Skilled player (1430)
Joined: 7/15/2007
Posts: 1468
Location: Sweden
Best I've got is 1.75 pixels shorter than the old pattern, going 2>3>3>3>5 with dash held on frame 1-3, 6-7, 10-11, 14-15, 18-22. I think that should be the best you can get with a 5 cycle stutter but I could be wrong. If it was possible to do a good 4 cycle pattern that could be shorter but I haven't found anything that worked. Also I noticed that more than just keeping the longer running animation from kicking in with speedbooster equipped, running without it actually makes the animation shorter, so the charge pattern is a few frames shorter. Still seems to be about 1.5 pixels too long to charge a shinespark in any new place that was impossible before.
Agare Bagare Kopparslagare
Skilled player (1318)
Joined: 6/27/2014
Posts: 87
Location: Finland
but now 2 taps can possibly be gotten in shorter distance :P (useful for spike-speedkeep)
Skilled player (1318)
Joined: 6/27/2014
Posts: 87
Location: Finland
Btw Cpadolf can you do this same test for pal version? I'm curious how short the charge can go there
Aran_Jaeger
He/Him
Banned User, Player (9)
Joined: 10/29/2014
Posts: 176
Location: Bavaria, Germany
Okay, so I think I have heard indirectly from Taco via Sniq that for the original input sequence for the TAS super short charge, a brute force algorithm was used to find the shortest distance in which a shinespark can be charged via running on a platform. (Provided this is true,) I assume that among these tests, the options that the algorithm went through were only those that are obtained by choosing for each relevant frame individually to press nothing or pressing Dash and/or Forwards (and/or Shot, since I am aware (due to crumble-block-bridge tests with remaining crumble-blocks after running past those with Speedbooster and then observing the pattern of remaining blocks) of the fact that the running pattern can depend on if one at least did shoot once or if one did not do so) (and/or pressing L or R, which would be very counterintuitive aswell as unlikely to help for minimizing the necessary distance, due to the known forwards-shift that those can cause, but well, at least they influence the running pattern somehow), but did maybe not involve the different motions given by turning Speedbooster on or leaving it off. Furthermore, since it at that time potentially might not have been known that the current level/value of the Boost-Counter (by that, I refer to the counter which, I guess, can have values among 0, 1, 2, 3 and 4 (when echoes will apear) ) is preserved as long as temporary BlueSuit movement restrictions are fulfilled, even if only the value 1, 2 or 3 but not yet echoes (that would correspond to the value 4) appear (*), it could have been that the brute force tests only optimized time/distance for the highest value that the Boost-Counter can obtain, but did not optimize individually for the 1st, 2nd and 3rd level. (*): As an example, by what I said above, I mean: If one, with Speedbooster turned on, runs to get to e.g. the first Boost-Counter level, then enters the angled crouch pose (as if having temporary BlueSuit), then this level is preserved as long as the usual temporary BlueSuit movement restrictions are fulfilled (and can be continued if one lands as ball on spikes and unmorphs appropriately, as it is known to be possible). Is this brute force program that was used for this instance still somewhere available so that one can use it for further tests regarding the super short charge (or maybe other actions that involve only a rather small time interval aswell as only a small number of different buttons that would be relevant for the frames involved, like climbing walls with walljumps or whatever could come to mind)? Generally, I would be interested in results (given probably by brute force tests) that optimize individually for the following things for both versions, NTSC and PAL, separately: Optimizing for (i) minimal distance or (ii) minimal time (frame-wise or in-game-time-wise), when trying to either get the 1st, 2nd, 3rd or 4th level for the Boost-Counter. Further ideas for getting a shinespark charged within little space or time were if one could somehow abuse right+left (and testing both, to run to the left or to the right, since there might be an asymmetry aswell) and/or combine it with Moonwalk (even if it might then not be relevant for TASing anymore); or, in the case of Samus being with her back aligned to a wall next to a (long) platform, to instead of starting the short-charge input sequence out of the static standing pose but start it from different poses that Samus has when moonwalking with her back against said wall and switch from that to the forwards motion since the time at which the magic running pose occurs might depend on that, theoretically. (Also, Sniq, you know there is an edit button to not have to double-post (although 1st, there usually is not that much action going on in this thread, so it does not break the flow of discussion and 2nd, I don´t know how strictly such things maybe or maybe not are handled on TASVideos)? Of course it is just a minor thing though. And yes, I am aware that if one (usually without changing settings) only edits a post, then it isn´t marked as such, e.g. as I did with my last post where I added the 14%PB-PB option that TAS has, which is not included in the graph.)
