Super Mario Bros.

Alko

2014-02-06 01:48:58 pm

Caution: This user contains Kana ^_^

I love PBs that end with a hammer jump. Congrats, mate :3

andrewg

Edit history:

andrewg: 2014-02-27 10:58:22 pm

andrewg: 2014-02-27 10:51:46 pm

2014-02-27 10:46:19 pm

Hi! I'm andrewg!

Haha, i don't know why people do this kind of thing... it's insulting:

EDIT: Well, I guess he uses savestates and thinks that's how you do a speedrun or what a speedrun is. Meh, just wish people would label things properly:

Also:

doicm

2014-02-28 01:28:23 pm

Quote from Daniel's Game Vault:

Aren't speedruns all about save-states and editors ? It's impossible to do such moves without frame-by-frame editing I'd say.

This made me chuckle.

zallard1

2014-03-01 03:30:23 am

B+Left, Left, Up+B, ★

Looks like this is cristian3131's youtube account! Couldn't tell at first because youtube lets you choose what name your account displays. I guess he wasn't content with badly editing up a 4:57 any%, so he went for the badly spliced warpless record Tongue

Blubbler

2014-03-01 04:29:39 am

Quote:

EDIT: Well, I guess he uses savestates and thinks that's how you do a speedrun or what a speedrun is. Meh, just wish people would label things properly:

I don't believe that. He tried to pass it off as legitimate and now pretends he didn't know.

For those who don't know: some of Hotarubi's SMB runs have been found
warpless in 19:24 (old WR, but the video was lost)
All-Stars in 37:29
Lost Levels SNES in 8:19

JackintheBox333

2014-03-01 07:58:52 am

Professional Shaq Fu Speedrunner

Oh wow, I though that really weird timer looked familiar. I guess his editing has gotten a little better since that poorly spliced together "4:57". Of course its still really badly spliced still on his new "video".

andrewg

2014-03-02 01:02:40 am

Apologies to Cak and Mitch in advance...told him I USED to have those records, not that I do now. :/ Anyway, beside that goof, this was pretty cool:

http://www.bostonglobe.com/metro/2014/03/02/quincy-gamer-quest-for-super-mario-perfection/lFGe6RONwJrWFqNMwfDjtJ/story.html

sdkess

2014-03-03 10:22:04 am

Jungle Rat

Don't feel too bad about the goof. In the old "All Thumbs" article about you, Billy wrote that I was the record holder on a bunch of games like Metroid, Zelda 2, SMB2, SMW, etc that I was the former record holder of. I thought that I made that clear when I spoke to him on the phone, but this kind of thing is actually very common in journalism. Nearly every news article that has been written about me has contained some fairly large errors. It makes me wonder how many non-gaming news articles are riddled with errors... Probably a bunch.

presjpolk

2014-03-03 10:49:11 am

HELLO!

Michael Crichton pointed out in 2002 that the whole paper is always full of errors, but yet we always read the parts of the paper we don't know enough about to see the errors in. That we then forget the paper is pro bably inaccurate, he called the Gell-Mann Amnesia Effect.

Named such because it was scientist Murray Gell-Mann who was noticing the paper is usually wrong about science...

Alko

2014-03-03 04:35:17 pm

Caution: This user contains Kana ^_^

I keep reading the corrections my weekly paper publishes. It's strange what kind of errors happen. Placing events in the wrong countries, for example. At least I can usually forgive them for being wrong about science; science being the fascinatingly complex thing that it is. The poor blokes who have to explain the Nobel prize in chemistry to their readers every year … xD

Other than that very nice article ^^

cak

2014-03-03 05:20:54 pm

twitch.tv/ccaakk

Congrats Andrew! And no worries...this sort of typo probably happens all the time.

sdkess

2014-03-04 12:44:55 pm

Jungle Rat

That's interesting stuff about errors and the Gell-Mann Amnesia Effect. I'm always pointing out video game related errors in TV shows and movies which sometimes annoys the people that I watch with, haha.

It is a great article. I really liked the old one Billy wrote for The Boston Globe several years ago about Andrew (All Thumbs), and also the Bizarro World one about Billy's wife setting a Tetris record at Funspot. Any articles about gamers like us are good, but it does sometimes bother me that facts often get messed up.

andrewg

2014-03-25 08:07:37 pm

Well, at least it's an improvement! It's only a matter of time now. I'm even more determined now! thumbsup

Theworm826

2014-03-25 08:28:19 pm

If we're looking up old hotarubi runs, here's a 36:50 Lost Levels SNES D-4 Warpless

http://firestorage.jp/download/8b6612626c0a2d5fb4df0c22296b651fa4ccf8aa

Probably one of the best speedruns I've watched.

Blubbler

2014-03-25 10:12:01 pm

In light of andrewg's new PB I think I may go into more detail about precise timing, because there is some confusion about it:
The idea is that RNG in this game is determined by what frame you enter a screen on. Now you can enter the last room on a number of different frames on an emulator, look at the pattern of the hammer bro, the two fireballs and Bowser's movement and hammers, compare that with the video and hit the axe on the same position. If you know what framerules the run was started on (on the title screen) and that the runner held B and right for the duration of the last room, you can get the frame-precise timing of the run.
Now you can ask if it's useful and I think honestly it isn't really, but it's interesting that it works and if two times ever got that close or if someone got that close to the 4:57 barrier that the inaccuracy of the capturing method matters, it could be resolved by this. I originally did this for Kosmic's run where the video length was known to be off by a significant amount.

