1 2
12 13
Posted: 6/11/2004 10:22 AM
(Edited: 1/23/2022 6:47 PM)
Quote
Bisqwit
Editor, Active player (297)
Joined: 3/8/2004
Posts: 7469
Location: Arzareth
I felt a bit awkward for beating Michael's movie with such small marginal - only 0.7 seconds is hardly an improvement in a 8 minute long movie. But we started the movies for this game as a competition anyway - Michael and I were both making movies for this game and we kept beating each other. So now was my turn, and I'm not a newcomer to this serie... There were some things I liked in this new movie. - In 1-2, at time=388, I took a shortcut through the wall. I think it saved some frames, although it resulted in a slow acceleration delay. - In 5-1, at time=337 and time=331, I stepped on two of three turtles. Some extra points this way. It's not easy. - In 8-4, at time=358, jumping through the piranha plant was cool. Mario goes to the pipe before the plant does. It's actually the same thing as in 1-2 at time=344. - In 8-4, at time=322, I finally managed to pass this point without any delays. Also, it was sort of cool to stomp a turtle with a hat. - In 8-4, at time=312 and time=287, Mario jumps through the flow of hammers. Unless you watch it in slow motion, it looks impossible. Other things: - In 1-1, at time=377, I skipped one coin to prevent the game from lagging. It saved one or two frames. - In 1-2, at time=323, I skipped one coin and did a walljump. I previously thought that it's slower than doing the regular way, but this time my experience proved otherwise. I tried something else here too, but it didn't work. - In 4-3, at time=365, we have previously always went the lower route, but this time I went the upper route. I got lots of more coins than usually. - In 4-4, at time=278, I tried many different ways of doing a walljump and I picked the one that was fastest. I also tried the through-wall trick (as seen in 1-2) here, but it was too slow here. - In 5-1, at time=317, I managed to do the walljump trick without any delays involved getting the favourable randomness. This is no news, but I was glad of having finally succeeded in it. - In 8-1, at time=381, I finally managed to pass this point without any delays. Previously I've always delayed a bit to get the turtles in a good position. - In 8-4, at time=392, I finally figured out how to get the fireball in a good tall position. Previously I always delayed a bit to avoid the ball. To be honest, here's the list of things that I didn't like when I reviewed my movie: - In 1-1, at the beginning, I perhaps should have gone the lower route instead, to get more coins. - In 1-2, at time=354, I feel like I jumped too much. - In 4-1, at time=375, the delay to get the spikeball in a better position was maybe a little too long. - In 4-2, at time=367, a week ago I stomped both the turtle and the bullet. This time I couldn't stomp the bullet. I couldn't figure out why. - In 4-3, at time=368, I'm not certain whether it was impossible to get those two coins while still going the upper route. - In 4-4, at time=273, I feel like this wasn't as fast as in Michael's video. But the time was better due to the walljump trick. - In 4-4, at time=268, I couldn't stomp the hammer brother, because Koopa (the castle boss) would have jumped to the opposite direction. I am not certain whether this is unavoidable. - In 5-1, at time=397, I have a feeling that it would have been faster to jump a bit later, despite getting only one coin in that case. - In 5-1, at time=324, I would have wanted to go the middle route, despite there's no gain in it. - In 8-1, at time=372 and time=364 and time=360, I couldn't stomp the hammer brother. - In 8-2, at time=368, braking disturbed me a lot. - In 8-3, at time=357 and time=354, I couldn't stomp the hammer brothers. - In 8-4, at time=396, I feel like my randomness control delay was too big. It's invisible though. - In 8-4, at time=366, I feel like it would have been possible to go a few frames faster. - In 8-4, at time=348, stomping the turtle would have gave a better score than displaying the hidden 1up. Yes, this is what I mean by being selfcritical. Ps: For the jumping period that starts from 8-4 at time=338, I copypasted my movements in the FMV file with a binary-safe editor. It serves no purpose but I thought it was funny. Edit: Fixed the time difference...
Posted: 6/11/2004 11:09 AM
Quote
tim333
Joined: 6/9/2004
Posts: 17
Wow, that's... incredibly obsessive. Good, though. These SMB2j videos are very amazing. It took me a month to beat that game, so it's quite shocking to see it done in 8 minutes.
