UPDATE: I finally manipulated Hammer Bros and am currently up to the Castle1.
Also, I found out two awesome boosts (both as Super Luigi)
One involves ducking, and the other involves the rope. When ducking, jump to the lower right corner of a block. If done correctly, Mario/Luigi would straighten himself up just as he hits the block, pushing him forward and boosting his speed to anywhere between 64-112.
As for the rope, Mario/Luigi gets a small boost upon grabbing the rope. And if they jump just before touching the rope, they do not lose any speed and still have that boost.
(sorry for the double post)
I've finished the run. I think its a little rough on the edges, but with some hex editing, it can definitely smoothen out. overall, i like how it all turned out. several spots i think can be optimized a bit further (i'm not that good at squeezing out frames from runs). but if you all see something that can be done, here is the .dsm file and video file. Other than that, what do you guys think of this run? Any comments or suggestions?
dsm file(if it desyncs, i may upload individual levels instead)
http://www.megaupload.com/?d=GUMSCQD2
video file (in wmv format because avi takes up too much memory.)
http://www.megaupload.com/?d=3ROAMPVE
Why did you get a blue shell in 8-3? I don't think I saw you use it anywhere. Also, both of the 8-F levels have slower times than what was listed in adelikats submission, was this just a mistake in timing on his part, or where you slower in these levels (seems unlikely as you are faster everywhere else, but still).
Other than that, it all looks pretty good, and it was quite entertaining (much more so than the published run).
Ok, now I noticed that the blue shell is faster under water and that adelikat measured the times in the fortresses from when he entered to doors to the bosses, so everything seems good. I say go ahead and submit unless you'd really like to try and tweak the movement optimization further.
8-T probably is slower because I was trying to manipulate Hammer Bros to move to 8-3 (to get Blue Shell which saves around 20 ticks). Since I was mini-Mario for most of 8-T, the only way I can manipulate is coin collection.
This was my WIP version from earlier this month
http://www.megaupload.com/?d=MU5L555P
8-T seems faster in that WIP vid, but Hammer Bros always seemed to move left to 8-2 instead of right to 8-3.
Joined: 6/13/2006
Posts: 3300
Location: Massachussetts, USA
Watched it, and it's much better than the published run. However, there's still improvements to be made*, and I think you'd be much better off if you restarted using a lag free version of desmume, because those periods in between levels are still boring. And they can be made faster! it really is a significant amount of emulation error lag, and it can, and should, be fixed.
I'd also like to say that your waiting antics during 8-2 were very good.
*
-Sometimes you don't land on the very bottom part of the flagpole when you could
-Your acceleration and rounding corners when underwater was very bad. Especially in that level where the big eel chases you... the beginning, you didn't round corners at all, and also you didn't approach the tubes that shot bubbles optimally either. If you swim upwards towards them or float down from above you want to clip the corner and hit those bubbles as fast as possible.
-Is ducking off a ledge when running with star power faster than doing a small jump?
-I mentioned to adelikat about when doing vertical jumps in a tower, if you need to make many short jumps, it may be better to duck and jump, for the added benefit of making shorter jumps and having faster gravity acceleration so that you land and can jump sooner.
-In 8-2, at 22 seconds, can you take damage and wall jump on those spikes to get to the door faster?
Watched it, and it's much better than the published run. However, there's still improvements to be made*, and I think you'd be much better off if you restarted using a lag free version of desmume, because those periods in between levels are still boring. And they can be made faster! it really is a significant amount of emulation error lag, and it can, and should, be fixed.
I'd also like to say that your waiting antics during 8-2 were very good.
*
-Sometimes you don't land on the very bottom part of the flagpole when you could
-Your acceleration and rounding corners when underwater was very bad. Especially in that level where the big eel chases you... the beginning, you didn't round corners at all, and also you didn't approach the tubes that shot bubbles optimally either. If you swim upwards towards them or float down from above you want to clip the corner and hit those bubbles as fast as possible.
-Is ducking off a ledge when running with star power faster than doing a small jump?
-I mentioned to adelikat about when doing vertical jumps in a tower, if you need to make many short jumps, it may be better to duck and jump, for the added benefit of making shorter jumps and having faster gravity acceleration so that you land and can jump sooner.
-In 8-2, at 22 seconds, can you take damage and wall jump on those spikes to get to the door faster?
I didn't land on the bottom of the flagpole because I didn't think there was any time loss as long as Mario/Luigi was below the actual flag. Plus, I believe there was manipulation of map movement with the score.
In the underwater levels, I actually did try to clip the edge of the bubble-shooting pipes (though it may not have been optimized). I think it looks like it wasn't clipping because I was constantly moving to the right and downward movement in the water is slow (if I went downward a frame earlier, Luigi's feet would touch the pipe, reducing speed; and if I went upward a frame earlier, Luigi would bump his head, and I would get into the bubble stream later).
As for ducking with star power, I wanted to get 72 Star speed for as much time as possible, so the quickest way to the ground is through ducking.