Now lets look at his new PB:
Andrew records with a DVD recorder and the biggest possible time difference between "recorded time" and "actual time" should be one NES frame (1/60.098814 s).
4:58.08 is 17867 frames at 59.94 fps (NTSC framerate)
The framerate of the NES is 60.098814 and within that 1 NES frame margin are
4:58.07 (17914 frames)
4:58.09 (17915 frames)
(Notice that 4:58.08 is an impossible time by counting NES frames)

Here are a few gifs of entering the last room on different frames:
17828

17829

17830

And here's the same part in andrew's run:

You can see how well it matches one of the gifs. Here is a closeup of the position on the switch:

Now what confuses me is that I did this with 3 of andrew's run and it was slower than the "recorded time" every time, that's quite a coincidence - unless it isn't a coincidence and there's a reason for it that I'm missing?

In light of andrewg's new PB I think I may go into more detail about precise timing, because there is some confusion about it:
The idea is that RNG in this game is determined by what frame you enter a screen on. Now you can enter the last room on a number of different frames on an emulator, look at the pattern of the hammer bro, the two fireballs and Bowser's movement and hammers, compare that with the video and hit the axe on the same position. If you know what framerules the run was started on (on the title screen) and that the runner held B and right for the duration of the last room, you can get the frame-precise timing of the run.
Now you can ask if it's useful and I think honestly it isn't really, but it's interesting that it works and if two times ever got that close or if someone got that close to the 4:57 barrier that the inaccuracy of the capturing method matters, it could be resolved by this. I originally did this for Kosmic's run where the video length was known to be off by a significant amount.

Now lets look at his new PB:
Andrew records with a DVD recorder and the biggest possible time difference between "recorded time" and "actual time" should be one NES frame (1/60.098814 s).
4:58.08 is 17867 frames at 59.94 fps (NTSC framerate)
The framerate of the NES is 60.098814 and within that 1 NES frame margin are
4:58.07 (17914 frames)
4:58.09 (17915 frames)
(Notice that 4:58.08 is an impossible time by counting NES frames)

Here are a few gifs of entering the last room on different frames:
17828
[IMG]http://i.imgur.com/lCWhrlp.gif[/IMG]
17829
[IMG]http://i.imgur.com/lduZNaC.gif[/IMG]
17830
[IMG]http://i.imgur.com/imC3iU0.gif[/IMG]
And here's the same part in andrew's run:
[IMG]http://i.imgur.com/sFOaUPr.gif[/IMG]
You can see how well it matches one of the gifs. Here is a closeup of the position on the switch:
[IMG]http://i.imgur.com/0Nqug72.png[/IMG]
Now what confuses me is that I did this with 3 of andrew's run and it was slower than the "recorded time" every time, that's quite a coincidence - unless it isn't a coincidence and there's a reason for it that I'm missing?

andrewg

2014-03-26 07:48:12 am

Hi! I'm andrewg!

Is that 4:58.09 then? I have no idea why its always faster than what it actually is (maybe how the dvd recorder rounds the last frame?)

blubbler

2014-03-26 07:54:59 am

Or maybe you time it differently from me?
Let's say the timer in 1-1 appears on frame 400 and Mario and Bowser get invisible on hitting the axe on frame 1200, the length of the run would be 800 frames, right?
Or maybe you can just tell me what you get, when you time the TAS to see if it matches?

andrewg

2014-03-26 12:13:23 pm

Hi! I'm andrewg!

i time the same way, so its probably sonething concerning tge dvd r

andrewg

2014-03-26 12:14:53 pm

Hi! I'm andrewg!

frame before visible movement to invisible

blubbler

2014-03-26 03:27:08 pm

Quote from andrewg:

frame before visible movement to invisible

So you stop timing one frame sooner?
Can you load the TAS into FCEUX and give me the frame numbers?
http://tasvideos.org/movies/fm2/happylee-supermariobros%2Cwarped.zip
I thought it was:
start 197
end 17868
If we just had different end times, that'd explain sth.

andrewg

2014-03-26 03:44:36 pm

I time from when "400" is on the clock to when bowser turns invisible upon hitting the axe.

Magnum66

2014-03-27 11:46:04 am

http://twitch.tv/Exisidis2

This is like the 10th time I've seen you say "I give up" then just went right back to it lol.

Patashu

2014-03-28 05:42:59 pm

Perhaps the difference in timing is in how you define 'timing a run'.

Let's say that at frame 0 '400' appears on the clock and at frame 1 bowser turns invisible upon hitting the axe (you somehow achieved the theoretical only warp speed run). Would you time this run as 33ms long, 16ms long or 0ms long?

cak

2014-03-28 10:53:56 pm

twitch.tv/ccaakk

Input control to loss of input control, or 16 ms.

It's clear that control begins upon appearance of 400, and reasonably clear that control is lost upon Mario disappearance.

andrewg

Edit history:

andrewg: 2014-03-29 11:03:49 am

2014-03-29 11:02:47 am

Hi! I'm andrewg!

We are all timing it the same way, there's just gotta be something odd going on in either how the dvd recorder rounds the final time or maybe how the encoding does... that, or it's just pure coincidence (1/8th chance with what blubbler observed).