Posted: 6/11/2004 11:22 AM
Quote
Locke
Joined: 3/29/2004
Posts: 224
Well done. BTW, Ever since I saw the first Mario time attacks on this site, I've been wondering how exactly the "stomp the turtle with your face/hat" trick works. Can anyone enlighten me?
Posted: 6/11/2004 11:29 AM
Quote
Bisqwit
Editor, Active player (297)
Joined: 3/8/2004
Posts: 7469
Location: Arzareth
Locke wrote:
I've been wondering how exactly the "stomp the turtle with your face/hat" trick works.
It's mostly about luck and right position, but when you are in upward motion at full speed, you can knock out flying monsters. In SMB2j and SMB3, doing this will gain you extra vertical speed boost. In SMB3, you can also stomp walking monsters (goombas, turtles, hammer brothers) this way.
Posted: 6/11/2004 11:32 AM
Quote
tim333
Joined: 6/9/2004
Posts: 17
These little glitches add a certain untouchable quality to SMB runs above simple speed... It's the "How the heck did they do that?" factor. Makes for a very entertaining video.
Posted: 6/11/2004 6:27 PM
(Edited: 6/11/2004 7:11 PM)
Quote
Michael_Fried
Former player
Joined: 3/8/2004
Posts: 1107
I didn't watch it yet, but in 8-4 were you able to get the plant to not be in the way so you don't have to wait for it to go down before entering the pipe? Edit: I just noticed that you cheated by stopping the fmv too early, so it's not really 1.5 seconds faster. Edit 2: It's actually only about 24 frames faster. Edit 3: I just noticed that even with the stopping earlier, your video was only 47 frames faster. 30309 - 30262 = 47. After some testing, I've determined that my video could have been stopped at approximately 30285 (assuming I just as early as possible to get Mario higher up before stopping the fmv), which makes your video only 23 frames faster. You may have found some good ways to save time, but I think it could have been a little more accurate in some places, especially now that we have the 60 fps hack. I'm also pretty sure it's possible for the plant in 8-4 not to get in your way which might save another second or two.
Posted: 6/11/2004 6:35 PM
Quote
Megaman951
Joined: 6/11/2004
Posts: 35
Location: Quebec, Canada
what's the file name of the rom you used to watch the movie ? I can't find a good rom of super mario bros 2 (J) anywhere...
Posted: 6/11/2004 7:13 PM
Quote
Michael_Fried
Former player
Joined: 3/8/2004
Posts: 1107
I think the one Bisqwit uses is called smb2j.fds, but I don't think that's the official name for that rom. Edit: There's also another way you could have saved time in a few places. Every time you do a wall jump to get up to higher ground, such as 1-2 and 5-1, you should do the walljump backwards since the corner doesn't slow you down as much. I haven't tried it yet in smb2j, but it worked in smb (like when I did the walljump in 8-4 backwards), so I assume it should work the same way in smb2j. Now with the 60 fps hack this works even better since you only have to press left for 1 frame so you barely slow down at all. If you did that in the movie you posted, maybe you would have been going fast enough to do the wall jump higher up. Another possibility that probably doesn't work (though I've never tried it) is to do 2 walljumps on the same wall. First do the walljump lower on the wall, and then do a second walljump higher up, and then you'll be able to reach the higher ground. This would obviously be too hard with the normal Famtasia, but maybe it could be pulled off with the 60 fps hack.
Posted: 6/11/2004 9:06 PM
Quote
Megaman951
Joined: 6/11/2004
Posts: 35
Location: Quebec, Canada
well I can't find that rom anywhere...and the link for the .avi file is broken. Can someone fix the link ?
Posted: 6/12/2004 12:06 AM
Quote
Bisqwit
Editor, Active player (297)
Joined: 3/8/2004
Posts: 7469
Location: Arzareth
Michael Fried wrote:
Every time you do a wall jump to get up to higher ground, such as 1-2 and 5-1, you should do the walljump backwards since the corner doesn't slow you down as much.
