Name: 100% OJ Co-op Solo (Star Devourer) in 55s (old PB) @jda7
Uploaded: 2026-09-18T08:17:00-05:00
Duration: 8 min 17 s
Description: 100% OJ Co-op Solo (Star Devourer) in 55s (old PB)

100% OJ Co-op Solo (Star Devourer) in 55s (old PB)

jda7 2026-09-18 | 100% OJ Co-op Solo (Star Devourer) in 55s (old PB)

OJ Co-op 4p Star Devourer 2m22s Voice&Game

jda7 2024-11-25 | Note* not a WR. just a PB for this set of players.
Game Audio Only version: youtu.be/xXVu8RudlWc
Buggle POV: youtu.be/nWY6ixW9UXU
Set on: 11/22/2024

Mixers: Aggressive Heroes, Evasive Heroes, Joker
Map: Cyberspace

Timing method: First frame of game to first frame of You Win fading in.

Forced revival strat to have a 1 hp big poppo got nerfed. Now you have to actually deal 6 hp after the revive instead of 1. This means we need one of the following to win:
- Take 50 stars from star devourer and defeat big poppo once
- Defeat big poppo once, revive them, deal 6 more hp

Also play of the gods doesn't use forced revival unless an ally is down. Big Poppo doesn't count.

The strat we went with here is to defeat big poppo once and use the stars from the defeat to norma to 3 so 2 marc hyper beats them. Granted, we just got lucky instead with a very large roll vs small roll at the end.

We also have some dark side of business to attempt the 1st win condition instead, but it wasn't the main focus of this strat. It was just to keep star devourer from getting 200 stars making defeating big poppo a moot point.

Since we have 4 players, we can manipulate play of the gods by bringing only 1 event card type and then we give tomomo the summer games hyper unlocked at lv 10 attacker. This means play of the gods only has forced revival to target.

We also have dash with joker mixer so other players can dash turn 1 to also attempt to fight big poppo besides just the closest.

OJ Co-op (Online) M10 (WR) 3m35s Voice&Game

jda7 2024-11-25 | Game Audio Only version: youtu.be/lu6IfQ1hTZA
Buggle POV: youtu.be/faIRjFt7rDA
Set on: 11/20/2024

Mixers: Blazing, Aggressive Heroes
Map: Cyberspace

Timing method: First frame of game to first frame of You Win fading in.

4 Atk Kae is very strong. And if they go down they have a much lower rec compared to a 2 base atk character. They also get an extra 2 atk from im on fire and then 2 more if you use their hyper blazing.

Party time is for getting the boss to you. Play of the gods helps play that for low cost and if even luckier you can instead get little war. Deck choices may be improvable in terms of low norma low cost card, but it was fine enough for this run.

OJ Co-op (Online) Haruo (WR) 3m14s Voice&Game

jda7 2024-11-25 | Game Audio Only version: youtu.be/dO87wevJSdw
Buggle POV: youtu.be/zMzRams4Jrk
Set on: 11/20/2024

Mixers: Blazing, Aggressive Heroes
Map: Cyberspace

Timing method: First frame of game to first frame of You Win fading in.

4 Atk Kae is very strong. And if they go down they have a much lower rec compared to a 2 base atk character. They also get an extra 2 atk from im on fire and then 2 more if you use their hyper blazing.

Party time is for getting the boss to you. Play of the gods helps play that for low cost and if even luckier you can instead get little war. Deck choices may be improvable in terms of low norma low cost card, but it was fine enough for this run.

OJ Co-op (Online) Big the Jonathan 3m27s (WR) Voice&Game

jda7 2024-11-25 | Game Audio Only version: youtu.be/XRAWjXpOnrA
Buggle POV: youtu.be/VMyqMmw5i-M
Set on: 11/20/2024

Mixers: Blazing, Aggressive Heroes
Map: Cyberspace

Timing method: First frame of game to first frame of You Win fading in.

4 Atk Kae is very strong. And if they go down they have a much lower rec compared to a 2 base atk character. They also get an extra 2 atk from im on fire and then 2 more if you use their hyper blazing.

Party time is for getting the boss to you. Play of the gods helps play that for low cost and if even luckier you can instead get little war. Deck choices may be improvable in terms of low norma low cost card, but it was fine enough for this run.

OJ Co-op (Solo) Swimming Coach 2m35s (WR) with mic

jda7 2024-11-24 | Game Audio Only version: youtu.be/MO8BpfJQ4F4
Set on: 11/22/2024
Mixers: Aggressive Heroes, Evasive Heroes, Blazing

The deck is a mix of draw power and survivability since the boss takes so long and you will basically take damage. So far all avenger tomomo is the best to likely live balloon jerk and then hit him.

Party time is to gather the goalposts in one spot to reduce the amount of times the CPUs fight them needlessly.

Tomomo's hyper is also quite good for swimming coach since just 1 teammate downed gives 3 damage and takes down the ringers. A 2nd use would also take out balloon jerk. Taking down ringers and balloon jerks are the highest priority with 2nd priority fighting the boss directly.

OJ Co-op (Solo) M10 49s (WR) with mic

jda7 2024-11-24 | Game Audio Only version: youtu.be/GSBTxZQfp1o
Set on: 11/20/2024
Mixers: Aggressive Heroes, Blazing

Very basic deck to high roll big damage at low norma & star cost. Play of the gods can play the party time or little war if lucky. Party time is good since putting everyone close together means m10 can challenge everyone that gets ahead of him.

Cyberspace can only be played on Normal or higher difficulty and it also gives a permanent boss panel for possible double fights.

OJ Co-op (Solo) Big the Jonathan 1:08 (WR) with mic

jda7 2024-11-24 | Game Audio Only version: youtu.be/50A4EXu-sao
Set on: 11/20/2024
Mixers: Aggressive Heroes, Blazing

Oops all kae. The deck was meant for m10 hence the long distance shot. I just didn't want to change it at the time. Party time is to gather all the CPUs and boss in one place. If lucky with play of the gods, you can play that little war. Both those could probably be different cards too.

Fluffy Bluff L in 2:19.95 (WR)

jda7 2024-02-24 | This was livestreamed on twitch.tv/jda7
If you want to hear the version with my voice click this: youtu.be/pcFAnjmkd8M

Done on emulator on the chinese version, since I don't own that version. I found that mario is slower at the beginning, but he saves that time back and more with the ending long jump.

This was a warm up to get back into speedrunning and test my new pc with streaming.

Fearsome Fleet C in 55.98 (WR)

jda7 2024-02-24 | This was streamed live on twitch.tv/jda7
For the live version with my voice follow this link: youtu.be/Py54koAvSpI

This beats the previous record by pottoww by a little under 4 seconds. A lower 55s time is definitely possible.

