PurpleNebulaSaves 1 A press as it holds a previous press in the level. You can fall off the edge and then, for some reason, slide backwards along the wooden rim, only landing once reaching the grass. This means you have the chance to spin closer to the star.
also I showed off how high the tilted sling star landing can jump
Cloudy Court Comet star grab 0.5xAPurpleNebula2021-04-29 | Saves 1 A press as it holds a previous press in the level. You can fall off the edge and then, for some reason, slide backwards along the wooden rim, only landing once reaching the grass. This means you have the chance to spin closer to the star.
also I showed off how high the tilted sling star landing can jumpFlash Black GS2 improved ending (30.76s)PurpleNebula2024-06-22 | thanks to keegansimyh (https://m.twitch.tv/keegansimyh/home) for helping me understand that cancelling a flutter in a certain way can affect fluttering later on. The crouch to cancel the flutter before the launch star was causing a delay on being able to flutter on the last planet. Taking that out saved 16 frames (am still able to crouch after cancelling to enter launch star 3 frames earlier). A few more frames were saved by figuring out how to airspin after the ground spin on the last planet; airspin before fluttering adds some extra initial speed. This was done by Wiimote spinning the ground spin (2 frames) and starting to Nunchuk spin in the middle of that shake to get as early a spin as possible.SMG2: Supermassive Mario float to launch star 0.5xAPurpleNebula2024-04-12 | Important for minimising Luigi ghost textboxes and also the Any% route.FAST Racing League: Rookie Beach ShortcutPurpleNebula2024-04-05 | Should save over 1 second. The first clip was the first time I got it, showcasing with a lower leap and landing into the black booster, the second clip gets a higher leap and so can turn much more before landing. Both these clips are roughly the same time but the second one probably has more timesave especially with the bonk at the end.[TAS] SMG2: Starshine Beach GS3 in 42.28sPurpleNebula2023-05-01 | Improvement on the previous vid on main channel: youtu.be/xmnvjrK1zgg
Turns out it is actually possible and faster to get a boost from a leaf and do the big shortcut with 3 clouds, albeit needing to ground pound the leaf to get a good enough boost from the rebound. Was tested before but barely didn't make it and didn't try backflipping straight onto that leaf, instead going to the droopier one further back and jumping-ground pounding on the higher one from there (this was the first route that worked but was slower than waiting for a cloud flower).[TAS] SMG2: Tall Trunk Green StarsPurpleNebula2022-11-30 | Green Stars TAS WIP Star order between GS2 and GS1 comes down to how their first planet routes work with a land cancel. GS2 benefits much more as the land cancel can combine with an instant sideflip for a faster start.
Green Star 2 - 21.70s IGT - 49:36.31 real-time split (1:19.78 segment)
As a brief summary, the sideflip & backflip route up the tree is more than a second faster than the best mega triple jump route, and on the second planet, Yoshiless is 6 frames faster than using Yoshi.
Green Star 1 - 8.51s IGT - 50:28.16 real-time split (51.85 segment)
This improved on my 2020 TAS by 19 frames, with much of the improvement being using buffered crouch acceleration at the start to get enough speed for an earlier jump to be able to chain into a mega triple jump. The mega triple jump strat was found by Vellopia; the original footage has been deleted, but here is an IL: youtu.be/qTax6Inw6fs
Green Star 3 - 1:12.86 IGT - 52:36.20 real-time split (2:08.04 segment)
Doing a mega triple jump to the first plank and then backflip walljumping up was the fastest route by 37 frames, the closest alternative being a direct mega triple jump spin and walljump spin. Crouch acceleration was also used here after the land cancel in order to get more speed for the double jump.
When on the slide, there isn't really any difference to what lines you take when grounded, as long as you are going forward - but even holding back only loses 1 frame per 5 or 6 seconds. This gives freedom for swaying about the place, slaloming through Wigglers, spiky plants and coins. The optimisation comes entirely in airtime between obstacles and, more specifically, reducing it.
For the two spiky hurdles after the first tunnel, a spin off the side saved a frame each over jumping. For the gap in the next tunnel, a frame was saved by doing a late spin or early jump between the left of the dirt line and the line of spiky plants. This was better than early spins, or spinning or jumping at a different position.
