SSBM Break The Targets [TAS]: Peach [7.08] [Former WR] tauKhan 2022-07-11 | First complete with all new 1 in 25 000 000 000 probability, 4x bomb pull route. No codes were used to affect Peach turnip pulls, or anything else. The run was done on PAL, from starting seed 0x4DA230AC. With perfect rng, the route would yield 7.00, 5 frames were lost in taking a slightly slower descend to bomb drop for target 7.The route requires a non-item turnip pull, followed by 3 consecutive bomb pulls afterwards. The 4th pull could also be a sword. Additionally, there's 5/12 chance that the 4th item drops from Peach getting hit by the explosion on target 7. The naive chance to hit all those events with a set of inputs, starting from a random seed is therefore 127/128 * (1/384)^3 * 1/256 * 7/12, which amounts to roughly 1 in 25 000 000 000 probability.Melee uses a simple LCG with a single 32-bit seed for all its RNG events, which is easy to emulate. Unfortunately though, the number of rng events occuring throughout a run is not constant, but depends on the random numbers generated for particular events in a chaotic manner. Because of that we can't yet fully predict exactly how RNG events progress. Additionally, the total number of available starting seeds is 'only' 2^32 = 4 294 967 296, so for any given set of inputs, it's very 'unlikely' there even exists a starting seed that would complete the route.To get around the limitation on number of starting seeds, I prepared slight alterations to run, that have no effect on finish time, yet they change rng events of the run. That essentially let me 'multiply' the number of available starting seeds. To this date, I have 150 individual runs with different RNG, that would all yield 7.00 on perfect match.I wrote python script to find seeds that would have highest likelyhood of succeeding on the pulls for different runs individually. As while I can't exactly predict the number of rng events happening during a run, each run still has a reasonably predictable distribution for number of iterations occurring between each pull. For each run, I filtered into around 10 000 'most likely' seed candidates based on the distance of bomb pulls with the seed compared to the average number of iterations occurring with the run. I then used another script to go through a list of candidates for a run by having the script load a savestate and then swap starting seed, then play the run. Rinse and repeat with next candidate until the list of candidates was exhausted.Time saves:Largest time save was of course the damage boost from exploding target 7. The events around target 7 are intricate: First at 5.55 Peach z-drops her bomb, then NILs on the ground below. Still on the same frame, the bomb hits and breaks target 7. Then on next frame, Peach is actionable as she just NIL'd, and immediately pulls another bomb. Then later that frame, the previous bomb explosion from hitting target 7 finally connects on Peach, removing the lag from the new bomb pulling animation entirely. So effectively, Peach gets to do SDI from hitlag, and damage boost from explosion knockback, while 'keeping' a bomb for target 9 in hand. That saves enormous amount of frames compared to previous TAS.At the beginning, a falling turnip pull followed by double jump into turnip throw for target 3 saves around 15 frames. This is mostly because double jumping before turnip throw has Peach ascend during the throw, instead of falling down very low. With previous TAS route it would take a lot more time to dj back up after the throw. The airdodge into edge-cancel following turnip throw also uses double sweetspotted edge-cancel, tech that I debuted in my Ness TAS wr. In this run, the edge-cancel is ridiculously precise and only a single airdodge angle, of the available 6456 different angles, actually works. Peach slips off the edge after landing only by a margin of 0.00009 ingame units. For reference, a phantom hit occurs in Melee if hitbox overlap is less than 0.01 units, for instance.After first bomb pull, a slightly delayed airdodge cancelled bomb throw was used. The delay was done because of how ECB bottom offset fluctuates during fall animation after the throw: With later throw, Peach is at earlier fall animation frame when getting just above the platform, and is actually able to NIL while still moving slightly upwards due to ECB bottom expanding down. This saved 1 frame over the previous TAS wr.Previous wr by PracticalTAS for reference: youtube.com/watch?v=g5HEmEVlZWk .