The main improvements are the new bullet bill lure strat and the flutter + jump shot from the hammer.

Here are the visual cue/strats I use in order:
-After first bullet shot, line up near the bults to the left of the yellow and black striped platform
-Shoot your bullet bill for hit 2 when he fires the first bullet
-Run to right point of the sling star shadow
-React to when the closest spikes from his left hammer (your right) go under the stage/out of vision. This is when you fire and hold forward for the double hit.
-Run a little past the yellow and black platform. When the bullets fire, start running the other way
-Shoot your bullet bill when the purple laser goes off the bottom of your screen.
-When the 2nd bullet bill hits, sideflip a little after. By waiting this long, you can fire the bullet almost immediately when you get to the peak of your flutter post cutscene.
-Last flutter is just face down right, jump and nunchuk spin so you can do the flutter sooner. If you made sure you weren't too far forward you can also get further on the hammer with the extra speed from grazing the ground with the flutter.
-After the shot, wait as long as possible before you spin. Also don't flutter after the spin. If your flutter gets cut off by the cutscene you get a delayed flutter.

Flip-swap 1 RTA Sim 28.78

jda7 2024-02-07 | I was going to make a more detailed video with commentary, potential pitfalls/mistakes while doing the strats, where to line up for jumps, etc. However, I have had this footage for over 2 years and haven't done that yet. I figured I will at least put this out there and also an imperfect comparison here youtu.be/qcYThHgALWM for the ending.

Maybe I'll get around to making that vid, but not for now. I'll probably do other stuff first.

Shiverburn C in 28.41 (No jump cancel)

jda7 2023-05-20 | Since you use the 1p wiiremote so little in the stage, you can just put it down somewhere close (i put it in my lap) and just play with nunchuk and 2p wiimote. This allows you to run with the rainbow star without spinning when 2p spins an enemy.

Cloudy Court S with Hungry Luma in 1:34.70

jda7 2023-05-19 | This is a 3.14s improvement over the time I got from jhay. This can still be improved to possibly a 1:33.xx. This also makes it so you don't have to walk to the cloud flower in cloudy court S if you do S first, which means it effectively saves an extra .5 or so

Improvements:
-Cut the run around for the 4 coins at the 2nd cloud flower
-Use backwards momentum on the drum section
-Small better cuts in the cloud section

Mistakes:
-The boost to the ? coin wasn't optimal. you can get all these coins with a much shorter jump. I just ended up short.
-I didn't optimize my ground pound spacing on the cloud flower at the drum section
-I could have created the cloud sooner and backflip to the ? coin while being lower for a faster bounce
-I went too far when collecting the second line of coins. You can ground pound much sooner.
-Lost running speed at the final silver star

Back ups:
-If you miss a coin, you can backflip on the goomba for an extra coin
-If you miss 2 coins, go high on the final drum bounce and boost to grab 1 coin. You can also spin to delay to get a 2nd coin on the way down.

Supermassive S - RTA Flower Planet Research

jda7 2023-05-18 | The TAS route for the planet is too hard, so I made a route that's actually possible for RTA.

To drift out of the launch star you need to wiggle the stick up and down in the direction you want to go. I don't know why it's necessary, but it is. Timing it sometimes works, but wiggling the stick is far more consistent even if sometimes there is slight variance where you end up.

The idea with this planet is to barely touch each flower's trigger to walk to smallest amount possible. The TAS uses a route with barely any cues on when and where to turn and makes it very easy to miss a flower. Every time I attempted it, I would often miss 2 or 3. I remedied this by creating a route where I would turn at specific points and memorizing which flowers pop up and where. For instance, I do that circle and hope to hit the small grass patch underneath the purple flowers in the beginning and then run to sunflowers. I then run down and do a couple zigzags to try and hit the pinkish and red tulips while making sure to hit the flower patches to the left as well. Once I get to the large pink flower patch that's when I know it is time to cut inwards towards the question mark block while barely grazing the large bush patch.

I noticed that this route also has a bit of time to save at the ending. I could have cut back downwards much sooner, so it is likely this only wastes .3s instead of .6s. Until that is shown in footage, however, I will consider it when routing as .6 slower.

For the ending, it turns out grabbing the vine as high as possible frame 1 is faster than grabbing the vine as high as possible.

I also showed a slower but more consistent RTA flower route at the end. This one loses all that time because it walks through the large bush patch instead of just grazing it like the harder route.

Boom Bunker - Launch Star Cutscene Shorten (Small Improvement)

jda7 2023-05-17 | There, now it is faster than the other strat definitively. I realized instead of spin canceling you could just do the 2p spin during the ground pound and focus on doing the long jumps as fast as possible. That's the whole improvement.

For this to work the following must happen:
1. Kick the boomerang bro off stage
2. Land neutral in front of the launch star, not beside it
3. Crouch after landing neutral

Shiverburn 1 - Wall Boost to CM comparisons

jda7 2023-05-16 | It saved less than I wanted since we do the level coinless, but this either adds a coin or makes the level a lot riskier since you have to take no damage from CM to the star.

I purposefully go farther than normal off the first cloud to avoid putting the planet on camera to keep the cycles frozen. This avoids some jank that can happen otherwise. My cue to spin is right when the camera is starting to turn.

This also can only be done on shiverburn 1 after shiverburn s (aka skipped cutscenes). The cycle doesn't line up otherwise. The cycle doesn't matter normally because the optimal strat is to just long jump out and spin for comet medal. For instance, both Doopsie and Goryuya do shiverburn 1 first in their PBs and HIME's TAS also just long jumps to comet medal and spins.

If there was a way to make it this wall boost work with cutscenes, then another .25s could be saved.

Tall Trunk 1 - TAS Sideflip visual cue and comparison

jda7 2023-05-16 | With neutral wall jump strain and this visual cue for where to wall jump, this trick is much easier now. If you are wondering what neural wall ump strain is, it is just returning the stick to neutral during your wall jump and then pressing the direction you want to move in after. The c button returns the stick to neutral so you'll hear me press it during these wall jumps while I simply hold where I want to go.

Doing a strain on both wall jumps saves almost a whole second.

Shiverburn 1 - RTA Improved Skate triple into enemy bounce

jda7 2023-05-15 | After discussion about keeping speed during ice skating, I decided to revisit this. From what I can tell you won't lose boost state until you jump while skating, but the cause of loss from jumping is the weird part. If it was a boosted jump, you tend to lose it upon landing after skating for a bit again. Long jumps sometimes take it away or just simply don't give enough speed when you do try to boost (possibly just not up to max skate speed while skating for 1.5x speed boost to be at its most viable).