For the next gap, it saved 2 frames to fall off early and spin, as Luigi reached the next ledge in such a way (sort of hitting the front of it) that it warped him forwards. For the double gap, it saved 2 frames to make the first spin late but the second as early as possible. For the gap shortly after, the best route was spinning early towards the left side of the next ledge, 1 frame better than going to the right. Finally, for star grab, spinning early saved a frame over jumping.
#supermariogalaxy2 #toolassistedspeedrunBlooper Launch into Starshine GS3PurpleNebula2022-10-01 | new ABC strat to save 2 A presses (0.5 required) Could it be a speed thing? Probably not, luring the Blooper up takes so much time and there is a fast cloud strat: youtu.be/hg8Tyo10qE8 This is tool-assistedDreadnought 1 OoB w/ Input VisualizerPurpleNebula2022-09-20 | controller tasing yay[TAS] SMG2: Hightail Falls Green StarsPurpleNebula2022-06-15 | WARNING: Some weird "flashing" (it's strobey but dark) when eating or single jumping with a Dash Pepper (or first person cameras). This is a graphical bug in the older version of Dolphin I'm using (5.0 stable). Green Stars TAS WIP
Despite the lag from an extra crystal break on the world map, it takes the same time to do Hightail Falls before Boulder Bowl than to do Boulder Bowl before Hightail Falls.
For details on GS3 and the majority of movement for the other two stars, see the description here: youtu.be/eFt2U7PtYkU
Green Star 3 - 57.23 IGT - 27:34.13 real-time split (1:53.28 segment)
Green Star 2 - 57.96 IGT - 29:15.45 real-time split (1:41.42 segment) The key difference from cutscene skips is the Paragoomba. It doesn't get to the good bouncing spot on the right, so it's faster to 2P freeze it on the left and go that side instead. For the ending, low rapid flutters are used to keep at higher flutter speed. Jumping off the side of the wall while facing upwards gets a similar amount of height as going straight up off the top, while being able to jump earlier.
Green Star 1 - 30.73 IGT - 30:29.55 real-time split (1:14.10 segment) As mentioned above, jumping off the side upwards is similar to jumping off the top, but it happens earlier. However, in this case, it didn't actually save time because it was so close to the top. It didn't lose time either. Dismounting into the star is 2 frames faster.
#SuperMarioGalaxy2 #ToolAssistedSpeedrunSMG2 World 1 Greens TAS (just the stars)PurpleNebula2022-05-23 | Test run for an encode idea once the TAS is finished. Cutting out at least 33 seconds of stuff between each level.Speed difference with camera change in Fluffy BluffPurpleNebula2022-05-13 | Update: Joselle actually has a doc on this stuff already: docs.google.com/document/d/14DAcbHw0OPSL2psmxpvAPtpAVdF_AYkYJyeE8GNZKjM/edit The findings here show that it is in fact to do with Luigi's direction in comparison to the camera (and parallel (east-west) is actually faster than diagonal).
Updated with extra footage of a fixed camera getting high speed if Luigi is going a different direction. Due to a constantly changing framerate when recording Dolphin, the speed values aren't correct for every exact frame, but I tried to at least sync them with each cloud jump and final landing, so they give an idea of the overall speed of each jump.
[1] Slopes against camera distance: youtu.be/7gjMQlb7KZY[TAS] SMG2: Rightside Down GreensPurpleNebula2022-05-06 | 2nd channel time
GS2 with description: youtu.be/Ikq9CPZoGRk GS1 (41.90 IGT) is much of the same which is why this is a 2nd channel thing.
The cutscene skip is different than I usually do it: for most levels, Luigi doesn't start moving for a little while if you immediately skip the fly-in, so for the sake of good IGT times and things looking a bit smoother, I delay the skip as far as Luigi's first movement frame, not losing any time. For this level however, as it starts in a pipe, it is faster to skip the cutscene immediately.
For the ending strat, a straight triple jump and three firespins to the star is possible but requires more work than it seems. The star's Z position is actually misaligned from the 2D restriction, closer to the back wall (likely because the intended way involves entering from the 3D section that's in line with the back wall). The edge of its hitbox is roughly -100 units on the Z axis. The 2D restriction goes up to where the star is, so to get to the star without entering the 3D area, you'd have to get above the star to move far enough over on the Z axis. A triple jump walljump is enough height, and distance is covered by first using neutral walljump strain to get more speed from the walljump, and then by getting a turnaround firespin (hold opposite direction 2 frames before spin happens) to move backwards in the air. This strat saved 22 frames over a direct triple jump + 3 firespins, and 1.1s over the intended way.