Brotato 182% Danger 5 Demon tauKhan 2024-09-07 | I discovered I'm a terrible Demon player, don't have a good handle on how to play the shops.
Brotato 182% Danger 5 Mutant - My First Ever True 182% win tauKhan 2024-09-04 | In the September 3rd beta patch, difficulty modifiers were changed to be multiplicative with other enemy health modifiers, such as Danger 5 extra damage and health. This means enemies now have a little bit more health and deal a little bit more damage on the highest difficulty settings.
Brotato Beta Patch Entrepreneur Endless Waves 120 - 134 tauKhan 2024-09-01 | Beta patch brought some fun new toys for endless; Scaling experience from Celery Tea allows for constant level ups deep into the run, Particle Accelerator makes enemies stop still with high engineering, Greek Fire deals health% dmg unmitigated in endless at the moment.Robot Arms give engineering + melee damage for constant Plank dps scaling with little extra help from Nail. Decomposing Fleshes gives lifesteal on level ups, which scales Sharp Tooth damage.In the 100+ waves the game definitely outscaled this build though as one can probably tell from watching; I'm only really alive because Greek Fire still kills bosses and elites eventually, and I have just enough tardigrades... for now.
SSBM btt mismatch TAS, Kirby/Roy 5.21 [WR] tauKhan 2024-05-13 | Previous wr by sockdude youtu.be/Y9DSHtU2EFM
Ssbm btt mismatch tas, Marth/ICs 9.82 [wr] tauKhan 2024-05-06 | previous wr by jenkem youtube.com/watch?v=Sg4cinMVKmU
SSBM Break The Targets [TAS]: Jigglypuff [5.77] [WR] tauKhan 2024-04-20 | Previous wr 6.78 (PracticalTAS) youtube.com/watch?v=r-DAzVunNvcThis run saves 61 frames over previous record, by using uptilts to reach mid/high moving boxes to trigger "burner glitch". Previous record was bottlenecked from having to wait for the lowest block to arrive onto the log bridge in the middle floor.The moving blocks in the middle floor are not burners, and they don't usually damage puff. To get the 2 damage boosts rightward I used burner glitch twice. The way it works is the glitch essentially tricks the game into acting as if puff had touched the burner wall on the top floor again, activating the burner damage. When a character is *pushed* by a wall, i.e. a wall is in the way of the characters destination such that the character has to be displaced, information about that collision is stored into the characters data. Rightfacing walls, leftfacing walls, floors and ceilings all have their own entries. When a wall is *touched*, the game looks which wall of that facing last *displaced* the character, and if that walls type was burner then the character is damaged accordingly. However, there's a way to "phantom touch" walls by such a thin margin that no displacement is needed, in which case the wall displacement data doesn't get updated at all. But, they still count as a wall touch, and trigger stage hazard check. So the game looks at the displacement data to determine if the character gets burned. Since the data was not updated by the "phantom touch", the character then gets burned if the *last* wall of the same orientation that did displace them was a burner.In this run, the only rightfacing wall that displaced puff is the burner collided into at the beginning. After that all rightfacing wall touches are avoided, except for the phantom touches with uptilts to trigger the burner glitches.
SSBM Break the Targets[TAS]: Mario [7.28][WR] tauKhan 2023-12-25 | Almost 2 years after the invention of this new Mario route that climbs to top immediately after clearing right side targets, the old 7.31 TAS wr was finally broken.Sockdude came up with the new route, and PracticalTAS managed to push it to exactly tie AJPs 7.31. I managed to cut 2 frames off PracticalTAS's run using a ceilingbonk ECB manipulation into dair autocancel to climb on top of the stage slightly faster. The idea wasn't new, I did try to do ceiling bonk variations before too, but only managed to tie 7.31 back then. I missed that for some at the moment inexplicable reason, in this instance Mario loses significantly less height when bonking of ceiling extension instead of the actual ceiling. Ceiling extension is a straight line strip that extends ~1 unit beyond the dimensions of the actual displayed ceiling, and moving through the extension makes you still hit the ceiling.The extra height after hitting extension allowed the dair ECB reset autocancel to work; when hitting the actual ceiling, Mario would not be high enough and couldn't get on top of the floor properly before ECB bottom lockdown ends.Previous wr by AJP: youtube.com/watch?v=AMcZrKGQQb4
SSBM BTT Mismatched TAS peach/puff 4.62 [WR] tauKhan 2023-10-30 | All item pulls coded, obviously.Previous wr 4.99 by jenkem66: youtube.com/watch?v=G12KLyFhckw
FoD stage failure tauKhan 2023-10-07 | Luigi falls thru FoD due to main stage collision group being pre-filtered out momentarily, as he misses tech on platform that descends below floor.