Anyway, I made this more RTA safe by showing the back ups you can do with a camera turn. You only have to trust in your ability to gauge the amount of speed your skate triple jump got and decide to back it up or send it.

Shiverburn 1 - Skate triple into enemy bounce and comparison

jda7 2023-05-14 | I decided to look into HIME's bounce strat performed in the any% TAS. I still don't fully understand why sometimes you get more or less momentum during skate jumps, but I have better guesses now.

Current guesses on keeping speed:
1. Once you start skating make sure every spin is canceled before a certain point to ensure your spin jumps maintain the correct speed.
2. Long jump without spinning removes the stored jump speed
3. (Possible placebo) backwards skate before the spin single jump improves the window

It could technically all be placebo and it is solely based on jumping out of spin on a certain frame window, but I do know that once you lose the speed you must start a new skate to get the speed back again.

For this demonstration, I decided not to figure out where to turn the camera to show the lines better. RTA it would be ideal to turn the camera for a short amount of time to have a faster backup if the skate triple doesn't get the correct momentum or you accidentally jump at the edge of the ice (causes a normal triple instead of skate triple).

Rightside down 1 - Falling crates section comparisons

jda7 2023-05-11 | I found a new RTA viable strat for this section. First, throw 4 to 5 crouch spin fireballs and then walk forward. This should clear the goomba and break the bottom crate.

Before you get to the line in the background, do a jump then nunchuk spin while holding a, then a double jump then nunchuk spin while holding a (I double tap a before landing to make sure I double jump), and complete with a triple jump. This triple can sometimes go too far forward, so I go neutral a little bit during it before pressing forward.

For the 2p freeze on the goomba, I use a nunchuk spin to throw the fireball earlier to not bonk on the crate from the distance I'm long jumping at.

Lava Lair 1+C lava wall boost comparisons

jda7 2023-05-10 | Upon uploading this I realized it saves even more in lava lair C than this. Jhay already went through the life up get animation so it probably saves about a second more. 1.67s is a lot more worth it considering you'd probably only lose like 3-4 if you fail.

Trick explanation:
I aim for the right most part of the wall because slightly left of that is weird geometry that can sometimes give you a glitchy wall jump leaving you incapable of wall jumping up.

Lava lair c:
After that, I long jump and line up with the line on the floor and do a jump and instant nunchuk spin to increase the window for the lava boost. I then anticipate when 2p's shadow will go off the edge and that is the moment that I ground pound. You want to hold forward the entire time so you don't accidentally hold neutral during the lava wall boost and lose time.

During the lava wall boost, you want to correct yourself up right a bit, I usually shift the stick up right twice and then back to up until I pass by the first question block. In lava lair 1, you also want to freeze the thwomp for a short amount of time so you can attempt the thwomp push.

Lava lair 1 part:
For the thwomp push, I do an a held double jump and A held wiimote spin and then a no a held nunchuk spin and try to aim for the arch of his slam. This never kills you, but sometimes won't push you as far as you see in this clip.

The rest is as normal.

Rightside Down S - 15s remaining crate game using 2p

jda7 2023-05-10 | I start the minigame turning the camera to keep certain crates on screen. I then do a long jump into nunchuk spin while also doing a wiimote spin for 2p to destroy the top box. You can leech the wiimote spin for a set of crates instead, but I have trouble when mixing different fireball spins and often make 2p spin when doing so (which since I hold a to buffer 2p lock-on, the spin won't cancel so 2p won't lock on). If it makes more sense the other way for you, then do it that way.

After the long jump, I shoot the top left and right lines and then use the busted crate below as a visual cue to shoot the bottom left and right lines.
I then shoot a fireball left and use 2p to break the lone left crate and then walk right until 2p is ready for the right crate.

I don't do any fireball crouch cancels because you can't with nunchuk spins. A 16 would be possible if you can juggle everything while mixing in a couple crouch canceled fireballs.

Cosmic Cove 1 - Goryuyas skate triple to frozen waterfall

jda7 2023-05-08 | Old goryuya strat that got lost when he deleted his channel. I recreated it using a pretty consistent method here and also compared it to hardcore to show how much it saves.

Shiverburn 1 - Prince pikante fight comparisons

jda7 2023-05-07 | The time save mostly comes from skipping one animation of him getting mad.

When baiting the shots I usually spin to make sure they don't land in bad spots nor go too high. For the coconut 2p has, I nunchuk spin that and then use 2p on the coconut he fires because I will land by the time 2p sends it to me. This means I get an easy timing to hit that back as well.

I also use the amount of distance to stop the coconuts from being hit too closely together because he is immune for a bit after getting hit.

For the triple, I missed the first coconut (on accident) and used 2p spin spam to hit the next one to me as a back up. To perform 2p spin spam, first spin and then press a to cancel the spin. Repeat this as much as you want to turn 2p into a beyblade.

Puzzle Plank C - 31.46 using adjusted route for 2p

jda7 2023-05-06 | It's the same pathing as my last 1p route, but I use 2p to grab 3 groups of 4 coins to allow me to take better lines.

Upside dizzy S - Slightly faster crate game using 2p

jda7 2023-05-05 | This is still improvable, but I went with what seems like it'd be the most consistent.

The strat here is to turn the camera, target the bottom right crate of the bottom right group of four crates while throwing a fire ball at the one above. This leaves 1 crate for 2p to take out while you take out a lot of the rest.

I then use the broken crates to line up the rest of the fireballs and ignore 2p. By using 2p on the final crate groups and also using diagonally thrown fireballs on them, you could save even more time. However, both of these can fail so I went with the more consistent method.

Supermassive - New koopa planet strat and comparison

jda7 2023-05-04 | I moved to the coin to ensure I hit the koopa, but the drill is very generous and might have been able to hit the koopa from the initial spot anyway. Someone will need to test it later since this is very useful for RTA runs in 120.

The strat is similar to purplesun's GS3 TAS, but I purposefully don't go for jump cancel for timing reasons. A bit before getting to that random brush after bouncing off the drill is when I switch which koopa 2p is freezing.

I love saving over 2.5 seconds for 3 different levels. Very cool.

Fearsome Fleet Boss Fight - New Strats and Comparisons to TAS

jda7 2023-05-03 | New strats that can be done 1p? What a novel concept. Anyway, improved first phase by .7s and 2nd phase by .1s.

