(Link to video)
Submission Text Full Submission Page
VS Super Mario Bros is a modified version of the super popular NES game, Super Mario Bros. This modded version is for the arcade machines, and so it is slightly harder, but easier than Lost Levels. This movie completes it in a hair under 10 minutes and 15 seconds. This is my third attempt of a TAS, the first two were slow WIP's that didn't deserve the name TAS. It is near perfect, but I'd personally like to see what I thought was impossible by TASing, be done.

Game objectives

  • Emulator used: FCEUX 2.2.1 (2.2.2 and 2.2.3 also work)
  • Uses Hardest Difficulty
  • Aims for speed rather than entertainment
  • Glitch abuse
  • Uses collision errors
  • Takes shortest route possible

Comments

VS SMB is a slightly longer and harder version. There are no warps to worlds 7 and 8, pipe warp maps may have changes, and in my ROM, the game has a faster timer. Some surprising results were found. This TAS has been in the works for almost a year. With that there are a few things to note.
1. Framerules are very screwy. They aren't always every 21 frames
2. Things that are usually faster, aren't always that way in this game.
3. Pipe warp in 4-2 has moved one pipe to the right, thus making the Wrong Warp near impossible.
4. Furthest warp in the game is 6-1.
5. Flagpole skips aren't always possible.
6. Physics engine is ever so slightly different. Not by much, but it can ruin your day.
With all that in mind, let's get started on the stage commentary. If there isn't special to note, and is just what this community expects, I won't say anything.

Stage by stage comments

1-1

We all know and love it. Little to nothing has changed in the level. Slightly different flagpole skip than the preferred one, but it gets the job done. Time: 357

1-2

Not much to say here, except two things: 1. There are more goombas before the set of three pipes. There are 9 goombas here. This many goombas causes one of the piranha plants to not spawn (notably the second one). 2. Slowdown to clip into pipe might be slow, but it was the fastest I could get.
Time: 324 Onto 4-1

4-1

It's all what we expect. Again, it's that different flagpole skip. This is the one used by speedrunners, and it seems to work every time. The TAS one I can not get to work. But, is it faster?
Time: 265
P.S No backwards acceleration here, because it would cause a 266, and a sad TAS community.(6 fireworks)

4-2

Notes #3 and #4 makes all of us sad. I don't see any visible improvements, but maybe that wall clip is still useful.
Time: 340 Onto 6-1

6-1, 6-2, and 6-3

Grouping these together because they have their notes similar.
Not much to say. A few things to say though:
I can't get the flag skip in 6-1. I feel that this can be done, but for right now, I have to say the the phrase hated by this community: I think it's impossible.
Building on top of that, I also think that the 6-2 flag skip is impossible. I do know for 100% certainty that it is impossible to hit the top of the flag. I'll say that the biggest improvement is in 6-2.
6-3's flag skip looks possible, but I am probably not a sophisticated enough TASer to do it. I'll be surprised if it can't be done. (By the way, does this ring a bell? Does this look like 5-2 in Lost Levels, or am I crazy?)
6-1 Time: 280 6-2 Time: 319 6-3 Time: 231 (1 firework, who cares?)

6-4

Unlike those three levels, I am happy with this level. Everything is so polished and smooth, and it looks great.
Time: 223

7-1

Not much to say, but I do have a challenge. If successfully done, it is an improvement.
Challenge: Get the flagpole skip in this level without getting fireworks or losing two timeunits. There are four ways to skip the flagpole: 1: TAS flag skip 2: Speedrunner flag skip 3: Bullet Bill glitch (if you can get it) 4: Despawn the flag. (somehow. I did that once)
Time: 282

7-2

What can I say that won't bore these people? Yeah, this level is pretty much what you'd expect. I have done the L+R to go faster. Oh, here's something:
Flagpole skip is very possible, I'm not sure what it would do the fish in the next level.
Time: 248 (Mario touched the flagpole at 249, but the timer drained for two frames.)

7-3

Not much here, BUT: Flagpole skip is almost certainly slower here. I know, it sounds odd, but that's what I got.
Time: 201 (1 firework, who cares?)