SSBM Break the Targets[TAS]: Kirby [10.87][WR] tauKhan 2023-02-15 | Done on PAL.Had left a relatively loose frame on the 10.88 run. Swapped upair to airdodge waveland in the middle to bair IASA double jump airdodge waveland. As IASA frames don't allow airdodge, but dj is possible, dj airdodge lets kirby airdodge sooner, and land one frame earlier. The earlier airdodge loses horizontal distance on the waveland, but with the new top right strat that takes out the 2 high targets first, the lost distance doesn't matter.Kirby is probably the closest of any character to cutting a frame of: It's possible to fair and fast fall toward target 3 a frame earlier, and whiff the target by 0.0007 units. Which is 7% of phantom hit distance, or less than 0.1% of Kirby's air mobility in a frame.
SSBM Break The Targets Mismatch TAS: Samus - Peach [5.82][WR] tauKhan 2023-02-13 | Cleaned up the run. Beginning improved 2 frames with jump bomb instead of running bomb.Cleaner timing on 2nd last bomb, and this time I don't turn around before last bomb, bc wall jump already gets me turned around for t10 fair (lol)
SSBM Break The Targets Mismatch TAS: Samus - Peach [5.93][Former WR] tauKhan 2023-02-12 | Route improved from jenkem's last record, that introduced the new ending with fair through the wall. Added bomb to zoomy left in the beginning, and the small zoomy in the end to save 0.25 seconds in total.
SSBM Break the Targets[TAS]: Bowser [7.46][WR] tauKhan 2023-02-05 | Done in PAL version, previous record 7.48 by PracticalTASRun is exactly same as PracticalTAS previous record, except with slightly improved ending. Instead of using double jump into diagonal waveland to slide off the pole before t7, I didn't double jump, and went for full horizontal air dodge waveland. This improvement delays the fall off by 1 frame, but in return I gain a lot of extra horizontal distance, as the extra horizontal slide frame on top of the edge takes me much further to right than aerial drift frame would. That in turn allowed a much earlier fast fall than before, so I slide off the bottom platform 2 frames ahead, despite being 1 frame behind in fall at the top. The horizontal distance off the bottom plat slide off was just a tiny bit off from improving the record with 2 frames, but in the end resulted in just 1 frame save.
SSBM Break The Targets Mismatch TAS: Samus - Pikachu [7.35][WR] tauKhan 2023-02-04 | Frame grinded jenkems previous wr, (youtube.com/watch?v=AmH2_3GMtAs) route remains same.Saved 1 frame early with delay on 2nd missile, that lands frame later but edge cancels. The 5 remaining frames were saved with grapple break bomb at the top that lands on the plat much sooner, so the daughter dash chain starts earlier, netting extra fall frames.
SSBM TAS - Samus Race to the Finish, last door in under 20 seconds tauKhan 2023-01-25 | This stage was quite fun to TAS for Samus, as it gives opportunity to showcase pretty much every super wavedash technique: repeated daughter dashes, zoomies, zoomy leaps and even 2 grapple reel in dashes.