I recommend the 1st phase strat RTA, but not the 2nd phase. 2nd phase strat sometimes the hammer blocks your tongue or the bullet dies on the hammer. I move my flutter from up left to up to try and stop that from happening, but it can still happen sometimes. Not recommended until it is more consistent (maybe a flutter reangle or something. idk it's like .1 idc enough to research).

Cosmic Cove 1 - Tighter Coin Route 2p comparison and full strat

jda7 2023-05-01 | roughly 2 seconds faster. showed both hardcore's and jhay's beginnings of cosmic cove 1.

if you end up with 6 coins coming out of the cave there is a coin on the platform near the hungry luma as a back up, which would still save time.

you need at least 4 starbits for this strat and there are plenty of places to take out a slime and get more if you miss the ones from the ring at the start.

Flipsville C - Getting Checkpoint using 2p and exploring

jda7 2023-05-01 | You could get a new coin high score using this. Also that coin I picked up was in fact invisible.

Flipsville C - 2p Optimizations 55.86 (EMU)

jda7 2023-05-01 | I couldn't save a cycle, so instead I optimized the ending by grabbing the 4 coins you would normal sideflip and spin for in the last room using 2p in various rooms. Once in the beginning, 2 in the middle room, then the last one while drilling into the star room. This could be better optimized still, but this is good enough for now.

Melty Monster 2 in 51.58 (WR)

jda7 2023-04-29 | For this level, I originally researched and found mario was faster. However, that research was old and may no longer be true. This TAS of GS3 by PurpleSun uses luigi and does movement better than previous luigi movement that I had seen: youtube.com/watch?v=MhqGMs8oWSw&feature=youtu.be

This level is mostly pure optimization of long jumps and lines to take with rock mario. I closely followed what the TAS does and then deviate to zig-zag skip at the end. This can still be lowered both through the TAS strat of rock jumping over the bridge and through hitting less crystals along the way to the star.

I tried the rock jump for a while and got it once over about 2.5 hours. I choked that run and settled for this. I didn't want to give myself repetitive stress injury. Here's the choke and also a funny bumper moment after: youtu.be/4EJ3PuncP-4

Update: This was done before easier rock jump method.

Flipsville 2 - Cycle Skip using 1 up textbox

jda7 2023-04-19 | New 4s time saver if you can dodge 1 ups until here.

I release the 1up on the 2nd beep before it switches.

Make sure you go quickly because it is possible to not be fast enough.
Also you do need to freeze that fuzzy with 2p otherwise it will hit you.

Boom Bunker 1 in 1:35.28 (WR)

jda7 2023-04-09 | After a few hours of research, attempts went surprisingly quick. This was the 3rd attempt in a row to make it to the boss in a 1.5 hour session. If only there was a consistent visual cue for the 2nd cannon shot, then it would have been much easier to grind and I wouldn't leave it with these few glaring errors.

For the first cutscene, I cut it at the 3rd note so the blue platforms don't interfere with my 2nd cannon shot.

For the second cutscene, I wait so I can grab both starbits from the broken megahammer's eyes.

For the bullet bill, you want him to turn sharply up so he spends the least amount of time turning as possible.

My movement to the glass dome is very specific so that the camera doesn't change much.

I use the green square at the bottom left of the cannon to line up my instant cannon shot. It's not that hard. When you backflip and ground, go to the center of the glass dome. This ensures the cannon instantly takes you in. You can tell when you are over it via your shadow.

There is technically a cutscene shorten on this launch star that I have not figured out how to implement. It just seems impossible to be standing still when killing boomerang bros.

This camera cancel makes the mines an issue, so you have to shoot down the one in the top left to clear the way for an instant shot. I jump and then fire starbits at the same time. This planet is jank when you are shooting starbits on it unless you have jumped. Then it works a bit more normally, but it is still hard to hit it consistently.

This instant cannon shot is very precise and has no real visual cue. You just hover between to meteors/rocks close to the top of the screen and fire immediately. It's basically luck.

If you hit just the right height, a spin reaches these rocks and allows for a ground pound. I didn't have the right height, but managed to jump cancel the landing which saved the run.

This boss fight is pretty cool, but quite challenging. I jump into the cloud flower so that I can stop and turn around faster, then I do a cloud sideflip up to bowser jr and backflip holding back. I then make a cloud before luigi reaches max height on the backflip and jump down. It is important to jump because sometimes bowser jr resets his panic animation if you walk onto him from the cloud instead.

You have to time your ground pound for when bowser jr starts spinning. This isn't too hard once you get used to it. The shrinking cloud will then cancel your animation and allow you to create another cloud. Unfortunately, I get a homing ground pound here which costs around 2 seconds over frame perfect ground pound cancel. When looking at the footage I ground pounded too early anyway, so the homing saved it.

The final phase is fairly standard. Luigi just needs 3 cloud backflips and a spin to reach bowser jr. I tried to ground pound earlier but I was slightly late on the input and had to jump before it instead which cost a bit of time.

If it weren't for the lucky shot, this would be a cool grind level. Maybe the comet is better as a grind level? I don't know.

Rightside down GS2 in 1:03.98 (WR)

jda7 2023-04-09 | I had to retype this description because youtube dumb.

This run had pretty good execution, but bad luck at the crate section. I'll probably try grindind this again some time since I just need to get better luck there to improve it to around a 1:02 time.

If you do a jump and nunchuk spin fast enough, jumping on this paragoomba becomes pretty easy. Doing it slower but releasing a sooner on the nunchuk spin can make it trickier to land on the paragoomba.

When jumping off the dandelion, hold neutral then press forward a bit before landing. You can get variable speeds from it based upon when you press forward, all the way up to full running speed.

I found that this spin canceled long jump into the goomba saves about 19 frames over the previous method. This long jump fails if it is too far or too short. Too far you hit the switch with the spin. Too short and becomes a backflip or bonks or the back of the switch stops you.

The backflip from this ? block makes it so you have to hold a for varying lengths of time based on where you start the backflip. The farther left from the center, the more you have to hold a in the spin. If you didn't grab the coin that I grab, then you held a for too long and will hit the top ledge before the slope.
Grabbing the coin is a toss up on whether the trick works or not because you can bounce on the goomba and still not make it sometimes.

When you go to the middle of the slope, hold up and slight forward to not cling the wall. If you hit the top, you can just glide down as normal.

I found that a clean double jump here is the best way we have of doing this section. There is still the possibility of one of the two TAS strats being used one day, but I had 0 successes in 2 hours of trying that. On the second jump of the double jump, I go neutral on the control stick to control my speed better.

This next section requires you to land far left on the wooden? block. If you land to far right the thwomps load early and you will have less speed approaching them causing you to get squished.

The nunchuk spin after the wall jump could have been done a bit faster.