16:49.91 real-time split, 1:24.86 segment[TAS] SMG2: Yoshi Star GS3 w/ cutscenes in 55.10sPurpleNebula2022-04-01 | Uploading this here, all Yoshi Star Greens will be uploaded in one vid when they're done.
The start of the flutter looks awkward because you need to flutter close to ground to gain speed with backwards fluttering. Spinning gives a higher flutter acceleration but can't be done immediately after mounting.
For optimising the remount, the issue with straighter lines is that you need to be far enough to the right when bouncing off the fruit to keep the camera straight. This lets you keep backwards flutter, keeping a straight line. Adjusting the facing angle further left would mean more rightwards movement into the fruit bounce, so the whole flutter line would sway out further right anyways. The second issue is that the second flutter will always have a curved line, so you want to be slightly off on the first to curve accurately in the second.
#SuperMarioGalaxy2 #ToolAssistedSpeedrun[TAS] SMG2: Cosmic Cove Secret in 0xA (2xA) [Mario]PurpleNebula2022-01-21 | Excludes the Hungry Luma dialogue (7xA), but gets the 15 coins necessary to prove they are 0x on a new file.
I used a double spin at the end, ground spinning to hit the checkpoint flag on the edge of the floor to get some airtime, and then doing an air spin.Gravity Gauntlet lava boost skip 0xA w/o Life ShroomPurpleNebula2022-01-05 | Collects coins so not useful for Comet, but skips the need for Life Mushroom in main star and GS2 - particularly helpful skipping it for main star as then the benefit of the textbox A press could be used elsewhere.[TAS demo] SMG2: Starshine Beach 2 in 30.98s [Mario, cutscenes]PurpleNebula2021-12-29 | 30 is possible, nice. Improves on Hime's Any% TAS run of 31.51s.
I was doing this to compare with the shore launch route, which I have struggled to get fast because of SMG2's scuffed water physics. Predicting a low 31 from it at best.
Main timesave was at the start, using a ground pound right on the surface of the water to instantly dive in. Main improvement would be the long jumps, I didn't optimise them properly. Backflip from cloud is possible by slowing down enough vertically to snap to the floor of the cloud, avoiding the landing 'stun' and allowing a couple of extra frames of movement.
#ToolAssistedSpeedrun #SuperMarioGalaxy2Boulder Bowl Rolladillo route comparisonPurpleNebula2021-12-10 | This is tool-assisted. Been meaning to test this for a while now, comparing the crystal skip route with the launch star skip (aka roundabout) route, both found early last year.
When first doing the roundabout route a while ago, I found that I couldn't avoid Rolladillo rolling into a ball on the second phase, but didn't test further to figure out why. It's because Mario gets teleported to wherever he entered the first boss cutscene, which was too close to Rolladillo to get away from it and avoid the attack. So the roundabout route needs to enter the fight in a certain area for the whole fight to be fast.
Also rolling into Toad while hitting the large crystal saves 0.3s for the second hit as Mario bounces immediately after the recoil so you can start the second roll much earlier. Not sure how much it saves overall, though, considering the movement after breaking the crystal.Flip-Swap GS2 Crystal float w/ input visualizerPurpleNebula2021-11-14 | reverse gravity floating on a crystal[TAS] SMG2: Twisty Trials GS2 in 26.96PurpleNebula2021-10-02 | Another one from the backlog of TASes that I forgot to upload, this one from August last year. Thing of note for the cloud jumping: the first jump didn't need a Z cloud cancel because the cloud was just about formed before leaving the ground. A later cloud formation had the possibility of warping down slightly and losing all running speed.[TAS] SMG2: Gravity Gauntlet GS1 in 12.03 (fast CS skip cycles)PurpleNebula2021-09-05 | One from the backlog of TASes to upload. IGT could be improved by delaying cutscene skips to optimise the fireball cycles. This TAS was made for the comparison of fast CS skip vs. full cutscenes between both GS1 and GS2 to figure out which is faster to do first in a Green Stars run. Due to fireball cycles being slightly better (4 frames) without cutscenes, it's faster to do GS2 first.