SSBM Break The Targets Mismatch TAS: Samus - Kirby [7.88][WR] tauKhan 2023-01-17 | Fun level to do, the strat evolved a lot in the making.The last super wavedash / zoomy is probably the longest single one in a Samus BTT yet. Its very precise. To be able to take advantage of the small slope, the upair for target 9 had to be delayed a frame, such that Samus would fall to right height at the end of her upair, instead of being already in fall. As her upair has a slightly higher ECB on the last frame, it allows her to get close enough to the ground to land quickly after morph ball. Even so, I was very barely able to make it past the first edge. A frame longer on the first floor would cost 5 frames, because then Samus would not land on the middle platform early enough to be able to goomy zoomy properly under the last target.
SSBM Break The Targets [TAS]: Peach [7.00] [WR] tauKhan 2022-12-24 | The perfect 1:25 000 000 000 odds Peach complete is finally here! With starting seed 0x9CF5BC17, Peach TAS wr once again matches best time with forced item pulls.Previous record(+5 frames): youtu.be/f4e2vMrtPX0Best time with previous route (by PracticalTAS)(+42 frames): youtube.com/watch?v=g5HEmEVlZWkSince everything is mostly same as with the previous record, except seed is now better and allowed uncompromised ending, the description is still relevant:"The route requires a non-item turnip pull, followed by 3 consecutive bomb pulls afterwards. The 4th pull could also be a sword. Additionally, there's 5/12 chance that the 4th item drops from Peach getting hit by the explosion on target 7. The naive chance to hit all those events with a set of inputs, starting from a random seed is therefore 127/128 * (1/384)^3 * 1/256 * 7/12, which amounts to roughly 1 in 25 000 000 000 probability.Melee uses a simple LCG with a single 32-bit seed for all its RNG events, which is easy to emulate. Unfortunately though, the number of rng events occuring throughout a run is not constant, but depends on the random numbers generated for particular events in a chaotic manner. Because of that we can't yet fully predict exactly how RNG events progress. Additionally, the total number of available starting seeds is 'only' 2^32 = 4 294 967 296, so for any given set of inputs, it's very 'unlikely' there even exists a starting seed that would complete the route.To get around the limitation on number of starting seeds, I prepared slight alterations to run, that have no effect on finish time, yet they change rng events of the run. That essentially let me 'multiply' the number of available starting seeds. To this date, I have 150 individual runs with different RNG, that would all yield 7.00 on perfect match.I wrote python script to find seeds that would have highest likelyhood of succeeding on the pulls for different runs individually. As while I can't exactly predict the number of rng events happening during a run, each run still has a reasonably predictable distribution for number of iterations occurring between each pull. For each run, I filtered into around 10 000 'most likely' seed candidates based on the distance of bomb pulls with the seed compared to the average number of iterations occurring with the run. I then used another script to go through a list of candidates for a run by having the script load a savestate and then swap starting seed, then play the run. Rinse and repeat with next candidate until the list of candidates was exhausted.Time saves:Largest time save was of course the damage boost from exploding target 7. The events around target 7 are intricate: First at 5.55 Peach z-drops her bomb, then NILs on the ground below. Still on the same frame, the bomb hits and breaks target 7. Then on next frame, Peach is actionable as she just NIL'd, and immediately pulls another bomb. Then later that frame, the previous bomb explosion from hitting target 7 finally connects on Peach, removing the lag from the new bomb pulling animation entirely. So effectively, Peach gets to do SDI from hitlag, and damage boost from explosion knockback, while 'keeping' a bomb for target 9 in hand. That saves enormous amount of frames compared to previous TAS.At the beginning, a falling turnip pull followed by double jump into turnip throw for target 3 saves around 15 frames. This is mostly because double jumping before turnip throw has Peach ascend during the throw, instead of falling down very low. With previous TAS route it would take a lot more time to dj back up after the throw. The airdodge into edge-cancel following turnip throw also uses double sweetspotted edge-cancel, tech that I debuted in my Ness TAS wr. In this run, the edge-cancel is ridiculously precise and only a single airdodge angle, of the available 6456 different angles, actually works. Peach slips off the edge after landing only by a margin of 0.00009 ingame units. For reference, a phantom hit occurs in Melee if hitbox overlap is less than 0.01 units, for instance.After first bomb pull, a slightly delayed airdodge cancelled bomb throw was used. The delay was done because of how ECB bottom offset fluctuates during fall animation after the throw: With later throw, Peach is at earlier fall animation frame when getting just above the platform, and is actually able to NIL while still moving slightly upwards due to ECB bottom expanding down. This saved 1 frame over the previous TAS wr."