The crate section here had bad luck. The strat I am using is very fast if the goomba walks to the left. He walked to the right instead, so I threw a standing fireball to deal with that. It unfortunately means I lose all speed. I also attempt a triple jump when I go under the crates, but it failed. Luckily mashing a happens to get this wall jump which is also fine.

Going to the 2nd crate stack, the fireball missed the goomba and hit the crates behind it which meant I had to do another standing fireball. These are the main time losses of the run.

I failed the same rslj that the old WR did. Back then nemi called it "fast long jump landing" or something like that, which shows how old that record was.

For the ending I enter the gravity much earlier and drift up right. The drift up right delays the transition into the gravity while not interfering with how fast you get on the same level as the star. This saves 1 second over the previous record.

Retyping this was a pain and I probably missed some stuff since this was no longer fresh on my mind. Ty youtube.

Fearsome Fleet GS2 in 23.60 (WR)

jda7 2023-04-09 | This level is very hard due to the position of the star and how you have to adjust your angle in real time based on how your previous flutters were angled and other factors.

I skip the opening cutscene letting only a tiny bit of the 2nd note play so that the fizzlet jumps. Any more and he will zap instead. Waiting too little causes him to jump too late.

The 2nd cutscene I wait so long to have luigi walk in the correct direction at the start.

Statue long jump can be finnicky, but it's the fastest way to yoshi according to PurpleSun. Maybe for RTA ILs the yellow platform boost he talked about is more consistent, but I just went for the fastest thing even if some runs would die to it frustratingly.

I tried approximately 3-4 different strategies for the flutters.

Strat 1: Crouch turn TAS strat (less compatible with other strats) youtu.be/boux_GCvNEU
-Small crouch turn down right. Slight right from up for most of the first flutter and shift to slight up above up right.
-Pray the fizzlet goes underneath the line you are going on
-Flutter to the star. I usually hold the angle still for 3 more flutters before shifting to an angle that will give more speed (further towards the right side of yoshi (player's perspective))
-Dismount near the mine and hope you get it. I never grabbed the star once, but I did miss several times.

Strat 2: Close shave flutter TAS strat youtu.be/o09L3z0sghw
-Similar to the video, you do an a held jump and a nunchuk spin (still holding a). Start the flutter holding up and then shift
to up right in the middle of the first flutter. You'll need to do height flutters, so make sure to cancel on yoshi's 6th "kick" until you have enough height.
-Pray the fizzlet goes underneath the line you are going on
-Same flutters as Strat 1 here post fizzlet bounce

Strat 3: Spin bonk flutter youtu.be/Y1C2Pt-Mg1k
-There is a piece of geometry on the platform that allows for a spin bonk
-You can just do a "fake" spin into it and hope it bonks up right, then do a flutter holding up and shift to slightly right of up. Continue for about 4 flutters to be safe before committing to an angle for better speed.
-A tip here is you need more height flutters than other strats because you usually are a bit lower than the other strats.

Strat 4: Double close shave flutter
-This is the strat I went for in the video
-Do a held jump nunchuk spin (still holding a). Start the flutter holding up and then shift to up right in the middle of the first flutter. Depending on how good the close shave was, you may need height flutters.
-Land close to the middle of the platform and perform the same trick again, but you'll want to shift the control stick further to the right than up right and do it relatively quickly (it is dangerously close to killing your speed).
-If you did it right, you won't even have to readjust your angles. For the height I go for, I try to be around the ? block in the background. I dismount around that one mine and hope that I just find my way to the star.

Close shave flutter: The name I gave to increased speeds when fluttering dangerously close to the ground from a spin flutter.
First used in nemi's perfect run/ultimate test IL speedruns. The closer the shave, the better the speed (usually).

You can technically go even lower with these close shaves by releasing and repressing a to cancel the spin early.
youtube.com/watch?v=dgZqwLfoJtY

Strat 2-4 I would try to decide which one I would do based on the flutter and conditions. Really good close shaves on the first flutter I would do Strat 2 and 3. Bad close shaves (lower speeds) I would go either for Start 2 and 4 or just 4 based on the angle and ofc the fizzlet position. Strat 2 is never guaranteed since it is purely RNG.

This was a 14 hour grind, so I'm not entirely thrilled with never getting TAS strat in the time, but at least I finally got the WR.

Boulder Bowl 1 in 1:09.08 (WR)

jda7 2023-04-09 | An improvement over the last WR by almost 6 seconds. This level is tricky but also fun.

For the cutscenes, I wait until luigi's left hand passes over the fence. This puts the cycles on the next planet exactly how I want them. The first one is skipped immediately.

This beginning section is tricky to optimize. I opted to try and focus on keeping running speed the entire time instead of getting better angles. I obviously didn't care if I failed to keep the speed due to the strats I was going for. This grind was more quantity over quality. I even got a ledge grab but kept going because that's how much I wanted more attempts at the hard later strats.

For this second planet, there's a way to bounce on this luma by entering the planet's gravity lower, but I often find where it puts me to be more trouble than it's worth. I can line up the next trick easier with a long jump so entering the gravity earlier is fine.

New trick. By facing backwards you can get into the sling star on this specific side. This means the slower methods of getting on the crystal are obsolete. I attempt to do it from the toad but the clip fails so I do a back up where I run at the crystal, jump, then when I see luigi is closer to the ground I then switch my control stick to half-throttle up left. This causes a turn around double jump when done correctly, which you can then do a spin and clip in as you would in the normal trick. There's also a method to clip in using a backflip and spin from underneath the rock the toad is mining, but it is slower.

Since I used the sling star, I opt to go around the front of this planet instead of the back. Did not time if it was faster.

When you skip the cutscene for the boss you get running speed almost instantly which allows for a long jump.

Boss fight time. The first hit to start the fight is obvious. When you hit him changes the camera angle for the fight so it's important to hit that first hit. The second hit I follow the large red line down until the coin comes on screen. When I see the coin I turn hard left. Waiting for some reason lets you turn sharper and hit him. For the third hit I go up left trying my best to line up with the center of rollodillo. I then do a rock triple jump to go out of his aggro range and land onhis back. I do it again on the last hit but now I have to instantly do the rock triple or I'll run into him. The method I use for the rock triple is to just shake the controller in my hand with my thumb over a. I find this usually gives me a rock triple. I also make sure to stop mashing/shaking when I think I've executed the triple so I don't do too many and overshoot.

I had a 1:08 pace run die at the ending here because I mashed a so hard a shook the controller and rock double jumped over it.

Freezy Flake GS3 in 44.88 (PB)

jda7 2023-04-09 | After giving up GS1 to pottoww's insane run, I went here and got this in a little over an hour.