Mega double/triple jumps weren't understood at the time; maybe there could be an improvement right at the start with this tech?Cosmic Cove GS3 cycles - IL vs cutscenesPurpleNebula2021-08-28 | This compares the water block cycles when skipping cutscenes vs not skipping cutscenes.
The first comparison uses the RTA movement for cutscene cycle, holding onto a stored long jump and waiting through the water blocks to long jump out to the star. This is 8.2s slower than the IL route which is also RTA viable (see Joselle's IL run: youtu.be/e3AaYj77buE?t=89).
But the cutscene cycle does have an extremely tight fast route, although it loses the potential water long jump as it has to take a single jump into the last water block. This is 1.1s slower than the IL route.
It's worth noting for RTA that, because the trigger is somewhere above the arrow, you could get a bit more leeway in a cycle by backflipping from the wall instead of the platform.
Green Stars order should be GS2 before GS3, to take advantage of the faster block cycle when skipping cutscenes.[TAS demo] Rightside Down 1 in 1:01.13 (Mario)PurpleNebula2021-07-10 | 138 starbits! And those last 4 could be duped too.
It's hard to see, but IR is used on the world map after moving right 1 space. This saves 11 frames over analog stick all the way.
I couldn't manage to get the instant wall release seen here with Luigi: youtu.be/CdOdK9QOn4U?t=13 I didn't test if a forward slope jump down the big coin slide would be faster (conserving momentum like in the prologue TAS).
The detour to the rock ring costs ~10 frames for 8 starbits (0.8 bits/frame), and the detour to where GS2 is costs ~35f for 40 starbits (1.14 bits/frame).[TAS demo] Spin Dig 1 in 2:14.73 (Mario)PurpleNebula2021-07-10 | Also with 104 bits. Missed a few bits on the 3rd planet, and this was before starbit duping was a general thing. Not sure if I checked Lumalee for 2P bits either. Maybe 110+ estimate?
For Captain Toad skip, I landed on the very edge of the rim on the planet before to set up launch star displacement. What makes this version of the skip faster is landing further along and then running under Lumalee and long jumping under it to get squeezed towards the launch star faster. Also means not wasting a starbit.Shore launch to complete Starshine Beach 2 without pressing APurpleNebula2021-06-28 | This was done 0.5xA only to have easier movement in water; 0xA should be possible.Gold Leaf C 0.5xA save as MarioPurpleNebula2021-06-26 | tool-assisted. not nearly as precise as the original Luigi strat, and also faster. this level would easily be 0.5x if it wasn't for Cosmic, which is why we aim to go for the tree strat. could it be 0x? sure, maybe, I guess?Ghostly 3 1x save using perforated planetPurpleNebula2021-06-26 | This is tool-assisted. Using the speedrun strat to save the A press from using a pull star to get to the launch star. I also held A from the previous planet - it is barely possible to do the previous planet 1xA without sling pods by bouncing and doing a walljump.Toy Time 1 Usernames mechakoopa ABC strat as Luigi 0.5xAPurpleNebula2021-06-26 | for the sake of ABC video footage for Luigi (user has Mario video on a Drive link, and did post a Luigi one on Drive but jumped as a setup: youtube.com/channel/UCCqf3TkCsvacOP3tTpbD3ZQ). also for the sake of showing the mechakoopa drop down to blue block, confirming no manipulation is necessary, it just drops down. used savestates but it could be an RTABC strat too - I didn't push the boundaries of where the mechakoopa can be.Luigi Purple Coin Chaos 1xA savePurpleNebula2021-06-19 | The coins on the ends of the rows are barely reachable with a running speed forwards airspin, weakspinning to grab the coin and switching backwards to land in the swamp right next to safe ground, which lets you get pulled up, as seen here: youtu.be/Qj2hvqlstp4?t=302
So getting the middle coins in 1 jump would save an A press, which was done here with a long jump, also landing in that pull-up edge of the swamp.[TAS] SMG2: Slipsand Comet in 0.5xA presses [Mario]PurpleNebula2021-06-12 | This run confirms that Mario is equal to Luigi in Slipsand ABC, meaning we can avoid the menuing A press for unlocking the Luigi ghost.