SSBM Break The Targets [TAS]: Ness [11.75] [WR] tauKhan 2022-06-02 | Recorded in PAL, this TAS breaks PracticalTAS's previous wr by 7 frames. Link to previous wr: youtube.com/watch?v=dLJ0BNoSGRoRoute is still mainly same with minor adjustments.Improvements:Teeter cancel into full jump djc back fair after pk fire for t2 saves 4 frames overall. It lands 2 frames faster than slipoff dj fair, and further to the right, cutting another 2 frames in sliding time.Very precise airdodge up onto the moving platform before target 5 lands 1 frame sooner, while Ness still has slight upwards speed, but his ECB moves down barely enough to allow landing.The last 2 frames are saved with the airdodge edge-cancel on the pillar after target 6. Ness uses double jump 1 frame later to gain extra height, losing a frame, but with the gained height slideoff after airdodge happens 3 frames faster, netting 2 frames total. The last edge-cancel utilizes a novel mechanic or technique, "double sweetspotted edge-cancel" I found. In Melee, grounded characters needs to move at least 0.1 units over edge to actually fall off, and they can stand on that 0.1 overhang extension. If coming from horizontally above ground on prior frame, a character may also land on that 0.1 overhang strip from air! However, theres a catch when the edge is in corner with a wall connected to it below. Because then if the character moves through the overhang strip, they (may) actually hit the wall connected to the edge, and the wall collision then pushes the character beyond the overhang. That makes the character miss the ground and fall beside the wall. Because of that, in previous TAS'es where diagonal airdodge was used to get up on top of a corner, and then drift back utilized to edge-cancel the other way, we always landed on the actual stage, and then had to slide over the 0.1 overhang. This requires at least 0.1 + character traction+ of speed toward the edge, which takes time and space to accumulate. Space as in one needs to then get some distance onto the top of stage, before drifting back because the character moves back while accelerating toward the edge. The novelty I realized is that if the character also "sweetspots" vertically accurately in addition to horizontally onto the edge, i.e. their bottom collision point barely peaks through the ground, then wall collision in the corner doesn't necessarily even happen, or if it does the push maybe tiny enough to not prevent landing. So with such very precise double sweetspotted edge-cancel, I was able to land very close to the end of the overhang strip, and only needed traction worth of velocity to then slide off. That could be done spending 3 frames less time on accelerating toward the edge.
SSBM Break the Targets[TAS]: Fox[5.21][WR] tauKhan 2022-05-23 | This run beats previous TAS wr by 1 frame. The frame was saved when getting up on top of pillar in beginning. Walljump waveland was swapped to upwards diagonal airdodge into drift teeter cancel. With the teeter cancel fox becomes actionable on the same frame as with the wj into waveland, but is on the left edge of the pillar and doesnt need to dash into jump, netting the 1 frame.This run was performed on PAL. While on PracticalTAS's previous record, both NTSC and PAL could finish on 5.23, for this run only PAL gets to 5.21. This is because Fox is slightly lighter in PAL and therefore gets more speed from burner damage boost in the end. On NTSC, the laser would break t9 on 5.23. With added distance from the damage boost in PAL, the laser manages to hit t9 by 5.21.PracticalTAS's previous record for reference: youtube.com/watch?v=mroUP08Y_GM
SSBM Break the Targets[TAS]: Kirby [10.88][Former WR] tauKhan 2022-05-20 | This run beats my previous record by 3 frames, by switching order of targets on top right corner. Thanks to earlier waveland, jumping up for the 2 top right targets first is slightly faster. Turnaround dual hit bair for those targets ended up being 2 frames faster than upair.Done in PAL
SSBM Break the Targets[TAS]: Kirby [10.93][Former WR] tauKhan 2022-02-14 | Follows the route of previous record by AJPanton. Saves 1 frame by wavelanding onto the right stage section with an earlier airdodge.Done on PAL.