I decided I could get the mid-air rock jump to this planet faster than I could find a frame to save in pottoww's run. I got 2 over to the planet with 1 hour of attempts. The first one didn't WR and didn't have a great angle/landed in a spot past the goomba statue up top which wasted tons of time.

This run, however, landed where I wanted so I could execute the route I came up with for the star chips. I made a few major mistakes here but still saved enough time to beat the old WR by about 2 seconds.

Major mistakes:
-Missing the crystal with a starchip. I hit it 80% of the time so this was unlucky.
-Missing the starchip in the snow tree and having to go back for it
-Super slow camera cancel lmao

Here's my mash method so far for infinite rock jumps:
Hold the wiimote sideways and put the nunchuk down. With your left hand curl your index finger along the top of the wiimote to the backside. Put both your left thumb and right thumb on the a button. When you are in rock form for best results jump once normally then shake the controller by raising it using your left index finger hooked behind the controller and press a at the same time, then begin 2 thumb mashing while continuing to shake the controller with your left arm/hooked left index finger.

Freezy Flake GS1 in 24.96 (PB)

jda7 2023-04-09 | After I touch the fire flower I let the control stick go neutral and hold z. Fireballs seem to have a bias towards the 8 notch angles, so I use neutral to throw a fireball at an angle it usually does not like to go. I also try to make sure that I am as low to the ground as possible so I can get the long jump quickly. With luigi, the long jump sometimes requires neutral to be held for a little bit in order to save the most time possible since the disc planet's gravity tends to slow him down if he has too much speed when transitioning.

I went to the right side of the starbit crystal to avoid the lil burrs. There is a potentially faster strat by going left side of the crystal and getting more fireballs out while still long jumping, but it only works if the lil burrs move out of the way, which seems random.

Every fireball I threw I tried to be as low to the ground as possible and then did a no a hold wiimote spin. Where it mattered most, however, was the last one. Keeping running speed here is crucial for sub 25 with right side crystal.

For the clip, I actually found a second method to the clip pretty consistently, but I am unsure if it is faster. It is actually possible to do a crouch sideflip to clip onto the slide, which in theory means you could long jump sooner. However the extra running that has to be done after the long jump to ensure the correct position for clipping usually waste time. Not to mention that
it is also a harder input so less runs complete. Here is a video showing that clip as well as some extra weird stuff that happened: (vid link).

Upside Dizzy 1 in 43.40 (WR)

jda7 2023-04-09 | This is only an improvement of .35s over the previous WR. This level is pretty challenging to be consistent at mostly due to the ending.

For the opening cutscene I make sure 1-2 notes of the opening jingle play as it lets the platform after the wacky wall kick give you a boost.

For the beginning, I do 2 running speed conserving long jumps manually (no chain) as you have to walk a bit on the ? block to get a good wacky wall kick. The timing of this wall kick is to press a after passing the
2nd coin. It's not the hardest thing ever, but pressing it too early can cause it to fail. On the platform, I use my conserved running speed from previous to long jump into the gravity a bit earlier.

Spinning into the up gravity field instead of after seemed faster to me but I did not time it. It could be placebo (or literally the same).

I do a neutral strain wallkick to pass the goomba without spinning it to save a few stun frames caused by spinning an enemy.

This last section is the hardest part of the level. First shoot all the enemies beforehand, then long jump at a certain part of the wall (i use a brick's line as a visual cue). A little bit after long jumping, switch the controller to up right and do consecutive up right long jumps until the part where I jump. If you conserved running speed to here, the double jump will make it to the platform. From here do long jumps holding up right once again. If everything went smoothly then you should clear the ceiling just right so you can walk off.

The only real mistake is here at the end. I probably should have continued to hold a off the long jump so my spin could go higher up the wall for the wall kick. With how low it was I got a ledge grab which cost a fair bit of time compared to just long jumping immediately. Even if neutral wall kick strain is needed to avoid the ledge grab, it would still save a good deal of time. Overall though, this was pretty solid and did everything I had a plan for (which was everything except the very end).

Flash Black C in 38.58 (WR)

jda7 2023-04-09 | This level is very challenging. Getting a mistakeless run like this took maybe 15 hours or so of attempts. I'm proud of the end result as it looks very clean. Tough starbit shots, nonstop movement, inputs being dropped, luck. This level has it all.

I'll be numbering octoombas based on when they are killed in this video in this description for clarity's sake

For the cutscene a little after the octoombas start moving left I skip the cutscene.

For the beginning, I perform a long jump and drift up right before doing a spin canceled long jump into the octoomba. I do this instead of the previous strat in order to spawn more starbits for the rest of the level to reduce the RNG factor a bit. This can still go wrong and can sometimes net 0 starbits, but more often than not you'll get 2 or more. In addition, absolute faith that the next 2 starbit shots hit the 2 octoombas is needed. If you hesitate, then the starbits are more likely to go off camera as you move away. I also recommend hitting the octoomba before the long jump to octoomba #5 on its left side to have better odds of the starbits being on camera.

After octoomba #6, you'll see my pointer near where octoomba #5 died. I am trying to grab his starbits, but I got unlucky and none dropped on screen. I still had enough for the run to complete without them, but often times I get 1-3 there.

The next section was very tricky to figure out. The starbit stun order here is very particular and also inputs are prone to being eaten here due to kicking octoombas. Every time you kick an enemy the game pauses for 2 frames and during those 2 frames no inputs are registered. This makes many of the sections way harder as if you accidentally press a button in that window then the run dies right there. That's why I shoot octoomba #8 and #9 one after the other before then hitting octoomba #11 in the distance (to stun him in a good spot) and octoomba #10. I needed to give myself enough time to avoid this input eating property.

During this ending section, I try to collect the static starbits at the end of the level while continuing my movement. This reduces the amount of luck needed while increasing the skill floor required for completion. I grab them at very deliberate times. The 2 by yoshi's crystal I grab around octoomba #14. The ones in the end area I grab when jumping to octoomba #18. The top of the screen is weird with the pointer, so I chose these times as they were easiest to grab at these points.

At the ending crystal, I forgo killing the enemy in front of the crystal and instead long jump into a fast yoshi grab technique and opt to eat him with yoshi instead. Sometimes this can make the jump to the last 2 octoombas awkward as the jump is prone to getting eaten due to the fact that eating 3 enemies takes more time and you end up cutting it really close. That's why I hesitate slightly instead of taking the most efficient route to octoomba #29. Also if you need help with the jump to octoomba #29, press a in reaction to the yellow star that shows up when yoshi eats an enemy. This jump also is prone to getting eaten unless you use this visual cue.