7-4

This is probably my best level in this TAS. The maze route is different, but slower. There are probably a few frames to save, but they would be for someone who would grind five years over a single frame.
Time: 320

8-1, and 8-2

The game of: Is it Possible: Flagpole Edition. I think it's possible, but I seem to be unable to pull it off in 8-1.
8-1 Time: 209
On 8-2 I got a very, very surprising, but possibly dissapointing result. Both flagpole skip and Bullet Bill Glitch are slower by at least one framerule. You could imagine the rage boiling in me at that. So, I just did a standard finish.
8-2 Time: 268

8-3

All the standard stuff we know. I got the flagpole skip first try, which was nice, but really nothing here to say.
8-3 Time: 219

8-4

You know the drill. We all do. A few improvements remain: 1. The turnaround to the pipe used for the Wrong Warp in 8-4 2. Pipe exiting the water section (1-3 frames) 3. Bowser boss. With the RNG I got, it was slow.
8-4 Time: 280

Other comments

I've had this TAS done for four months, but I felt at the time that there was something missing. So, I went back to it in late May, and experimental evidence showed that the only things missing were some swag.

This TAS aims for speed and not really entertainment.

I give permission for anyone to use my TAS as a "base" TAS.
Please use frame 33521 of the TAS as a screenshot.

Masterjun: Judging.
Masterjun: Replaced file with one that starts from power-on instead of a savestate near the beginning.
KnucklesMaster368: I have made the final decision. ROM is faulty, TAS is suboptimal, and I'm at 7-2 with the best ROM we have, reworking it. Decision: cancelled