If I take damage on the spikes, I would end up fighting Bowser Jr without Fire Flower. Also, I think there is time loss in the "power down". I'll look into it. I'll also look into the duck-jumping to see what it can do to the run.
Cpadolf wrote:
Ok, now I noticed that the blue shell is faster under water and that adelikat measured the times in the fortresses from when he entered to doors to the bosses, so everything seems good.
I think in the current published run, he entered the door as early as possible. Its actually faster to enter doors as close to the middle as possible.
Joined: 9/2/2008
Posts: 289
Location: United Kingdom
In 8-2, at 22 seconds, can you take damage and wall jump on those spikes to get to the door faster?
You can't wall jump on spikes, and you'll be facing a longer battle with Junior that would last for 10 seconds if optimised conpared to 3 seconds with 6 Fireballs and a stomp.
I think in the current published run, he entered the door as early as possible. Its actually faster to enter doors as close to the middle as possible.
What I meant was that the amount of timeunits he reported to have left on the clock when competing the fortresses where measured from when he entered the boss door, instead of when the boss died which I did. That's the reason it seemed slower (because at first I didn't actually compare to his run by looking at it, I just looked at the times he had reported in the submission text). And so when I actually compared to his run I saw that you where noticeably faster in each fortress.
http://www.youtube.com/Noxxa
<dwangoAC> This is a TAS (...). Not suitable for all audiences. May cause undesirable side-effects. May contain emulator abuse. Emulator may be abusive. This product contains glitches known to the state of California to cause egg defects.
<Masterjun> I'm just a guy arranging bits in a sequence which could potentially amuse other people looking at these bits
<adelikat> In Oregon Trail, I sacrificed my own family to save time. In Star trek, I killed helpless comrades in escape pods to save time. Here, I kill my allies to save time. I think I need help.
I have a question, preferably toward adelikat or terrotim.
When I try to enter the pipe to the secret exit in World 1-2, I'm not allowed to do so until a certain amount of time upon landing has passed. All the runs that have been published enter the pipe immediately. How do you avoid this delay?
I have a question, preferably toward adelikat or terrotim.
When I try to enter the pipe to the secret exit in World 1-2, I'm not allowed to do so until a certain amount of time upon landing has passed. All the runs that have been published enter the pipe immediately. How do you avoid this delay?
It probably has something to do with which frame you land on... I could be wrong though...
This game is ultra hard to predict (at least for me...). And apparently, red ? block manipulation has something to do with firmware settings...
I have a question, preferably toward adelikat or terrotim.
When I try to enter the pipe to the secret exit in World 1-2, I'm not allowed to do so until a certain amount of time upon landing has passed. All the runs that have been published enter the pipe immediately. How do you avoid this delay?
Its been a while since I TAS'ed this game, but I believe you can enter a pipe faster if you are still moving horizontally as you go through the pipe (ie running into the pipe). If you are still in the air as you hit the pipe, your horizontal movement hits 0 until you land. And since horizontal movement is 0 when you landed, you get Mario/Luigi landing animation.
Of course this is just a theory.
mindnomad wrote:
This game is ultra hard to predict (at least for me...). And apparently, red ? block manipulation has something to do with firmware settings...
This is new to me. I was not aware of how firmware settings can affect red block movement.
New Super Mario Bros. "any%" TAS WIP
I'm working with mindnomad to create an improved "any%" TAS. This WIP goes up to the start of World 5.
We are currently 116 frames ahead of adelikat. However, we've still got a few problems: the pipe in World 1-2, and the pipe in World 1-Tower. If we could figure out what causes the delay before entering pipes, then we would be able to save about 6 frames more. From our experience, the delay seems to be reduced the more you hold left or right (without interruption).
If it's true that you need to hold left or right for a certain amount of time, then the pipe in 1-2 probably can't be improved because of how we need to stay in the Blue Shell whenever we can make a super corner boost (pressing left or right while in the Blue Shell will reduce your speed every frame if it's above 44). The gain from the super corner boosts really outweighs the delay. ;)
If you can find any improvements besides these, please tell us - we'd love to know!
Emulator Coder, Site Developer, Site Owner, Expert player
(3584)
Joined: 11/3/2004
Posts: 4738
Location: Tennessee
Nice WIP! You managed to use the star in 1-1 to save time, awesome. And super corner boosting in 1-2, for some reason I never thought of using that in 1-2, dur :S
The WIP desyncs for me after 1-2. Are you sure you are using Desmume 9.6? I hope you aren't using some interim/svn build. It needs to be able to sync on an official release.
I tried to improve the pipe in 1-2, but I couldn't . I couldn't figure out why you have a delay, it makes no sense. I thought maybe it was weird x subpixel positioning so I varied subpixels, 1 subpixel at a time, and no effect. I thought maybe it was y positioning, and jumped down there with small variations in y position, and nothing. I think you guys are just a victim of TAS Rule #34 All Improvements will have worse uncontrollable luck than their predecessor.
Also, you were 30 frames ahead upon entering 1-1, what is up with that? Did I screw up walking to 1-1?