There isn't too much more to be done with this level. There may be room for improvement by getting running speed conserving long jumps, but I found the cycles weren't fond of them and often would be too tight or just not there at all. It would make this level way harder than it already is. Would probably become something like a 100 hour+ journey.

Honeybloom S in 15.33 (WR)

jda7 2023-04-09 | This beats the previous WR by .02s. This run honestly was not that special. TAS is now down to a 14.93 which means there is still a lot of room for improvement. Most notable is the entry into the wall jump section and the long jump to the star.

The beginning section is interesting. You don't want to hold a to bounce high off the first goomba. I don't know if it's necessary, but I make sure to strain this wall kick (however not with c because the c button also delays spin inputs). You do an a held nunchuk spin around the flower and continue holding a the entire time. One of the main areas I think some time save here is just getting off the swing. I think I was a couple frames late. My bounce onto the swing also seems unoptimal which may have been why I went with a later release.

For the wall jump section, I just guessed what kind of strain angle might be needed here (still unsure). I winged the first 3, I went with down and away from the walls that I am kicking until the final one where I strain up left. The spin in the wall jump section is a no a held wiimote spin, however holding a for a tiny bit during the spin is recommended for enough height
(even TAS does it). To be honest, I don't even know which jumps the TAS strains besides the last wall jump.

When coming out of the wall jump section, I returned the stick to neutral before landing (c button doesnt work) and then reheld toward the star. I was a little slow on starting my long jumps, so this is where another good chunk of time is.

I don't know if humans will see sub 15 some day considering how specific every height has to be, but this can definitely get to a low 15 with a longer grind. I only put in around 2 hours to get this so a more dedicated grind could do it.

Slipsand 1 in 2:45.08 (WR)

jda7 2023-04-09 | This level is annoying to speedrun. I find the best way to actually long jump in the sand is to just mash a quickly and pray.
Timing the jumps works less than mashing.

The path I did for the first planet is as follows:
-Hold slightly above left
-After the first long jump in the sand hold up and continue the chain
-move slightly up left during the final section to ensure no fireballs hit you

The execution on the first planet is just enough to make it through this next section cleanly.

My biggest mistake in the run is on this planet. I get an angle that doesn't allow for a wall jump and have to do a ground pound into the sand which allows me to backflip (because programming is hard). It wastes about 2-3 seconds meaning a sub 1:42 could be possible and a sub 1:43 definitely is.

These rolling rhomps can push you forward, so I decided to try for it on the left side of as many of them as possible. This one was bad until the very end where I managed to stick to it very well. You don't want to get pushed for too long by that one or you'll go through the floor like this: youtu.be/ey2roXIX7U0 .
It's safer than I thought it would be to go for these pushes, just don't panic with a jump and fly off the level.

At the end of the slide, you can just go left and jump a bit after the camera pans to go right into the launch star.

I still don't know the best way to trigger this boss fight. I hold right then switch to up and long jump. The camera messes with angles so much here.

In the boss fight I go into first person. It does nothing, it's just funny that you can. It's nice to not always have to look at squizzard or his sandy sinkhole.

When the boss is beaten, hold slightly left of up until the cutscene ends. For whatever reason up makes you walk not up, but slightly away from up doesn't. So just react with holding directly up after the cutscene and long jump.

Rolling Masterpiece C in 1:23.16 (WR)

jda7 2023-04-09 | So, Pottoww beat my previous run without knowing by a whole lot. Apparently the launch pads can launch you quicker or slower depending on how you enter them. It's essentially random for a human due to how precise everything is. It is very hard with pure motion controls to get the same inputs every time. Anyway, that's the main time saver he did as well as the jump at 0:39 that has a 1 frame window.

I found and used a couple extra time saves for this level so I didn't have to get as lucky as him to win.

For the beginning I found out you can get an instant mount by doing a no a held wiimote spin onto the ball. This saves around 40 frames (.66s). A good launch out of the first launch pad is a 1:57, but I'd accept high 1:56 as well.

You can tell how good a launch pad is by the number of "vwoop" sounds. The lowest I've heard is 2 vwoop. It's technically possible in theory to 2 vwoop all of them but a human won't be able to do that (TAS might not either since entry for each launch pad is different).

After the launch I turn the wiimote completely on its side (front part facing right) and then tilt the wiimote forward. Before hitting the wall, point the wiimote left and shift the wiimote back to normal (front part facing up again). This will usually guarantee that you keep a bunch of speed from the launch pad.

Since I found a strat to keep speed, I use a different route on this crayon planet. Make sure to hit this key more on the left side because how you hit this key changes the camera in different ways. The best way to make sure you hit it on the correct line is to line up with these starbits at (timestamp). You'll have to learn to manuever around the red crayons while keep almost full speed, but I found this route to be slightly faster than WR route. It is close, however, so it isn't the most important if this route is too tricky.

The launch pad on the crayons planet on average gives an 8 vwoop (and i've seen 4 before).

For the jump out of the hole on the next planet, I angle the wiimote up right and hope I jump on the right frame. The first 2 frames just kill you unfortunately. The 4th frame either eats the jump or hits the wall so you get a pathetic jump out which wastes 1 second compared to hitting the strat.

To time the jump up to the ? panel puzzle, I wait until a green starbit behind the grating of the elevator on the right side goes past the elevator. Then I tilt forward and jump when I can see the blue of ? panels for a bit (slight hesitation because I've failed the jump even with these cues before by being too fast).

When entering this launch pad strat on, I twist the right side of the wiimote towards me a tiny bit about a 3rd of a second or so before I enter it. This seems to often give a 4 vwoop. It only works with a very direct line and is still difficult, just less so than the others.

This planet is very hard due to the positions of the bob-ombs being random (including how many fell off). There are 2 paths that can be taken. Based on the bomb positionings, I choose a path each run so I can complete as many as possible.

I've seen this planet give a 4 vwoop before. Since chain jumps seem pretty fast, I usually just jump in and pray.

Here's how to 3 hit chomp like a pro:
-Tilt down left when being launched in
-Switch to left after hitting the wall
-Jump into the chomp aligned slighty down from his immediate right side.
-Jump into him 2 more times and win.

The reason you hit him slightly lower than just hold right the entire time is because you don't want the chomp to touch any of the walls. If he does, he'll barely hang on when attempting the 3 hit and kill the run. So I try to tap him so that he flys up left and rolls back down away from the wall and lines himself up for the next 2 jumps. Learning this 3 hit definitely helped consistency on the ending.