I actually jumped backwards from the wall to be able to accelerate before the corner is reached. This helped in 4-4.
Megaman951 wrote:
and the link for the .avi file is broken. Can someone fix the link ?
What do you mean the link to the .avi file is broken? I don't see any problem.
Posted: 6/12/2004 1:25 AM
Quote
Megaman951
Joined: 6/11/2004
Posts: 35
Location: Quebec, Canada
nevermind, the problem was coming from my side. Seems my BitTorrent was messed up and gave me an error msg whenever I clicked on a bitorrent link.
Posted: 6/13/2004 3:30 AM
Quote
Michael_Fried
Former player
Joined: 3/8/2004
Posts: 1107
Bisqwit, are you going to fix the smb2j movie description where it says how much faster it is?
Posted: 6/13/2004 1:54 PM
Quote
Bisqwit
Editor, Active player (297)
Joined: 3/8/2004
Posts: 7469
Location: Arzareth
Michael Fried wrote:
Bisqwit, are you going to fix the smb2j movie description where it says how much faster it is?
mfried2-smb2j.fmv is 60762 bytes, bisqwit4-smb2j.fmv is 60668 bytes. 762-668 is 94. That's where the number comes from. But FDS recordings have two bytes per frame, so the frame difference is 94/2 is 47. Sorry about that. Fixed the problem. You noticed some flaws in my movie. I'm not going to play the game for a week or two, so I welcome you to beat it if you want.
Posted: 6/13/2004 6:01 PM
Quote
Michael_Fried
Former player
Joined: 3/8/2004
Posts: 1107
Fixed the problem. What about the problem that you stopped the fmv too early? It was really only 23 frames faster. If you want I could edit my fmv by stopping it earlier and send it to you so you can see for yourself. Anyway, if by your rules it's ok to stop the fmv before you actually beat the game, should I also edit my SMB speed run to make it shorter and then send it to you?
Posted: 6/13/2004 8:08 PM
Quote
Bisqwit
Editor, Active player (297)
Joined: 3/8/2004
Posts: 7469
Location: Arzareth
Michael Fried wrote:
Fixed the problem. What about the problem that you stopped the fmv too early? It was really only 23 frames faster. If you want I could edit my fmv by stopping it earlier and send it to you so you can see for yourself. Anyway, if by your rules it's ok to stop the fmv before you actually beat the game, should I also edit my SMB speed run to make it shorter and then send it to you?
The only requirement I've had is that the movie must contain a sequence that leads to the ending without user input required. I'd also prefer if the movie is ended so that when the "Movie end" text is displayed, the user who watches the movie can't control the game so that the end isn't reached. I'm not certain whether this is true in my video (can you kill Mario by pressing LEFT when the movie ends?), but there are already some published movies (such as Phil's Castlevania 3 movie) where this is not true. Regardless, the only meaningful time I use is the length of the sequence in the FMV file. I do not measure distances between events in the game.
Posted: 6/14/2004 12:17 AM
Quote
Michael_Fried
Former player
Joined: 3/8/2004
Posts: 1107
So if I stop my smb2j fmv earlier and send it to you, would you change the description to say it's 23 frames faster? Also, should I do the same with my smb fmv so you can say the length is shorter? The only requirement I've had is that the movie must contain a sequence that leads to the ending without user input required. By my rules, the fmv should be stopped when user input has no effect on the game (which would be when Mario touches the axe, not half a second before). At first your rules might seem to somewhat make sense, but if you think about it more you would see it's nonsense. The perfect example is Arkanoid. What if you position the paddle in such a way that the ball just keeps bouncing without having to move the paddle any more. Let's say you stop the fmv, and then the ball keeps on bouncing for another 20 seconds, and then finally breaks the last brick? When is the game really over? Edit: By the way, I think I tested your fmv and found that I could kill Mario by pressing left. Trying to figure out the exact moment that Mario can't be killed by pressing left wouldn't make sense, there needs to be a fine (and clearly seen) line between when the game is over and when it's not, and the axe provides that line.