Nice WIP! You managed to use the star in 1-1 to save time, awesome. And super corner boosting in 1-2, for some reason I never thought of using that in 1-2, dur :S
The WIP desyncs for me after 1-2. Are you sure you are using Desmume 9.6? I hope you aren't using some interim/svn build. It needs to be able to sync on an official release.
I tried to improve the pipe in 1-2, but I couldn't . I couldn't figure out why you have a delay, it makes no sense. I thought maybe it was weird x subpixel positioning so I varied subpixels, 1 subpixel at a time, and no effect. I thought maybe it was y positioning, and jumped down there with small variations in y position, and nothing. I think you guys are just a victim of TAS Rule #34 All Improvements will have worse uncontrollable luck than their predecessor.
Also, you were 30 frames ahead upon entering 1-1, what is up with that? Did I screw up walking to 1-1?
It syncs alright for me, but I noticed that the WIP has 1 more frame of lag than the published run.
By the way, I checked the dsm and apparently, you were already 17 frames behind after clicking the "Mario Game". Guess not everything was done at first possible frame :P.
Edit:
Y0SH1 wrote:
New Super Mario Bros. "any%" TAS WIP
I'm working with mindnomad to create an improved "any%" TAS. This WIP goes up to the start of World 5.
We are currently 116 frames ahead of adelikat. However, we've still got a few problems: the pipe in World 1-2, and the pipe in World 1-Tower. If we could figure out what causes the delay before entering pipes, then we would be able to save about 6 frames more. From our experience, the delay seems to be reduced the more you hold left or right (without interruption).
If it's true that you need to hold left or right for a certain amount of time, then the pipe in 1-2 probably can't be improved because of how we need to stay in the Blue Shell whenever we can make a super corner boost (pressing left or right while in the Blue Shell will reduce your speed every frame if it's above 44). The gain from the super corner boosts really outweighs the delay. ;)
If you can find any improvements besides these, please tell us - we'd love to know!
I haven't looked at how to save time in 1-2 or 1-T in depth, but I agree that the secret pipe in 1-T can be improved by breaking the bottom two bricks rather than the top two so you can technically have a running start towards the pipe.
I'm looking into those two levels though as we speak. I'll post what I find.
Edit2: After much fooling around in the area after the teeter-totter in 1-2, I was able to save 2 frames after the teeter-totter thing (1 frame from avoiding the walking animation shortly before entering the pipe and 1 frame from entering the pipe almost immediately) I theorize that this delay problem has to do with the jumping animation and where you collide with the pipe. Certain frames in the first, second, and third jump cause much delay and some hardly cause any.
Also, it seems like there is a small area of the pipe that, if you can collide into it, you enter the pipe the very frame after both x-speed and y-speed hits 0, even if you hold the dash button. It's a tiny area above the ground. For some reason, the x speed, upon colliding, goes from 48 to 16 then 0; the y speed stays at -64 until x hits zero, at which the speed becomes -23 then 0 at the next frame.
I've had average luck trying to recreate it (most of the time merely saving 1 frame instead of two).
It seems like there are some spots on the teeter-totter thing where you can get instant 48-50 speed instead of 25-30 speed, but I haven't been able to fully take advantage of it.
By the way, did you try wall-jumping off the red block that had the blue shell in it? I'm just wondering because you were able to do so with the star in 1-1.
The WIP desyncs for me after 1-2. Are you sure you are using Desmume 9.6? I hope you aren't using some interim/svn build. It needs to be able to sync on an official release.
Check you firmware settings. They should be at the default settings
Nickname: yopyop
Message: DeSmuME makes you happy!
Favourite Color: Turquoise
Birth Month: July
Birthday: 15
Language: English
If it's not syncing, then your other settings should be at default.
Also, you were 30 frames ahead upon entering 1-1, what is up with that? Did I screw up walking to 1-1?
TRT wrote:
By the way, I checked the dsm and apparently, you were already 17 frames behind after clicking the "Mario Game". Guess not everything was done at first possible frame :P.
Yeah... I've seen this problem happen very often. During the making of your movie, eventually your input file will add a few frames to the beginning for no reason.
TRT wrote:
I haven't looked at how to save time in 1-2 or 1-T in depth, but I agree that the secret pipe in 1-T can be improved by breaking the bottom two bricks rather than the top two so you can technically have a running start towards the pipe.
The 1-Tower problem has already been solved, so it's fine. :)
TRT wrote:
By the way, did you try wall-jumping off the red block that had the blue shell in it? I'm just wondering because you were able to do so with the star in 1-1.
We've tried that before and weren't able to do it. We were able to wall jump off of World 1-1's Red ? Block, however. (Maybe it has to do with the way it moves, or even the place it's positioned...)
TRT wrote:
It seems like there are some spots on the teeter-totter thing where you can get instant 48-50 speed instead of 25-30 speed, but I haven't been able to fully take advantage of it.
The first one was taken advantage of. However for the second one, I'm not sure if it'd help - that seemed to depend more on how long you stay on it.
It's great to hear that you reduced the time upon landing. All you had to do was time the last jump so that you jump with any of the three jumps and land at a certain animation state, right?
Also, an extra lag frame? That's not good...