Fireball spamming is done by constantly shaking the controller and mashing Z. I had to get back onto the rock after the first slide as it didn't have enough speed to make it. Moving backwards in the air just before landing for the second slide provided an extra bit of speed to get the star.Slipsand 1xA saved with lj + spiny bouncePurpleNebula2021-06-08 | This is tool-assisted. Saves having to do two jumps, one to get onto the blue ring and the other to get onto the dirt ring. Fire damage meant both invincibility to not get damaged by the spiny plant, and also it gets more speed somehow. After bouncing off the spiny plant, I held up-left, sideways to Luigi's facing angle, which affected how Luigi moved up the planet, gaining some more distance to make this trick possible.Battlerock 1 OoB Backwards CannonballsPurpleNebula2021-05-08 | Notes for this route: - I needed to 2P jump once out of the disc's gravity, as a spin would push forwards away from the balls. Should be able to optimise the strength and tilt of the mega tj (landing a bit further up to jump higher up) to get around this. - That first barrier is annoying; might just be better to go down and long jump for a bit. - Backwards slope jump was somewhat helped by pressing and holding Z near the end of the slide. - If sliding angle can be improved, that backwards slope jump could lead to a bouncing route to get onto the final bounced ball without 2P jump, and from there get to the final barrier without a 2P slope jump. - I was able to freeze a ball behind the final barrier, but couldn't figure out how to freeze it enough to bounce on it, a strat which would mean only one more slope lj is needed rather than two. Although speed is decreased from an lj bounce, so would it be faster?
more than 30s faster than autoscroller. tool-assistedDreadnought 1 ending strats comparison (tool-assisted)PurpleNebula2021-05-05 | Side-By-Side: LEFT - Cutscene skip, triple jump 2P mid-air to beamer, wallkick w/ neutral strain going forward, spin cancel to avoid land stun, lj spin to star RIGHT - Jump cancel just to be further up but play cutscene, run and backflip 2P mid-air onto beamer, backflip into star
Learned about potential cutscene skip from StainedGlass: twitter.com/Mario_Speedrun/status/1380514755106893828?s=20 Beamer backflip was inspired by one that PodzDragon did off a lower beamer to get to the planet itself, in a Discord embed.
Should find a better movement for right side than just run to the backflip 2P spot. perhaps even the tilt of a 2P ground jump + spin could do better.Dreadnought 1 Star Grab (tool-assisted)PurpleNebula2021-05-04 | First, no cutscene skip so the beamers are moving. Then a mega triple jump spin up to the beamer (likely possible with a normal tj, especially as Luigi). Then I waited until the beamer started moving toward the star (hard to tell from up there) and backflipped.
This is tool-assisted. I kept the IR up to exactly replicate what StainedGlass did even if it made little sense that collecting starbits would have an effect. To clarify, it doesn't, this is done with a jump land cancel, pressing A at the frame-perfect time to miss the trigger. Said frame is 2 frames before landing. However, the trigger continues around some of the planet; I had to delay the spin so that I wouldn't trigger the cutscene in mid-air, getting far enough up the side to be able to land for a moment. The backflip was also frame-perfect (meaning a single jump has a 2-frame window), any longer would have triggered the cutscene.SMG2: Freezy Flake going to Secret: star 1 vs star 2PurpleNebula2021-04-11 | This is tool-assisted. I should have used Mario but, for such a short distance with no high jumps or long gaps, the differences are minimal.
Optimising the first planet with Fire Flower is strange - throwing fire in the air hitting the statue earlier still ended up slightly slower than the final result. Maybe the way the camera gets slightly blocked off by the planet means the launch star takes a bit longer to become active?
Rock could be optimised some more with a better initial angle, but the end result worked out fine.[TAS demo] SMG2: Flip-Out 1 in 57.98sPurpleNebula2021-04-10 | bad. Besides the obvious movement optimisation and the 2P textbox (I only connected P2 when entering star select lol), the main improvement here - if cycles are manipulated - is the spike jump that is still yet to be understood and replicated consistently: youtu.be/OyqIcsAJ4_I?t=53 and youtu.be/UplyeyR5ILp9k?t=3758
The megasideflip was surprisingly easy to pull off - there is a lot of area there that will get the trick working; the only times I didn't get it were when I got the wrong joystick angle.