Posted: 6/14/2004 11:12 AM
Quote
Nach
Emulator Coder
Joined: 3/9/2004
Posts: 4588
Location: In his lab studying psychology to find new ways to torture TASers and forumers
IMHO, if you set up your actions that they will lead to the end of the game without actively getting there, all the better. Arkanoid is a perfect example, if you can set it up that it can beat the level for you 20 seconds later doesn't that show better skill?
Warning: Opinions expressed by Nach or others in this post do not necessarily reflect the views, opinions, or position of Nach himself on the matter(s) being discussed therein.
Posted: 6/14/2004 11:28 AM
Quote
Locke
Joined: 3/29/2004
Posts: 224
I agree, it seems more "perfect" when you can just let go of the controller and the game will finish itself.
Posted: 6/14/2004 10:10 PM
Quote
Michael_Fried
Former player
Joined: 3/8/2004
Posts: 1107
So what if it shows more skill? The game isn't over yet. Let's say that by moving the paddle in a certain way, you can break the last brick in only 15 more seconds, even if though you need to put in user input for a longer time. Which is the faster speed run, the one where you broke the last brick sooner or the one where you stopped moving the paddle sooner?
Posted: 6/14/2004 10:38 PM
Quote
Boco
Former player
Joined: 3/30/2004
Posts: 1354
Location: Heather's imagination
Michael Fried wrote:
Which is the faster speed run, the one where you broke the last brick sooner or the one where you stopped moving the paddle sooner?
The one where the last brick is broken sooner. The time is measured from first imput to when input ceases to affect the game in a meaningful way.
someone is out there who will like you. take off your mask so they can find you faster. I support the new Nekketsu Kouha Kunio-kun.
Posted: 6/14/2004 11:07 PM
Quote
Bob_Whoops
Former player
Joined: 3/19/2004
Posts: 710
Location: USA
How about this: After the movie ends, it should be impossible to not beat the game Beside this, the author can end it whenever they want. However, this can be used to compare two movies. For instance if one person get 1000 frames, and another person gets 990 frames, but ends it 15 frames earlier, the first person is still faster. I mean, it doesn't really matter if your time is 12313 frames, or 12308. For most intents and purposes, they are the same thing,
Posted: 6/15/2004 12:19 AM
Quote
Michael_Fried
Former player
Joined: 3/8/2004
Posts: 1107
After the movie ends, it should be impossible to not beat the game What if you can't avoid beating the game, but you can prolong beating the game. In smb2j, I could have held down the jump button and he would have reached the axe later. I think that all that matters is that input no longer affects gameplay, which would be at the exact moment that Mario touches the axe and is no longer moving.
Posted: 6/15/2004 8:36 AM
Quote
todd
Joined: 4/4/2004
Posts: 66
From my perspective (as someone who has watched many videos but never made one), it seems right to stop the movie when user input no longer affects gameplay. When I watch a movie that says it ends but retains the illusion of activity, I start to think things like "Hey! He could've still fallen in that pit or let his time run out! Sure, if he just drops the controller, he'll beat the game, but he doesn't HAVE to drop the controller!". At first, I thought Bob Whoops' "impossible to not beat the game" criteria was best, but I discovered two flaws with that criteria. Michael Fried pointed out one: what if you can prolong beating the game? Another problem is that sometimes it is difficult to determine whether or not it truly is impossible to avoid beating the game. Just as clever ways to save time are often not found until the 5th version of a video, a clever way to not beat the game might be discovered. So now I favor Michael's "input no longer affects gameplay". It's usually easy to tell when input does no longer affect gameplay, but if there's a game where this is difficult, perhaps it wouldn't be bad to fall back on the "no more input required" criteria.
Posted: 6/15/2004 8:55 AM
Quote
Bisqwit
Editor, Active player (297)
Joined: 3/8/2004
Posts: 7469
Location: Arzareth
I agree for this definition. I will fix this in the next release. This is for Phil too: he has been doing this in quite many movies... most notably Castlevania 3, where he "saved" about 15 seconds by positioning the character correctly waiting for the ball to drop down later.
Posted: 6/15/2004 1:37 PM
Quote
Sarmu
Joined: 6/15/2004
Posts: 9
a clever way to not beat the game can simply be pause the game
1 2
12 13