the8thark SMB's physics are weird. As long as you're descending, enemies can't hurt you, in fact most of the time, if you touch an enemy from the side while descending, you'll end up stomping that enemy. That's why he jumps into the block; to be bounced back down and avoid damage.
If it was tool assisted then it doesn't count. If it wasn't, that's really impressive.
@Emperor-Palpsy
Spoiler alert, run was done in real time, not tool assisted.
I watched him live in person at AGDQ 2016, winning the Super Mario Bros race.
@Shigmin
4:57.02 is a timer that had to be stopped manually. It was there for stream monsters to get an idea whether or not he was ahead or behind his Personal Best.
4:56.878 as the title says is basically after watching the run through a program like Virtualdub and count the number of frames, showing a more accurate time.
@Shigmin Defeats Bowser at 4:56.878, gets to the Princess at 4:57:02.
@Greenalink Figured it might be something like that, thanks!
Oh my gosh. This is unpossible!
man, the level of suspicion and disbelief in this short comment section is astounding... ô.Ô
@the8thark
In the original NES SMB1, when you are moving downwards in the air, you are invincible to stompable enemies.
http://media.tasvideos.org/smbtricks/smb_hatstomp.gif
http://media.tasvideos.org/smbtricks/koopajump.gif
@Yoshi I know this wasn't a TAS.
I'm commenting that not all TAS runs are necessarily made by bots
Do they use the Japanese version of this game like they do for Zelda etc?
@Anti-Matter What? You only speedrun after you've already beaten the game normally, so it adds more replay value.
Cool story.... BRO!
Call Miyamoto, my favourite comment.
Congratulations for the record!
@AlexSora89
LOL XD
That the real reason why Mario do speedrun. ( OMG, I have to go to restroom quickly !) XD
@Anti-Matter
"Yet another gaming mystery solved thanks to Brawl in the Family."
The hit box detection glitches when the screen is constantly scrolling at high speed otherwise when Mario is running at full speed constant. maybe the hit box is smaller while screen is scrolling a full speed? - making it easier to clip through edges of enemies?
He could have shaved off some more time during the platform jumps to the flag pole by floating to the bottom of the flag pole.
Show Comments
Leave A Comment
Hold on there, you need to login to post a comment...