TASVideoAgent
They/Them
Moderator
Joined: 8/3/2004
Posts: 14899
Location: 127.0.0.1
BigBoct
He/Him
Editor, Former player
Joined: 8/9/2007
Posts: 1692
Location: Tiffin/Republic, OH
On the first ROM I found, md5sum B4E90B95A3E18F2BD79BBECF5009A28B, you desync at the beginning of 1-2. This might be a dealbreaker; I don't remember what the site rules are on using alternate dumps. Anyway, watching now.
Previous Name: boct1584
billybobfred
Other
Joined: 12/2/2005
Posts: 76
The Movie Rules page says "Use a good ROM dump if available." Going just by that, then, I'd say that if "VS Super Mario Bros (VS) [!].nes" exists, then the submission will have to be redone on that ROM. Tangentially, I'm kinda unclear on how alternate dumps become a thing in the first place.
I'm just a spectator. There was a time when I tried to participate, but I'm really lazy, so. she/her/hers
Player (12)
Joined: 6/17/2006
Posts: 501
Conjectures about the alternate dump issue: - At least one of them is a bad dump, but which one could not be verified back then. The author mentioned a faster timer in his version and imperfect frame rule, which could be signs of that. - There are different versions of the game with different ROMs. In particular, according to The International Arcade Museum, VS Super Mario Bros. has a dedicated arcade unit version and a conversion board version. I wonder if someone more familiar with this game, or the internals of Super Mario Bros. and The Lost Levels, could shed some light on the matter.
Joined: 5/1/2015
Posts: 96
Location: Florida
billybobfred wrote:
The Movie Rules page says "Use a good ROM dump if available." Going just by that, then, I'd say that if "VS Super Mario Bros (VS) [!].nes" exists, then the submission will have to be redone on that ROM. Tangentially, I'm kinda unclear on how alternate dumps become a thing in the first place.
I have not been able to find such ROM with an [!]. I've heard that other versions exist, but have been unable to find anything other than an a1 version.
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Joined: 5/1/2015
Posts: 96
Location: Florida
boct1584 wrote:
On the first ROM I found, md5sum B4E90B95A3E18F2BD79BBECF5009A28B, you desync at the beginning of 1-2. This might be a dealbreaker; I don't remember what the site rules are on using alternate dumps. Anyway, watching now.
Wait, a desync? I didn't get that. Though I can't get what the md5 sum of my rom is. Are you sure you have an [a1]?
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Joined: 5/1/2015
Posts: 96
Location: Florida
I'd just like to mention that this is my first serious TAS submitted here. On the topic of "VS Super Mario Bros (VS) [!].nes", I have found no such thing. Can someone tell me why alternate dumps are/could be bad? I'm checking to see if it works with "VS Super Mario Bros (VS).nes". I'll come back afterwards. EDIT: It does not. But, it has a faster timer, too, but not as fast. On the [a1] dump, my 1-1 time is a 357. On the normal dump, it is a 363. But it is as fast as a 370/369 on an original game. I have heard that the dipswitches do something about this. The desync is caused by the timer. At 1-1 it ends with a time of 363, which is causing 3 fireworks. This is from the thread on the game: Switches 1/2/3 control pricing: 000 (0) - 1 credit per coin 001 (1) - 1/2 credit per coin 010 (2) - 1/3 credits per coin 011 (3) - 2 credits per coin 100 (4) - 3 credits per coin 101 (5) - 4 credits per coin 110 (6) - 5 credits per coin 111 (7) - Free Play Switch 4 controls the number of lives you start with: 0 - 3 lives 1 - 2 lives Switches 5/6 control how often you get an extra life: 00 (0) - 100 coins 01 (1) - 150 coins 10 (2) - 200 coins 11 (3) - 250 coins Switch 7 controls the timer speed: 0 - Slow 1 - Fast Switch 8 controls how many lives you get when you continue: 0 - 4 lives 1 - 3 lives Could someone experiment with this? It doesn't work on the a1.
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Masterjun
He/Him
Site Developer, Skilled player (1971)
Joined: 10/12/2010
Posts: 1179
Location: Germany
Try finding a database of good and bad dumps (such as gamedb/gamedb_vs.txt in a BizHawk folder):
sha1:91FA719B4B05ADBAC0B9D507D2051ED361D1DED4	U	VS. Super Mario Bros. (VS) [a1]	NES		board=MAPPER099;system=VS;palette=2C04-4;
sha1:C81B9D98851E9BBC6172A10FFB218823D3C3B223	B	VS. Super Mario Bros. (VS) [b1]	NES		board=MAPPER003;system=VS;palette=2C04-4;
sha1:DD09C4FD6B440FBA948A5DC5269F5302A4DE75A2	B	VS. Super Mario Bros. (VS) [b1][o1]	NES		board=MAPPER003;system=VS;
sha1:3E73109098A478F6AC3EA3CCCCA19BA3A31010C6	B	VS. Super Mario Bros. (VS) [b2]	NES		board=MAPPER099;system=VS;palette=2C04-4;
sha1:0C538B47CBB6CE03988B80ACF8A9AFEB68A6A179	B	VS. Super Mario Bros. (VS) [b3]	NES		board=MAPPER099;system=VS;palette=2C04-4;
sha1:EB9E0F838B4688C241B30516BE132E723EA21044	O	VS. Super Mario Bros. (VS) [o1]	NES		board=MAPPER099;system=VS;palette=2C04-4;
sha1:B21AA940728ED80C72EE23C251C96E42CC84B2D6	G	VS. Super Mario Bros. (VS)	NES		board=MAPPER099;system=VS;palette=2C04-4;
Your a1 version is marked as U, meaning it's unknown what kind of dump it is. The only confirmed good dump is the one with a headerless sha1 hash of B21AA940728ED80C72EE23C251C96E42CC84B2D6. And this one happens to be the one boct1584 already found, which unfortunately desyncs.