I used neutral walljump strain for lower walljumps immediately after for that extra speed going into a long jump.
P2 can freeze spikes, and if you freeze before landing on spikes, you can run and jump off them. Also you can land and jump off the edge of spikes. Makes sense, as you can do a similar thing with lava.
Instant crystal break yayClockwork Ruins GS3 TAS w/ input displayPurpleNebula2021-04-09 | Disclaimer: Due to the fluctuating framerate from how Dolphin dumps frames, the input display might not be accurate to the frame in some places.
Also the places of low optimisation are much more obvious as I mash A a lot for wallkicks and long jumps lol
Featuring Xander's input display: github.com/MikeXander/WiimoteInputVisualizerMelty Monster Launch star skip (not fast)PurpleNebula2021-03-22 | this isn't fast but it's cool, should be able to do it without spinning so that would save a spin under the Shake ruleset.SMG1: Dusty Dune 3 accessing 2nd stars final planetPurpleNebula2021-03-22 | I was messing around with the OoB used for star chip skip and found that, underneath the planet when it is all the way down (the star chip tower is all the way up), there is a sideways gravity pull that falls towards the final planet of the 2nd star. I didn't go right when underneath the first planet because the change in gravity is much earlier and doesn't get as good a result before getting stuck in the gravity loop.
I didn't do much experimenting with trying to get to safe ground but I'm not sure if there's any platform close enough to make that work. This is tool-assisted, mainly just savestates and pausing.SMG: Dusty Dune 2 chips route (TAS)PurpleNebula2021-03-02 | This was 3 or 4 seconds faster than the normal way. But it might not matter in the end if pull star skip is proven. I went for launch star instead of pipe because I wanted to try out sign floating to the star.Dusty Dune resurrected shell homingPurpleNebula2021-03-02 | I assume the respawned shell started homing in automatically? This is a strange thing.SMG1: Freezeflame Secret no Fire Flower strat as MarioPurpleNebula2021-02-20 | (tool-assisted) vid is only of reaching the summit, but he can climb the summit too as seen in speedruns I did it as Luigi before, now as Mario. 1P maybe? But idk. That second 2P jump wasn't necessary, I just fell off the slope like a dodo and couldn't get back on with a regular spin.[TAS] SMG1: Honeyhive L in 28.65s [Mario]PurpleNebula2021-01-26 | did this ages ago (late 2019) but only showed it off as an mp4 embed in tas server. improved a frame from that old version by 2P jumping into the star. Attempted earlier shots, particularly using 2P IR instead of 1P as it comes from the left of the screen, but nothing worked out. I'm confident it will work but couldn't be bothered to keep testing. Xander's gonna smash this TAS up when he gets there. Interestingly I couldn't save a frame on textbox by turning the camera just before talking. Perhaps you could save something with a different talking position, but you need to make sure it's close enough to the star for an instant 2P jump at the end.[TAS] SMG2: Rightside Down GS2 in 55.80sPurpleNebula2021-01-23 | Everything is the same as GS1 up until the final Fire Flower crate at 0:49, check that description for details on movement: youtu.be/CdOdK9QOn4U Plus this improvement suggested by Yaasd: youtu.be/3t_9nZJqyJk
When branching off, most of the movement is long jumping, but there are a couple of important things to note. Double jump + single fireball (animation gets cancelled by landing) was a few frames faster than another long jump with 2 fireballs, and had enough momentum to instantly long jump unlike the long jump idea.