collect, analyse, categorise. "Mathematics - When tool-assisted skills are just not enough" ;) Don't want to be taking up so much space adding to posts, but might be worth mentioning and letting others know for what games 1) already some TAS work has been done (ordered in decreasing amount, relative to a game completion) by me and 2) I am (in decreasing order) planning/considering to TAS them. Those would majorly be SNES games (if not, it will be indicated in the list) I'm focusing on. 1) Spanky's Quest; On the Ball/Cameltry; Musya; Super R-Type; Plok; Sutte Hakkun; The Wizard of Oz; Battletoads Doubledragon; Super Ghouls'n Ghosts; Firepower 2000; Brain Lord; Warios Woods; Super Turrican; The Humans. 2) Secret Command (SEGA); Star Force (NES); Hyperzone; Aladdin; R-Type 3; Power Blade 2 (NES); Super Turrican 2; First Samurai. (last updated: 18.03.2018)
Skilled player (1430)
Joined: 7/15/2007
Posts: 1468
Location: Sweden
Well, I found the shortest pattern by manual testing and then it was later "confirmed" by bruteforcing. I think that was just done via a lua script, and only tested for the shortest pattern to reach the first booster level, and only tested for holding forward, dash, forward plus dash or holding nothing, so it didn't account for any weird quirks that could happen one way or another by using any other button combinations (or as we see now, starting without speedbooster). I don't think the script was posted but if you wanted to test for all that other stuff you'd probably have to write a completely new one anyway.
Sniq wrote:
Btw Cpadolf can you do this same test for pal version? I'm curious how short the charge can go there
I'll give it a try.
Agare Bagare Kopparslagare
Skilled player (1318)
Joined: 6/27/2014
Posts: 87
Location: Finland
Alright, thanks to the new shorter charge, it is possible to get 2 taps in 3 blocks wide platform now. It could maybe be used somewhere in the game for speedkeep with spikes that weren't wide enough before. http://dehacked.2y.net/microstorage.php/info/686665924/colosseum%20test%203.smv small edit: its not possible to jump unfortunately.. samus falls off ledge just as the 2nd speedcounter is executed. If it would be possible to shorten the charge a bit more, shinespark could possibly be charged in colosseum
Skilled player (1430)
Joined: 7/15/2007
Posts: 1468
Location: Sweden
The best I could find for PAL so far is 4-4-5, with dash held on frames 1-5, 8, 10-11 and then 14 and hold until you hit the first booster level to make the running animation as short as possible. It's just under 2.5 pixels shorter than the old pattern (and about under 6.5 pixels shorter than the new pattern for NTSC).
Agare Bagare Kopparslagare
Skilled player (1318)
Joined: 6/27/2014
Posts: 87
Location: Finland
Cpadolf wrote:
The best I could find for PAL so far is 4-4-5, with dash held on frames 1-5, 8, 10-11 and then 14 and hold until you hit the first booster level to make the running animation as short as possible. It's just under 2.5 pixels shorter than the old pattern (and about under 6.5 pixels shorter than the new pattern for NTSC).
Great! http://dehacked.2y.net/microstorage.php/info/1043706183/PAL%20ridley%20platform%20charge.smv almost worked at pal :P
Skilled player (1318)
Joined: 6/27/2014
Posts: 87
Location: Finland
We finally have an use for moonwalk! I named this trick "Hyperwalk". It allows turning around while keeping all previously gained speed and / or speed counter. The trick is to unmorph same time as getting hit by spike and then using moonwalk on the same magic frame where flash suit glitch works. Great idea by taco. http://dehacked.2y.net/microstorage.php/info/708173985/hyperwalk.smv http://dehacked.2y.net/microstorage.php/info/417521542/hyperwalk%20speedkeep%20turnaround.smv http://dehacked.2y.net/microstorage.php/info/193147810/hyperwalk%20stuff.smv http://dehacked.2y.net/microstorage.php/info/1883847329/hyperwalk%20stuff%202.smv
Joined: 5/12/2009
Posts: 748
Location: Brazil
I don't believe it. This is too awesome to be true!
Joined: 6/28/2007
Posts: 67
Sniq wrote:
We finally have an use for moonwalk!
Hey, don't forget that one frame it saves in Ceres! It wouldn't be too surprising if moonwalk can be abused in other ways. I don't think many people have ever really taken the idea seriously.
Player (88)
Joined: 11/14/2005
Posts: 1057
Location: United States
Now the question is how much time is saved during a run, compared to how much time it costs to enable the option itself. Anyway, I am sure there will be some epic techniques that come out of this discovery. Great job!
They're off to find the hero of the day...