At the end of the level, I waste 1 second jumping over the goal to hit the flag and turn around. This is so I can face backwards towards the flag and get backwards speed. This saves back 72 frames, which means it saves .2s alone. In addition, it's possible to 4 vwoop the goal using this strat as well to save an extra second.

This level wasn't great to run, but once I got consistent strats it was only a matter of time.

Slimy Spring 1 in 2:07.13 (WR)

jda7 2023-04-09 | The main improvements on this level over the old one is optimal swimming, better lines, a squeezed backflip at the end, and JP text.

Optimal swimming for luigi is just a held nunchuk spins.

After the launch star, I purposefully delay when I go for the spin canceled ground pound so that the shell is in reach after 3 spins.

I found the squeezed backflip ending to be difficult to get as it was very particular. Too far away and you get a normal backflip, too close and the backflip gets squeezed too much directly back into the ground. Pressing a on the first frame of regaining control also normal jumped. I found a somewhat decent alternative method of squeezing a single jump into nunchuk spin before hitting the ground (very fast inputs, basically a rising nunchuk spin). Unsure how effective it is at saving time, but could be usable RTA.

This run is certainly improvable. I missed fast dive at the beginning (probably loses .1-.2?). I went too far to the center before the launch star wasting a few frames. The last shell did not angle back up properly and took me lower than I wanted which also wasted time. Finally, the first toad textbox was a little slow. A 2:06.xx is definitely possible.

Starshine Beach GS1 in 27.20 (WR)

jda7 2023-04-09 | I had this idea when looking at what stars I might try to WR. I wondered if this lakitu always flew the same path. It turns out he does. That means we can avoid stopping on the slope and flutter from the lakitu to the star. This path does waste a bit of time though because of the more roundabout angle needed to get to the lakitu. This feels like it could be a consistent strat for RTA as well. The biggest problem is the approach can change the camera angle making getting the side flip angle harder or easier depending on the approach.

The flutter goes sideflip from up left, then spin flutter holding directly up. When I bounce on the lakitu, I actually hold slight up right for the first flutter and then go back to up to maintain speed.

Sky station GS1 in 3.16 (WR)

jda7 2023-04-09 | This time is still improvable by several frames according to purplesun's TAS: youtu.be/b-yFYZLkmps.

To start, the backflip is very jank. In order to get the angle I got, you need to hold slightly right of directly up on the control stick. One would assume you'd want to backflip opposite the direction you want to go, but not this time! Anyway, I actually time the backflip approximately when the screen is all green. It's not exact, but eventually my muscle memory caught on to how to do it. The reason I want to time it is to try and get the backflip on the first possible frame. Later frames give different angles.

After I backflip, I do a nunchuk spin and release a sometime during the rise (i just guess it). This is to land on the post as early as possible. Next I have to run a bit to get a long jump. If I remember, I also try to bring the control stick into a half throttle position upon landing. By doing so and landing early enough it is possible to get the low long jump TAS did RTA. I've only gotten the low long jump twice in 2.5 hours of attempts. It only saves a frame, so just trying to land earlier is more important.

Short star so I don't have much more to say.

Stone Cyclone GS1 in 5.73 (WR)

jda7 2023-04-09 | 1 frame off the TAS time. This trick is very hard. I really had to dissect the TAS in order to improve my chances here. The 1 frame I lost was likely the cutscene skip being 1 frame early.

When starting, you need to slightly adjust yourself up left for a tiny amount of time to get into a position to not get crushed. This trick seems to have a 1 frame window to press a to get the launch properly. For the launch itself I held neutral before landing on the tox box to avoid turning around. You can hear when I go neutral because I used the c button to do it. I don't know how specific the timing for everything is, but since this an IL I could just throw attempts at it until I eventually got the result I wanted.

Flipsville GS2 in 12.66 (WR)

jda7 2023-04-09 | Going around the wall is easier by using an angle between up left and left.
The ground pound needs to be far to the left of the grate to save time.
The long jump is an angle above left, then shift to left after long jumping.
Change camera when you have gone a bit above the tan dot on the wall.
The spin can be removed similar to the TAS, but is way harder.
TAS: youtu.be/pUWDsUgGFPw
The long jump to the star can be improved as well.

Beat Block C in 35.10 (WR - No Co-star)

jda7 2023-04-09 | This is a brutal level to do very quickly. I tried both mario and luigi on this level and I can say for sure doing this level with luigi is harder. Mario's ability to long jump quickly from a standstill saves him a lot of headache if any mistakes are made or the cycles are slightly off. Luigi's larger long jumps also puts him at weirder spots on the planet, which means he has a stricter cycle to make.

For the crystal strat, make sure you don't do your long jumps too well or too poorly leading up to it or you'll clip the blocks from the green blocks to the yoshi and lose a bunch of speed. Also make sure to land closer to the edge of the crystals as this gives more leniency.

This flutter is difficult to get full sideways speed in because yoshi isn't facing directly left. He is actually facing slightly up left (despite hitting a notch). This is because the camera is slightly different in both areas. For this particular attempt, I actually got a pretty bad flutter. The spin came out very late and my flutter speed was only average.

The ending section I forgot how to do from my any% practice, but I figured it out again. It's just a bunch of long jumps directly to each silver star. Be careful near the 1-up because a bad angle can send you off the level. A lot of my attempts died at the very last part because of the strategy I was doing. A long jump into the silver star then backflip after is much safer time wise as you don't risk ledge grabbing the block and wasting time.

Chompworks GS2 in 37.30 (WR)

jda7 2023-04-09 | First run that made it past the slope jump at the beginning got WR. Nice.
I tried to improve it by doing an extra hour of attempts, but these strats
are not very consistent.

The beginning requires a speed conserving long jump into an angle jump with another speed conserving long jump. I found chaining the long jump after instantly to be more consistent at getting the slope jump. The slope jump requires z to be held and is basically rng. Buffered jump doesn't get it consistently. Having full running speed here is important as it allows for a very fast switch hit.

One of the new time saves over the old WR is this cutscene shorten tech: youtu.be/33bWxAUqBmk By crouching, it is still possible to get the insta-walk after without time loss. I got lucky with the bounce off the gearmo as I had not practiced it, but it was still kinda slow. Just lucky the run didn't die there.

I missed the jump cancel, but the TAS strat saved enough time to make up for it. The angle for this jump is really hard to get. I find a bit more success doing a slight turn up before getting to the edge and then performing the inputs for this run off spin, reangle, long jump boost. I have gotten one to go directly into the star during practice, but this was the best in my actual 1.5 hours of attempts.

In terms of improvements, the opening can be done faster by letting the chomp roll more in. The next two are ofc a better launch and a jump cancel.