I delayed the second long jump so that I landed in the 2D section as early as possible, and while the star grab looks like it could be improved by fireballing earlier, the effect of entering the 2D section makes earlier fireballs redundant. When you enter 2D from the front, you get slowly dragged in with no way to change the speed, so entering early is the best option, and it means that earlier fireballs would still have to wait to be dragged into the star. In fact, fireballing earlier was a couple of frames slower than what I did, possibly because I hit the bottom corner of the star which, based on understanding from Fleet Glide GS1, has an extended hitbox.Yaasds Rightside Down bounce (0.5s saved)PurpleNebula2021-01-21 | Yaasd Palala suggested this in the GS1 TAS. Still need to iron out the kinks but even this attempt saved 0.5s.SMG2 Peewee 1:44.93 (TAS)PurpleNebula2021-01-08 | this was done to give a rough benchmark for Shadow64 in his Any% TAS, which he improved substantially by 1.07s (pretty big margin for a TAS) : youtu.be/IzcchUgsDMo
avoiding coin by delaying ground pound would add something like 0.25s to the time, but would save time in the post-level, so Shadow's improvement is 1.32s everywhere else since he goes for coinless.
textbox skip is weirdly more difficult to get with cutscenes, and also has a different effect in that the box actually appears for a moment. I had to do a tiny jump on the roof to get into the right spot, I couldn't manage to get a really high spin from the wallkick into the right spot. Also, while the cutscene camera panning is much slower than optimal (maybe 4 or 5f slower), it is necessary to get the camera to end up in a way that the trolleys go off-screen during the launch star launch, in order to manipulate their cycles.
Going further right at the start of the second planet keeps the yellow platform on longer, so that there isn't as much time to waste before it rises.
Again, that coin in the Luma bounce could be avoided with P2 by delaying the ground pound so the camera tilts up, so there's enough time for P2 to grab the coin before Mario hits the ground.
The trolley cycles were perfect. I didn't really do much to optimise the backflip on the penultimate planet.
I forgot that starbits are useful with Peewee, despite randomly grabbing some beforehand. More time to save with that weird starbit trick Shadow got? I was sure the earliest time to hit Peewee bare was 35f after the egg-break lag is finished, but both these hits took 39f.
Spin into star is faster by a couple of frames.Beach Bowl S No Ice updated (TAS)PurpleNebula2021-01-07 | with the allocation and the understanding of the lack of understanding
from what I've tested over several different tests in the past year or so, by jumping into water so the camera zooms in, and then getting out and also jumping up a bit, the camera will adjust itself just right to get a huge megasideflip at the start. Also this one has to have the mid-air angle sit outside the y-axis' central deadzone by no more than 3 joystick values i.e. with the deadzone being 113-143, y can only be 110, 111, 112, 144, 145, or 146.
also, I learned from this testing session that for the second megasideflip, the camera is affected by where you grab the ledge beforehand, and possibly also the time you jump out of the grab. At first I struggled to get a good megasideflip like I have in previous videos, but I tried to grab the same part of the ledge as a previous video and got better height from the same position and angle as the failed attempts. So there's definitely something there. I also switched from right to left during the megasideflip to increase the height even further. This was enough to 2P mid-air to skip the sling star. The angle for this one can be within the deadzone, but I also found that using those same values just outside the deadzone can slightly affect the height gain, possibly because they go slightly forward or backward.
What looks like a 2P wallkick is a regular mid-air from a spin cancel ground pound after wallkicking. I started Nunchuk shaking before wallkicking to get an earlier spin.
Surface spin probably isn't faster but it's fast.Gusty Garden 2 Mario Floaty Fluff 0.5xA SavePurpleNebula2021-01-07 | could it be 0x? idk. was tight 0.5x but the key part is how you swing around the planet which might not be so different. this is tool-assisted.[TAS] SMG1: Good Egg P in 0.5xA [Luigi]PurpleNebula2021-01-04 | supposedly you can get on top of the huge crystals for floating but who needs that when you can ground spin in the air from getting a bump off the crystal, use that ground spin to go backwards, and then air spin into the star.
evidence of 0.5xA Good Egg P, confirming the second* SMG1 multi-star galaxy to be possible without pressing A (excluding menus and dialogue). Honeyhive was first once it was confirmed you could get a flipbug down from the fountain area for the Purple Coins.[TAS] SMG2: Cosmic Cove GS2 in 1:06.71PurpleNebula2020-12-20 | The first couple of sections are the same as the GS3 run. The difference only comes in the last 2D section; instead of jumping further up to the Goomba, the route instead takes a water long jump across to the right water column. Going under the used block was 0.2s faster than going over, particularly because the jump and spin wouldn't have so far to drop onto the star. I had tried to get a spin knockback off the used block, but it barely fell short of the star, so a backwards jump and spin was the best strat.