Warning: Might glitch to credits I will finish this ACE soon as possible (or will I?)
Joined: 5/1/2015
Posts: 96
Location: Florida
Question about ROM dumps What makes a ROM dump good? It was a question that I was going to ask on my Super Mario Advance 4 TAS, which got rejected before I could ask it. What qualifies a ROM to be "good"?
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
dhxo
He/Him
Joined: 11/21/2016
Posts: 62
Location: Germany
What was said in the forum is that there is no frame rule in this game. Anyway, getting a meh from me, because there is still a little room for improvement, from my point of view.
You are reading this.
Skilled player (1707)
Joined: 9/17/2009
Posts: 4952
Location: ̶C̶a̶n̶a̶d̶a̶ "Kanatah"
KnucklesMaster368 wrote:
Question about ROM dumps: What makes a ROM dump good? It was a question that I was going to ask on my Super Mario Advance 4 TAS, which got rejected before I could ask it. What qualifies a ROM to be "good"?
I literally searched that up, and found it in the first result http://forums.qj.net/rom-discussion/6946-verified-good-dump-whats-mean.html
That means that it is a good rom that someone has checked and said that there should be no problems with it. Your problem may be emulator related.
In other words, likely means the file is an exact copy of the original, with no bit errors in it. Edit: https://www.emuparadise.me/help/romnames.php Some examples of "bad" dumps at the bottom of the page. I'm not sure if I'm allowed to link that site, since it's not directly to a rom. But if it's not allowed, the quote:
STANDARD CODE NOTES [a] This is simply an alternate version of a ROM. Many games have been re-released to fix bugs or even to eliminate Game Genie codes (Yes, Nintendo hates that device). A bad dump often occurs with an older game or a faulty dumper (bad connection). Another common source of ROMs is a corrupted upload to a release. [f] A fixed game has been altered in some way so that it will run better on a copier or emulator. [h] Something in this ROM is not quite as it should be. Often a hacked ROM simply has a changed header or has been enabled to run in different regions. Other times it could be a release group intro, or just some kind of cheating or funny hack. [o] An overdumped ROM image has more data than is actually in the cart. The extra information means nothing and is removed from the true image. [t] A trainer is special code which executes before the game is begun. It allows you to access cheats from a menu. [!] Verified good dump.
Joined: 5/1/2015
Posts: 96
Location: Florida
jlun2 wrote:
In other words, likely means the file is an exact copy of the original, with no bit errors in it.
Could a TAS, made for one good rom dump, desync on another good rom dump? If not, then my [a1] TAS should be rejected because of a bad dump. If so, then this TAS is still in play. Regardless, I am creating a new TAS with the verified ROM dump. Currently at: 6-3
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Post subject: Re: #5564: KnucklesMaster368's NES VS Super Mario Bros in 10:14.42
fsvgm777
She/Her
Senior Publisher, Player (221)
Joined: 5/28/2009
Posts: 1185
Location: Luxembourg
TASVideoAgent wrote:
This TAS is giving the warning that it starts from a savestate. I can safely confirm that the TAS can be loaded on power on with no warning. It does not need the savesate.
You selected "Record from now" instead of "Record from start". By the way, the savestate is actually embedded into the FM2.
Steam Community page - Cohost profile Oh, I'm just a concerned observer.
Post subject: TAS FlagPole Glitch
Joined: 5/1/2015
Posts: 96
Location: Florida
I have been given several notices that my method for flagpole glitch is slow/RTA/not perfect. The problem is, when I look up how to do this trick the TAS way, it seems to be confusing. What should I look for when performing it, and what inputs are required and when? This is for future reference, as Mario is at the stairs on 4-1 in my reTAS
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Joined: 9/8/2006
Posts: 19
Although the [a1] ROM I used worked fine and played it through to the end, I felt it was appropriate to check the run with the savestate removed and replaced by an insert-coin-and-press-start sequence, plus an appropriate amount of frames so that at least World 1-1 would sync. It was fine until 6-4, at which point Mario runs smack into the first podoboo and it all goes pear-shaped, which makes me think I need to pad it more for a different frame rule or something. EDIT: Padded a few more null frames at the beginning. This is what I've got now.
Joined: 5/1/2015
Posts: 96
Location: Florida
Eclipsed Moon wrote:
Although the [a1] ROM I used worked fine and played it through to the end, I felt it was appropriate to check the run with the savestate removed and replaced by an insert-coin-and-press-start sequence, plus an appropriate amount of frames so that at least World 1-1 would sync. It was fine until 6-4, at which point Mario runs smack into the first podoboo and it all goes pear-shaped, which makes me think I need to pad it more for a different frame rule or something. EDIT: Padded a few more null frames at the beginning. This is what I've got now.