SSBM Break The Targets [TAS]: Samus [6.77] [WR] tauKhan 2021-05-17 | Long overdue, this record was achieved in February 2021.It improves my previous record by 1 frame, by going from landing to 1 frame of crouch into morph ball, instead of directly starting the morph ball from the landing at 1:83, prior the first goomy zoomy. For some obscure reason, Samus skips 2 first frames of her morph ball when moprh ball is done out of SquatWait ("full crouch"). So it's a frame faster to do 1 frame of crouch (SquatWait from landing) into morph ball rather than starting the morph ball directly.Feel silly having missed that frame initially. But I guess others missed it too.
SSBM Break The Targets [TAS]: Roy [4.25] [WR] tauKhan 2021-05-15 | Achieved on May 15, 2021, 4f improvement over previous record by PracticalTAS youtube.com/watch?v=XVoOBuUQHFQThis record uses invincibility bypass glitch.The time is saved by substituting target 3-4 jab with an upb into double flipper hit to bounce upwards. While it takes 2 frames longer to start the 2nd upb for target 5, a lot of height is gained compared to the previous strat, which allowed roy to hit the upper flipper much earlier, and avoid hitting the bottom flipper on the way up, saving on hitlag frames too.For the bounce up from bottom flipper, I needed to have initial launch slightly leftwards, to face rightward for two reasons. For one, if Roy faced left, he'd hit the bottom burner and endure hitlag frames. Also, Roys hurtbox reach is much more favourable when facing rightward, allowing more upwards trajectory. The up and slightly left trajectory was DI'd to almost completely upward launch, to end up at high altitude, while narrowly evading the upper flipper.While I get knocked by the 2nd flipper 5 frames earlier than the previous record, I wasn't able to do so at quite as optimal position, because I'd whiff target 5 otherwise. I couldn't get a great leftward launch directly at all. Instead, I had to resort to SDI into phantom off of the first touch. Then I had to double SDI (first outward then up and inward) from the phantom into proper 2nd hit for the final launch leftward.
SSBM Yoshi Btt TAS 6:52 (WR) tauKhan 2021-04-19 | With few new improvements, this run saves total of 17 frames over the last published record (6:80 by PracticalTas)Unreleased wrs over last few days:6:62 (Me) [-11f]6:60 (Me) [-1f]6:58 (PracticalTas) [-1f]The new beginning on targets 1 to 4 with teeter cancel didn't actually end up saving time.Most of the time save comes from the very precise eggtoss for target 8.The unreleased records use full jump backair for target 8 instead, which is about same speed as the eggtoss, but the eggtoss leaves yoshi facing backwards towards the remaining targets, allowing faster finish with djc bair instead of fair.
SSBM Break The Targets [TAS] Samus [6.78][Former WR] tauKhan 2021-01-12 | This is the 4th iteration of new Goomy Zoomy routes I came up and co-developed. Needless to say, I'm pleased with the result of my first TAS project, as the previous world record was smashed by 19 frames. Obviously at this point I can't say for how long this record will hold though.Shoutouts to Stadium Discord peeps for pushing me further in optimization, and PracticalTAS for individually coming up with this final route and proving it was faster, while I previously made a preliminary run with it, and incorrectly discarded it as lower potential.Also, I really did luck into insanely good rng for ending on first my try, when the expectation prior was to have to finish the run on thousands of different RNG seeds to find the last frame.
![SSBM Break The Targets [TAS]: Peach [7.08] [Former WR]](https://i.ytimg.com/vi/f4e2vMrtPX0/maxresdefault.jpg)