Joined: 6/28/2007
Posts: 67
To provide an exact number, moonwalk takes 64 frames to activate. That is significantly less than some estimates I've seen (in fact, I'm pretty sure that number has never been mentioned anywhere since nobody has taken the possibility seriously, including me). Very minor frame saves throughout the game could be enough to make up for that, nevermind any moonwalk-specific glitches that are discovered.
creaothceann
He/Him
Editor
Joined: 4/7/2005
Posts: 1874
Location: Germany
hero of the day wrote:
I am sure there will be some epic techniques that come out of this discovery.
Someone just has to go ahead and Beat It.
Joined: 5/12/2009
Posts: 748
Location: Brazil
That's not that much if you can find at least a few good aplications. Also, you have to see how much you'll have to delay the intro to get perfect Ceres, if not any frame, you're already saving a few frames, if more, than the moonwalk option may cost a few more frames.
Joined: 6/28/2007
Posts: 67
Eye Of The Beholder wrote:
That's not that much if you can find at least a few good aplications. Also, you have to see how much you'll have to delay the intro to get perfect Ceres, if not any frame, you're already saving a few frames, if more, than the moonwalk option may cost a few more frames.
Well, in a realtime-oriented Ceres, that new staircase steam boost would be used unless near-perfect luck is obtained. The steam in that room is much easier to manipulate than the second to last room and I can't imagine more than 2-3 extra frames being lost on it. Either way, if just one or two good room strategies with the moonwalk glitch are discovered, it's probably worth it.
Joined: 5/12/2009
Posts: 748
Location: Brazil
Taco wrote:
if just one or two good room strategies with the moonwalk glitch are discovered, it's probably worth it.
So let's dig this game a little bit more.
Skilled player (1430)
Joined: 7/15/2007
Posts: 1468
Location: Sweden
Gaining any use for moonwalk is amazing, if it has any good spots for application this trick could potentially save quite a bit of time. It's just too bad that some of the cooler trick in the game seem to be limited to spots where you can bounce and take damage on spikes or other stuff with spike type damage (speedkeep, the new flashsuit, this hyperwalk).
Agare Bagare Kopparslagare
Joined: 11/26/2010
Posts: 444
Location: New York, US
Sniq wrote:
We finally have an use for moonwalk! I named this trick "Hyperwalk". It allows turning around while keeping all previously gained speed and / or speed counter. The trick is to unmorph same time as getting hit by spike and then using moonwalk on the same magic frame where flash suit glitch works. Great idea by taco. http://dehacked.2y.net/microstorage.php/info/708173985/hyperwalk.smv http://dehacked.2y.net/microstorage.php/info/417521542/hyperwalk%20speedkeep%20turnaround.smv http://dehacked.2y.net/microstorage.php/info/193147810/hyperwalk%20stuff.smv http://dehacked.2y.net/microstorage.php/info/1883847329/hyperwalk%20stuff%202.smv
Wow that is amazing to here that the Moonwalk might actually have use in a TAS, nice find sniq!!! Also I can't play the files for some reason what version of the emulator are you using?
My name is 4N6/Forensics.
Skilled player (1318)
Joined: 6/27/2014
Posts: 87
Location: Finland
lxx4xNx6xxl wrote:
Sniq wrote:
We finally have an use for moonwalk! I named this trick "Hyperwalk". It allows turning around while keeping all previously gained speed and / or speed counter. The trick is to unmorph same time as getting hit by spike and then using moonwalk on the same magic frame where flash suit glitch works. Great idea by taco. http://dehacked.2y.net/microstorage.php/info/708173985/hyperwalk.smv http://dehacked.2y.net/microstorage.php/info/417521542/hyperwalk%20speedkeep%20turnaround.smv http://dehacked.2y.net/microstorage.php/info/193147810/hyperwalk%20stuff.smv http://dehacked.2y.net/microstorage.php/info/1883847329/hyperwalk%20stuff%202.smv
Wow that is amazing to here that the Moonwalk might actually have use in a TAS, nice find sniq!!! Also I can't play the files for some reason what version of the emulator are you using?