It does indeed work! Should I resubmit the file, or just continue to let Masterjun judge it?
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Masterjun
He/Him
Site Developer, Skilled player (1971)
Joined: 10/12/2010
Posts: 1179
Location: Germany
I can replace the file if you want.
Warning: Might glitch to credits I will finish this ACE soon as possible (or will I?)
Joined: 5/1/2015
Posts: 96
Location: Florida
Masterjun wrote:
I can replace the file if you want.
Please do so, though I don't think it will affect your decision. The community will see.
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Joined: 5/1/2015
Posts: 96
Location: Florida
I wrote:
I have been given several notices that my method for flagpole glitch is slow/RTA/not perfect. The problem is, when I look up how to do this trick the TAS way, it seems to be confusing. What should I look for when performing it, and what inputs are required and when? This is for future reference, as Mario is at the stairs on 4-1 in my reTAS
Never mind, I brute-forced it in 4-1. (could do it in 1-1 but 363 happens)
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Joined: 11/15/2004
Posts: 804
Location: Canada
I'm not sure why there's so much confusion about what a good ROM dump is. Good means that the ROM was copied successfully from an unaltered source. It's a bit-perfect copy of the original. As for the question of there being more than one good ROM dump for a given game, usually that only happens if there was more than one version (i.e. v1.1 fixes a bug that was present in v1.0), or if the game was released in more than one region (NTSC 60 Hz vs PAL 50 Hz, for instance). It's not okay to use a good dump from one region in a machine for another region -- using a PAL ROM in an NTSC console/arcade machine will artificially speed up the game. Other than that, you're generally free to use whatever version makes your TAS the fastest, i.e. using v1.0 because you can exploit a time-saving bug that was removed from v1.1, or using the Japanese version because the text scrolls faster than the US version.
TASing or playing back a DOS game? Make sure your files match the archive at RGB Classic Games.
BigBoct
He/Him
Editor, Former player
Joined: 8/9/2007
Posts: 1692
Location: Tiffin/Republic, OH
hopper wrote:
...or using the Japanese version because the text scrolls faster than the US version.
Actually, according to the Movie Rules, that sort of time gain is NOT considered to be "faster" because it is not an improvement to the gameplay.
Previous Name: boct1584
Joined: 5/1/2015
Posts: 96
Location: Florida
There is a grand total of 1 difference between "VS Super Mario Bros (VS)" and "VS Super Mario Bros (VS) [a1]", and that difference is: The timer runs slightly faster in the a1 version. Here are a few examples: 1-1: (VS) 363, (VS a1) 357 1-2: (VS) 335, (VS a1) 324 4-2: (VS) 348, (VS a1) 340 Keeping this activated the a1 version probably IS a good rom dump, with dipswitch 7 (the one controlling timer speed) activated.
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Post subject: Should I add the video?
Joined: 5/1/2015
Posts: 96
Location: Florida
I have a YouTube video of this, but it has a few caveats: 1. It was before the file replace, meaning it is 10:14.42, not 10:14.92 2. It was a direct recrod from OBS, meaning it looks not too great. But it has some pluses: 1. It was recorded on the day it was originally finished (2/12/17) 2. It has IL commentary done by me on closed captions. It however, is mostly for non-speedrunners and non-TASers. Should I edit the post to add it?
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD
Masterjun
He/Him
Site Developer, Skilled player (1971)
Joined: 10/12/2010
Posts: 1179
Location: Germany
KnucklesMaster368 wrote:
Keeping this activated the a1 version probably IS a good rom dump, with dipswitch 7 (the one controlling timer speed) activated.
No it isn't. The DIP switch settings are not part of the ROM and only change register values. The difference between the good ROM and the [a1] ROM is the timer indeed, but here is the difference: The good ROM normally decreases the timer after 20 frames, or after 17 frames with DIP switch 7 set. The [a1] ROM normally decreases the timer after 17 frames, or after 13 frames with DIP switch 7 set. The timer values in the ROM are changed, making it a faulty dump.
Warning: Might glitch to credits I will finish this ACE soon as possible (or will I?)
Joined: 5/1/2015
Posts: 96
Location: Florida
Masterjun wrote:
The good ROM normally decreases the timer after 20 frames, or after 17 frames with DIP switch 7 set. The [a1] ROM normally decreases the timer after 17 frames, or after 13 frames with DIP switch 7 set. The timer values in the ROM are changed, making it a faulty dump.
Could a TAS set DIP switch 7 before starting it up in the future, if possible, or is that not allowed? Also, this makes the dump bad. (VS Super Mario Bros (VS) [b4])
If a map had all the information, it would look like this: (shows black construction paper) "Now that's a map!' XD