I'm using both snes9x 1.43 rerecording and 1.51 rerecording. Those smvs are for 1.51 rr
Joined: 6/28/2007
Posts: 67
Moonwalk can be used as an improved form of acceleration storage. When Samus begins moonwalking, her momentum is immediately set to 32768 and remains there until moonwalking is canceled. If Samus begins running forward to cancel moonwalk, her momentum then builds up starting at 32768 instead of 0. These two smvs will show what I mean: Normal acceleration storage -- Starting at x = 37.0, Samus runs forward for four frames, releases forward for one frame, and then continues running. Max speed (2.0) is reached at x = 137.36864. Moonwalk acceleration storage -- Starting at x = 38.32768, Samus moonwalks backward for one frame, runs forward for two frames, releases forward for one frame, then continues running. Max speed is reached at x = 135.53248. Samus must begin at 38.32768 to avoid colliding with the door and losing momentum. The same is true if this trick is performed on a platform of some sort; Samus must begin a pixel and a half ahead of the edge to avoid falling off*. Maybe this could be used to reduce short charge distance even further, too? EDIT: * -- Apparently, this is not true. [6:21:47 PM] Sniq: it works!!! [6:21:51 PM] Sniq: as long as subpixel is 0 [6:21:55 PM] Sniq: samus can start 1 pixel from air [6:22:00 PM] Sniq: lets say [6:22:04 PM] Sniq: samus normally falls at 59 65535 [6:22:12 PM] Sniq: you can start running from 59 00000 [6:22:13 PM] Sniq: with moonwalk [6:22:14 PM] Sniq: lol [6:22:20 PM] Taco: I was wrong then! [6:22:23 PM] Sniq: previously 60 00000 [6:22:26 PM] Sniq: was the minimum [6:22:29 PM] Sniq: now you can start from 59 00000 So it's even better than I thought!
Player (88)
Joined: 11/14/2005
Posts: 1057
Location: United States
Seems like this will save quite a bit of time over the course of a run. Very much looking forward to seeing this implemented!
They're off to find the hero of the day...
Joined: 6/28/2007
Posts: 67
hero of the day wrote:
Seems like this will save quite a bit of time over the course of a run. Very much looking forward to seeing this implemented!
Well, there is a downside: the most commonly used pattern is to run for two frames. release for one, and then continue running. I compared to the four frame run in that post since it was the best known pattern distance-wise, but this moonwalk pattern takes one extra frame to set up compared to the two frame run. That makes it not as incredibly ridiculously insane as it may sound :) But it will certainly have its uses. I really believe moonwalk may be worth taking...
Aran_Jaeger
He/Him
Banned User, Player (9)
Joined: 10/29/2014
Posts: 176
Location: Bavaria, Germany
Cpadolf, provide us with the optimal distance-wise optimized patterns for NTSC and PAL, individually for all cases among (i) only adding to turn on Speedbooster suitably (maybe your previous results are already optimal though?), (ii) only adding the Moonwalk-start, or (iii) combining both. So, I think there would now altogether 6 interesting TAS short charge limits exist. Go, go, go! :D
collect, analyse, categorise. "Mathematics - When tool-assisted skills are just not enough" ;) Don't want to be taking up so much space adding to posts, but might be worth mentioning and letting others know for what games 1) already some TAS work has been done (ordered in decreasing amount, relative to a game completion) by me and 2) I am (in decreasing order) planning/considering to TAS them. Those would majorly be SNES games (if not, it will be indicated in the list) I'm focusing on. 1) Spanky's Quest; On the Ball/Cameltry; Musya; Super R-Type; Plok; Sutte Hakkun; The Wizard of Oz; Battletoads Doubledragon; Super Ghouls'n Ghosts; Firepower 2000; Brain Lord; Warios Woods; Super Turrican; The Humans. 2) Secret Command (SEGA); Star Force (NES); Hyperzone; Aladdin; R-Type 3; Power Blade 2 (NES); Super Turrican 2; First Samurai. (last updated: 18.03.2018)
Skilled player (1430)
Joined: 7/15/2007
Posts: 1468
Location: Sweden
I'll see if I get some time to do it, but otherwise it's not really that hard to find good pattern, it just takes a bit of work. First you need to see what the shortest running animation is (with speedbooster you do that by walking, without it you do it by running), and make sure that whatever pattern you use maintains that fast animation. Then you need to make sure that any dash frames that are necessary to add are added as late as possible while still maintaining the fastest running animation, and that your momentum always lands on either 0.32768 or 1.0 when you release forward (or slightly above if not possible). This is because momentum ticks down in increments of 32768 subpixels every frame after releasing forward, and start ticking up again after it's "below" zero. So your momentum would go the following way given different initial values: 28672 > 0 > 12288 32768 > 0 > 0 > 12288 36864 > 4096 > 0 > 12288 As for the running pattern you just test it in iterations and try to find something that's the right number of frames so that your booster level increases to 1 the frame after you release forward the last time. Generally you should end on a longer cycle (that gets your momentum to 1.0) and start on a shorter one, and in the middle you keep to 3-4 frames of holding forward. And you try to keep the number of cycles down as you move forward a full frame every time you start to hold forward.
Agare Bagare Kopparslagare